Professional Documents
Culture Documents
Corporate Finance
This impressive collection defines the current status of corporate governance and corporate
finance from a European perspective. The book brings together a comprehensive range of
contemporary and classic articles, with a major emphasis on recent research, accompanied
by a new and authoritative editorial commentary.
Topics discussed include:
Addressing both the theory and practice of corporate governance and corporate finance,
this book is an invaluable resource for scholars and students engaged with managerial
finance, financial economics and business law, as well as the role of the corporation in the
modern economy. It is also fascinating reading for practitioners interested in managerial
finance.
Edited by
Ruud A. I. van Frederikslust,
James S. Ang and
P. Sudi Sudarsanam
First published 2008
by Routledge
2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN
Simultaneously published in the USA and Canada
by Routledge
270 Madison Ave, New York, NY 10016
This edition published in the Taylor & Francis e-Library, 2007.
“To purchase your own copy of this or any of Taylor & Francis or Routledge’s
collection of thousands of eBooks please go to www.eBookstore.tandf.co.uk.”
List of tables ix
List of figures xv
About the editors xvi
Preface by Floris Maljers xviii
Acknowledgements xx
General Introduction 1
PART 1
Alternative perspectives on corporate governance systems 7
1 Michael C. Jensen 11
THE MODERN INDUSTRIAL REVOLUTION, EXIT, AND
THE FAILURE OF INTERNAL CONTROL SYSTEMS
PART 2
Equity ownership structure and control 133
PART 3
Corporate governance, underperformance and management
turnover 255
PART 4
Directors’ remuneration 367
PART 5
Governance, performance and financial strategy 487
PART 6
On takeover as disciplinary mechanism 645
Ruud A. I. van Frederikslust, until 1 January 2007 has held various positions at RSM
Erasmus University since the graduate management programme was first established in
1970. His most recent position was Associate of Professor of Corporate Finance. He is
author of the work Predictability of Corporate Failure (Kluwer Academic Publishers),
and editor-in-chief of the volumes: Science, Management and Entrepreneurship (Kluwer
Academic Publishers), Mergers & Acquisitions and Corporate Restructuring and
Recovery (Elsevier Business Intelligence). He has participated in the organisation of
leading conferences in Europe and the USA and also presented numerous research papers
at the conferences. He has published in leading journals such as the Multinational Finance
Journal and the Journal of Financial Transformation. He is a former member of the
Board of the European Finance Association.
The last few years have probably seen more articles and books on both Corporate
Governance and Corporate Finance than ever before. Whilst much has happened in the field
of Corporate Governance following a number of serious accidents such as Enron and Ahold,
there are still some very fundamental issues, which will have to be addressed. Moreover this
is a field where globalisation of rules and attitudes has not, or not yet, been achieved.
First of all there is the fundamental difference between the US approach of strict and
often detailed legislation on the one side and the European attitude that a Company
Board can deviate from the required – or, more appropriately, recommended – course of
action by giving an explanation which then has to be accepted by the shareholders. Or to
put it in the vernacular, the well known ‘comply or explain’ possibility most European
codes offer has a counterpart in ‘comply or go to jail’ on the other side of the Atlantic.
Sarbanes Oxley is a case in point and European companies with a New York listing
frequently are confronted with this.
A second issue is the question of whether more supervision by independents (however
defined) also means more effective supervision. The problem remains that in the end the
supervision of a group of relative outsiders can never be complete but will depend on the
work of other outsiders, notably the external auditors. As an aside I would like to mention
that the almost endemic inclination to change accounting conventions and definitions with
disturbing regularity has been a further impediment to clarity in reporting. Even the most
active and experienced Audit Committee will have to accept that there are limits to what
they can control. In other words, in the end the supervisors will have, above all, to judge
the integrity of the company management and in case of doubt be prepared to take action.
The third fundamental question is whether more and better supervision leads to better
company results. Here some more research seems to be in order but the few articles I
have seen are inconclusive.That is a serious matter because the new approach to corporate
government, notably Sarbanes Oxley, allegedly generates very high costs and amounts of
a $100 million and more are mentioned as the cost of introducing the system in major
multinational companies.
Many publications address other important issues, such as the legitimacy of Boards,
whom they represent, how they should be nominated/elected or how their independence
can be assured. Over the past decade the Anglo-American tradition of giving primacy of
place to the shareholder has by now in fact been accepted in most European countries,
with the probable exception of Germany, where codetermination still has in important place
in corporate governance. Another topic for discussion is the question of the requirements
for a good Board member. Integrity, transparency and activity – meaning the willingness
to take action when necessary – are probably the most important for independent
Board members. For the sake of completeness I mention the European debate on the
PREFACE xix
one-tier/two-tiers Board, an issue that, in my view, will gradually disappear with the de
facto convergence between the two systems, which one can see in practice. This would
especially be the case for UK Boards where the recommended split of the CEO and the
Board Chairmanship (duality) in practice means that the board functioning is in many
respects comparable to the Rhineland model.
Corporate Finance and notably the merger and acquisition aspect of it, has also
recently drawn more attention from the public, the experts and the lawmakers.This is not
the place to discuss the dangers and merits of the growth of private equity groups and
hedge funds, two phenomena that are rather different, though sometimes confused in
popular perception. Suffice it here to mention that one of the reasons for the rapid growth
of private equity, which is mentioned with increasing frequency, is the increasing burden
of corporate governance requirements as imposed by the authorities.
There is obviously another more direct relationship between Corporate Governance and
Corporate Finance, certainly in the case of a takeover by a financial – as opposed to an
industrial – party. The financial party will also be inclined, at least initially, to be hostile,
although in the course of the process this may gradually change by applying techniques
such as the ‘bear hug’. A threat of a takeover or acquisition by a financial consortium will
often – though not always – be a reflection of the market judgement of the performance of
the management. The bidders feel, rightly or wrongly, that they can use the company’s
assets more productively than the present management. In effect the hostile bidders
criticise by implication the attitude and judgement of the independent directors. They are
seen as responsible for maintaining and tolerating a performance that outsiders apparently
consider below the required standards. A new owner, allegedly, would be able to improve
profitability by changing the existing situation, probably by first replacing (part of)
the management and then proceeding by hiving off non-core assets, however defined.
A further step is splitting the company in parts and selling these to interested parties, the
‘buy and break’ scenario. Burdening the acquired company with considerable loans at
relatively heavy costs would then usually finance the investment.
The risk of an undesired offer from a hostile financial partner increases therefore if the
Board has not been able or willing to exercise a proper level of quality control on the
performance of the senior executives.To avoid this, the non-executive Board members have
to be absolutely independent and sufficiently courageous to take action if and when
necessary. Related to this is the problem of formulating and implementing a fair
remuneration policy, which encourages and rewards good performance but does not accept
compromises if the target results are not achieved. The recent reports on the development
of some practices with the pricing of options in the US demonstrate that not every Board
is sufficiently aware of its responsibility to hold the management to established targets.
The subjects of Corporate Governance and Corporate Finance are therefore related in
a number of interesting and often complex ways and this book makes a timely contribution
to the discussion.
FLORIS MALJERS
NOTE
* Floris Maljers is former Chairman of the Advisory Board for RSM Erasmus University and
former CEO of Unilever. With his outstanding business experience and contacts he still helps
companies like Philips, KLM, Guinness and BP in a series of non-executive roles. He is emeritus
professor of strategic management at RSM Erasmus University.
Acknowledgements
The editors and publishers wish to thank the authors and the following publishers who
have kindly given permission for the use of copyright material.
Academic Press Inc. for article: Julian Franks, Colin Mayer and Luc Renneboog (2001),
‘Who disciplines management in poorly performing companies?’, Journal of Financial
Intermediation, 10(3–4), 209–248.
Blackwell Publishing Ltd for articles: Michael C. Jensen (1993),’The modern industrial
revolution, exit and the failure of internal control systems’, Journal of Finance, 48(3),
831–880; Andrei Shleifer and Robert W. Vishny (1997), ‘A survey of corporate gover-
nance’, Journal of Finance, 52(2), 737–784; James S. Ang, Rebel Cole, and James Wuh
Lin (2000), ‘Agency costs and ownership structure’, Journal of Finance, 55(1), 81–106;
Martin J. Conyon and Kevin J. Murphy (2000), ‘The prince and the pauper? CEO pay in
the United States and United Kingdom’, The Economic Journal, 110(Nov), 640–671; Jay
Dahya, John J. McConnell and Nickolaos Travlos (2002), ‘The Cadbury Committee,
corporate performance and management turnover’, Journal of Finance, 57(1), 461–483;
Jorge Farinha (2003), ‘Dividend policy, corporate governance and the managerial
entrenchment hypothesis: an empirical analysis’, Journal of Business Finance &
Accounting, 30(9–10), 1173–1210.
Elsevier Ltd for articles: Julian Franks and Colin Mayer (1996), ‘Hostile takeovers and
the correction of managerial failure’, Journal of Financial Economics, 40(1), 163–181;
Wayne R. Guay (1999), ‘The sensitivity of CEO wealth to equity risk: an analysis of the
magnitude and determinants’, Journal of Financial Economics, 53(1), 43–71; Marco
Becht and Ailisa Röell (1999), ‘Blockholdings in Europe: an international comparison’,
European Economic Review, 43(4–6), 1049–1056; Rafael La Porta, Florencio Lopez-
de-Silanes, Andrei Shleifer and Robert Vishny (2000), ‘Investor protection and corporate
governance’, Journal of Financial Economics, 58(1–2), 3–27; Mara Faccio and
M. Ameziane Lasfer (2000), ‘Do occupational pension funds monitor companies in which
they hold large stakes?’, Journal of Corporate Finance, 6(1), 71–110; Luc Renneboog
(2000), ‘Ownership, managerial control and the governance of companies listed on the
Brussels stock exchange’, Journal of Banking & Finance, 24(12), 1959–1995; John E. Core
and David F. Larcker (2001), ‘Performance consequences of mandatory increases in exec-
utive stock ownership’, Journal of Financial Economics, 64(3), 317–340; Tim Jenkinson
and Alexander Ljungqvist (2001), ‘The role of hostile stakes in German corporate gover-
nance’, Journal of Corporate Finance, 7(4), 397–446; Paolo F. Volpin (2002),
‘Governance with poor investor protection: evidence from top executive turnover in Italy’,
Journal of Financial Economics, 64(1), 61–90; Mara Faccio and Larry H. P. Lang (2002),
ACKNOWLEDGEMENTS xxi
Multinational Finance Society for article: Gertjan Schut and Ruud van Frederikslust
(2004), ‘Shareholder wealth effects of joint venture strategies’, Multinational Finance
Journal, 8(3–4), 211–225.
Oxford University Press for article: Marc Goergen and Luc Renneboog (2003), ‘Why are
the levels of controls so different in German and UK companies? Evidence from initial
public offerings’, Journal of Law, Economics and Organization, 19(1), 141–175.
Springer Science B.V. for articles: Jim Lai and Sudi Sudarsanam (1997), ‘Corporate
restructuring in response to performance decline: impact of ownership, governance and
lenders’, European Finance Review, 1(2), 197–233; Hiroshi Osano (1999), ‘Security
design, insider monitoring, and financial market equilibrium’, European Finance Review,
2(3), 273–302; Erik Lehmann and Jürgen Weigand (2000), ‘Does the governed corpora-
tion perform better? Governance structures and corporate performance in Germany’,
European Finance Review, 4(2), 157–195.
In addition the editors and publishers wish to thank the Library of RSM Erasmus
University for their assistance in obtaining these articles.
General introduction1
The articles in this volume are organised under six subject headings:
NOTE
REFERENCES
Ang, J.S., Cole, R. and Wuh Lin, J. (2000), ‘Agency costs and ownership structure’, Journal of
Finance, 55(1): 81–106.
GENERAL INTRODUCTION 5
Bebchuk, L. and Fried, J. (2004), Pay without performance, Cambridge, Mass: Harvard University
Press.
Becht, M. and Röell, A. (1999), ‘Block holdings in Europe: an international comparison’, European
Economic Review, 43(4–6): 1049–1056.
Faccio M. and Lang, Larry H.P. (2002), ‘The ultimate ownership of Western European corporations’,
Journal of Financial Economics, 65(3): 365–395.
Goergen, M. and Renneboog, L. (2003), ‘Why are the levels of controls (so) different in German
and UK companies? Evidence from initial public offerings’, Journal of Law, Economics and
Organization, 19(1): 141–175.
Lai, J. and Sudarsanam, S. (1997), ‘Corporate restructuring in response to performance decline:
impact of ownership, governance and lenders’, European Finance Review, 1(2): 197–233.
Lehmann, E. and Weigand, J. (2000), ‘Does the governed corporation perform better? Governance
structures and corporate performance in Germany’, European Finance Review, 4(2): 157–195.
Manne, H.G. (1965), ‘Merger and the market for corporate control’, Journal of Political Economy,
73: 110–120.
Osano, H. (1999), ‘Security design, insider monitoring, and financial market equilibrium’, European
Finance Review, 2(3): 273–302.
Schut G.J. and van Frederikslust R. (2004), ‘Shareholder wealth effects of joint venture strategies’,
Multinational Finance Journal, 8(3–4): 211–225.
Shleifer A. and Vishny R.W. (1997), ‘A survey of corporate governance’, Journal of Finance, 52(2):
737–783.
Sundarsanam, P.S. (2003), Creating value from mergers and acquisitions. The challenges, an
international and integrated perspective, New York: FT Prentice Hall.
Part 1
Alternative perspectives on
corporate governance
systems
INTRODUCTION
Shleifer and Vishny survey corporate governance systems around the world covering
shareholders’ legal rights and their enforcement regime, the monitoring role of banks as
creditors, and concentrated ownership structures that result from leveraged buyouts
(LBOs). They identify three principal governance systems – US and the UK with strong
shareholder rights, Germany with permanent large shareholders and weak shareholder
rights and Japan that falls between the other two. Comparing corporate governance
systems around the world, Shleifer and Vishny conclude that legal protection of share-
holder rights in combination with large shareholders provides the elements of
effective corporate governance and cite the US, UK, Germany and Japan as examples of
such optimal combination. They note, in contrast, how weak shareholder rights and con-
centrated ownership by family interests have reduced Italian firms’ access to stock and
loan markets.
In Chapter 3, La Porta, Lopez-de-Silanes, Shleifer and Vishny follow the same legalistic
line of exposition by elevating shareholder rights protection as a central framework for
examining corporate governance systems in different countries. They define corporate
governance as a set of mechanisms by which outside investors – shareholders and creditors –
protect themselves against expropriation by insiders (i.e. large controlling shareholders
and managers). By rendering expropriation technology expensive or ineffective, the legal
mechanism increases the access corporations have to external finance. La Porta et al.
compare the efficiency of private contracts subject to enforcement by the court (the Coase
theorem, 1937) to that of laws and regulations that explicitly limit the scope for expropri-
ation and argue that even in countries with a judiciary that can arbitrate private contracts,
such laws and regulations can be efficient and add value.They classify countries into four
groups based on their legal origin – common law in the UK, the US and the British
Commonwealth, French civil law, German civil law and Scandinavian civil law. They find
that common law countries afford better investor protection than civil law countries
while, within civil law countries, Germany is more efficient than France in enforcement of
investor rights. La Porta et al. explore the impact of investor protection on ownership
structure of firms, financial development and resource allocation and conclude that the
impact is positive.They also find the classification of governance systems as bank-centred
or stock market-centred is unhelpful in evaluating their relative efficiencies or in explain-
ing the prevalent patterns of corporate financing. They advocate strong legal protection
of investor rights, a reliable enforcement mechanism and, in the absence of these, a
robust market regulator such as the SEC in the US to enhance corporate governance
that promotes financial development. The paper thus clearly establishes the linkage
between shareholder rights as an element of corporate governance and corporate
finance.
While agency cost resulting from the divorce of managerial control of a firm from its
ownership has emerged, since Jensen and Meckling’s seminal paper in 1976, as a central
problem of corporate governance and its reduction a measure of the efficiency of a
corporate governance structure, plausible proxies for agency cost have eluded empirical
researchers. As Ang et al. (Ch.4) argue, the base case of a firm free of any agency cost is
one owned and managed by the same person/s. However empirical data on such firms have
been difficult to obtain since such firms are not publicly listed and are small. Using unique
data from the National Survey of Small Business Finances in the US, the authors’ test
many of the implications of the Jensen and Meckling model. They are able to use proxies
for agency costs between owner-managed firms and firms with nonmanaging shareholders.
They find that the former are significantly more efficient in using their assets (i.e.
they display less shirking and incur significantly lower levels of operating costs, that is, they
display less perquisite consumption). Ang et al. also report that agency costs decrease as
ownership becomes more concentrated and creditors provide additional, but much
weaker, monitoring of managers.These results are consistent with the benign role of large
ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS 9
REFERENCES
Michael C. Jensen
THE MODERN INDUSTRIAL
REVOLUTION, EXIT, AND THE
FAILURE OF INTERNAL
CONTROL SYSTEMS
Source: Journal of Finance, 48(3) (1993): 831–880.
ABSTRACT
Since 1973 technological, political, regulatory, and economic forces have been changing the
worldwide economy in a fashion comparable to the changes experienced during the nineteenth
century Industrial Revolution. As in the nineteenth century, we are experiencing declining costs,
increasing average (but decreasing marginal) productivity of labor, reduced growth rates of
labor income, excess capacity, and the requirement for downsizing and exit. The last two
decades indicate corporate internal control systems have failed to deal effectively with these
changes, especially slow growth and the requirement for exit. The next several decades pose a
major challenge for Western firms and political systems as these forces continue to work their
way through the worldwide economy.
the passage of antitrust laws restricting material ruin. . . . The fruits of the toil of
combinations, and in the late 1980s the millions are boldly stolen to build up colos-
reregulation of the credit markets, anti- sal fortunes for the few, unprecedented in
takeover legislation, and court decisions the history of mankind; and the posses-
that restricted the market for corporate sors of these in turn despise the republic
control. and endanger liberty. From the same
Although the vast increases in productivity prolific womb of government injustice are
associated with the nineteenth century bred two great classes of tramps and
industrial revolution increased aggregate millionaires. (McMurray (1929), p. 7).
welfare, the large costs associated with
the obsolescence of human and physical Technological and other developments
capital generated substantial hardship, mis- that began in the mid-twentieth century
understanding, and bitterness. As noted in have culminated in the past two decades in
1873 by Henry Ward Beecher, a well-known a similar situation: rapidly improving pro-
commentator and influential clergyman of ductivity, the creation of overcapacity and,
the time, consequently, the requirement for exit.
Although efficient exit—because of the
The present period will always be memo- ramifications it has on productivity and
rable in the dark days of commerce in human welfare—remains an issue of great
America. We have had commercial dark- importance, research on the topic has been
ness at other times. There have been relatively sparse since the 1942 publication
these depressions, but none so obstinate of Schumpeter’s insights on creative
and none so universal . . . Great Britain destruction.1 These insights will almost cer-
has felt it; France has felt it; all Austria tainly receive renewed attention in the
and her neighborhood has experienced it. coming decade:
It is cosmopolitan. It is distinguished by
its obstinacy from former like periods of Every piece of business strategy acquires
commercial depression. Remedies have its true significance only against the
no effect. Party confidence, all stimulat- background of that process and within
ing persuasion, have not lifted the pall, the situation created by it. It must be
and practical men have waited, feeling seen in its role in the perennial gale of
that if they could tide over a year they creative destruction; it cannot be under-
could get along; but they could not tide stood irrespective of it or, in fact, on the
over the year. If only one or two years hypothesis that there is a perennial
could elapse they could save themselves. lull . . . The usual theorist’s paper and
The years have lapsed, and they were the usual government commission’s
worse off than they were before. What is report practically never try to see that
the matter? What has happened? Why, behavior, on the one hand, as a result of
from the very height of prosperity with- a piece of past history and, on the other
out any visible warning, without even a hand, as an attempt to deal with a situa-
cloud the size of a man’s hand visible on tion that is sure to change presently—as
the horizon, has the cloud gathered, as it an attempt by those firms to keep on
were, from the center first, spreading all their feet, on ground that is slipping
over the sky? (Price (1933), p. 6). away from under them. In other words,
the problem that is usually being visual-
On July 4, 1892, the Populist Party plat- ized is how capitalism administers existing
form adopted at the party’s first conven- structures, whereas the relevant problem
tion in Omaha reflected similar discontent is how it creates and destroys them.
and conflict: (Schumpeter (1976), p. 83).
than doubled after 1981 from 1.4 percent Indeed, Marty Lipton, prominent defender
per year in the period 1950 to 1981 to of American CEOs, expresses a common
3.3 percent in the period 1981 to 1990.2 view of the 1980s when he states that “the
Nominal unit labor costs stopped their takeover activity in the United States has
17-year rise, and real unit labor costs imposed short-term profit maximization
declined by 25 percent. These lower labor strategies on American Business at the
costs came not from reduced wages or expense of research, development and
employment, but from increased productivity: capital investment. This is minimizing our
Nominal and real hourly compensation ability to compete in world markets and
increased by a total of 4.2 and 0.3 percent still maintain a growing standard of living
per year respectively over the 1981 to at home” (Lipton (1989), p. 2).
1989 period.3 Manufacturing employment On average, selling-firm shareholders in
reached a low in 1983, but by 1989 had all M&A transactions in the period 1976 to
experienced a small cumulative increase of 1990 were paid premiums over market
5.5 percent.4 Meanwhile, the annual value of 41 percent,9 and total M&A trans-
growth in labor productivity increased from actions generated $750 billion in gains to
2.3 percent between 1950 and 1981 to target firms’ shareholders (measured in
3.8 percent between 1981 and 1990, while 1992 dollars).10 This value change repre-
a 30-year decline in capital productivity sents the minimum forecast value change
was reversed when the annual change in by the buyer (the amount the buyer is willing
the productivity of capital increased from to pay the seller), and does not include
1.03 percent between 1950 and 1981 to further gains (or losses) reaped by the
2.03 percent between 1981 and 1990.5 buyer after execution of the transaction.11
During the 1980s, the real value of It includes synergy gains from combining
public firms’ equity more than doubled the assets of two or more organizations
from $1.4 to $3 trillion.6 In addition, real and the gains from replacing inefficient
median income increased at the rate of governance systems, as well as possible
1.8 percent per year between 1982 and wealth transfers from employees, commu-
1989, reversing the 1.0 percent per year nities, and bondholders.12 As Shleifer and
decrease that occurred from 1973 to Summers (1988) point out, if the value
1982.7 Contrary to generally held beliefs, gains are merely transfers of wealth from
real research and development (R&D) creditors, employees, suppliers, or commu-
expenditures set record levels every year nities, they do not represent efficiency
from 1975 to 1990, growing at an average improvements. Thus far, however, little
annual rate of 5.8 percent.8 The Economist evidence has been found to support
(1990), in one of the media’s few accurate substantial wealth transfers from any
portrayals of this period, noted that from group,13 and it appears that most of these
1980 to 1985 “American industry went on gains represent increases in efficiency.
an R&D spending spree, with few big Part of the attack on M&A transactions
successes to show for it.” was centered on the high-yield (or so-called
Regardless of the gains in productivity, “junk”) bond market, which eliminated
efficiency, and welfare, the 1980s are mere size as an effective deterrent against
generally portrayed by politicians, the takeover. This opened the management of
media, and others as a “decade of greed America’s largest corporations to monitoring
and excess.” In particular, criticism was and discipline from the capital markets. It
centered on M&A transactions, 35,000 of also helped provide capital for newcomers
which occurred from 1976 to 1990, with a to compete with existing firms in the
total value of $2.6 trillion (1992 dollars). product markets.
Contrary to common beliefs, only 364 of High-yield bonds opened the public
these offers were contested, and of those capital markets to small, risky, and unrated
only 172 resulted in successful hostile firms across the country, and made it pos-
takeovers (Mergerstat Review (1991)). sible for some of the country’s largest firms
16 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
current capital stock and organization. they been stated explicitly, would not have
Wal-Mart and the wholesale clubs that are been acceptable to the rational investor.”
revolutionizing retailing are examples of There are clues in the history of the nine-
such change. These new, focused, large teenth century that similar overshooting
scale, low-cost retailers are dominating occurred then as well. In Jensen (1991), I
old-line department stores which can no analyze the incentive, information, and
longer compete. Building these new low-cost contracting problems that cause this over-
stores means much current retail capacity shooting and argue that these problems of
becomes obsolete—when Wal-Mart enters boom-bust cycles are general in venture
a new market total retail capacity expands, markets—but that they can be corrected by
and it is common for some of the existing reforming contracts that currently pay pro-
high-cost retail capacity to go out of busi- moters for doing deals, rather than for
ness.15 More intensive use of information doing successful deals.
and other technologies, direct dealing with
manufacturers, and the replacement of
high-cost, restrictive work-rule union labor Current forces leading to excess
are several sources of the competitive capacity and exit
advantage of these new organizations.
The ten-fold increase in crude oil prices
Finally, excess capacity also results when
between 1973 and 1979 had ubiquitous
many competitors simultaneously rush to
effects, forcing contraction in oil, chemi-
implement new, highly productive technologies
cals, steel, aluminum, and international
without considering whether the aggregate
shipping, among other industries. In addi-
effects of all such investment will be
tion, the sharp crude oil price increases
greater capacity than can be supported by
that motivated major changes to econo-
demand in the final product market.
mize on energy had other, perhaps even
Sahlman and Stevenson’s (1985) analysis
larger, implications. I believe the reevaluation
of the winchester disk drive industry
of organizational processes and procedures
provides an example of this phenomenon.
stimulated by the oil shock also generated
Between 1977 and 1984, venture capitalists
dramatic increases in efficiency beyond the
invested over $400 million in 43 different
original pure energy-saving projects. The
manufacturers of winchester disk drives;
original energy-motivated reexamination of
initial public offerings of common stock
corporate processes helped initiate a major
infused additional capital in excess of $800
reengineering of company practices and
million. In mid-1983, the capital markets
procedures that still continues to accelerate
assigned a value of $5.4 billion to 12 pub-
throughout the world.
licly traded, venture-capital-backed hard
Since the oil price increases of the
disk drive manufacturers—yet by the end
1970s, we again have seen systematic
of 1984, the value assigned to those com-
overcapacity problems in many industries
panies had plummeted to $1.4 billion. In
similar to those of the nineteenth century.
his study of the industry, Christensen
While the reasons for this overcapacity
(1993) finds that over 138 firms entered
nominally differ among industries, I doubt
the industry in the period from its invention
they are independent phenomena. We do
in 1956 to 1990, and of these 103 subse-
not yet fully understand all the causes pro-
quently failed and six were acquired.
pelling the rise in excess capacity in the
Sahlman and Stevenson (p. 7) emphasize
1980s, yet I believe there were a few basic
the lack of foresight in the industry: “The
forces in operation.
investment mania visited on the hard
disk industry contained inherent assump-
tions about the long-run industry size Macro policies
and profitability and about future growth,
profitability and access to capital for each Major deregulation of the American econ-
individual company.These assumptions, had omy (including trucking, rail, airlines,
18 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
telecommunications, banking and financial produce boxes with 1,000 times the capacity
services industries) under President Carter for a given price. Consequently, computers
contributed to the requirements for exit in are becoming commonplace—in cars,
these industries,16 as did important toasters, cameras, stereos, ovens, and so on.
changes in the U.S. tax laws that reduced Nevertheless, the increase in quantity
tax advantages to real estate development, demanded has not been sufficient to avoid
construction, and other activities.The end of overcapacity, and we are therefore witness-
the cold war has had obvious ramifications ing a dramatic shutdown of production
for the defense industry, as well as less lines in the industry—a force that has
direct effects on the industry’s suppliers. wracked IBM as a high-cost producer.
In addition, two generations of managerial A change of similar magnitude in auto pro-
focus on growth as a recipe for success duction technology would have reduced the
caused many firms, I believe, to overshoot price of a $20,000 auto in 1980 to under
their optimal capacity, setting the stage for $20 today. Such increases in capacity and
cutbacks, especially in white collar corporate productivity in a basic technology have
bureaucracies. Specifically, in the decade unavoidably massive implications for the
from 1979 to 1989 the Fortune 100 firms organization of work and society.
lost 1.5 million employees, or 14 percent of Fiberoptic and other telecommunica-
their workforce.17 tions technologies such as compression
algorithms are bringing about similarly
vast increases in worldwide capacity and
Technology functionality. A Bell Laboratories study of
Massive changes in technology are clearly excess capacity indicates, for example, that
part of the cause of the current industrial given three years and an additional expen-
revolution and its associated excess diture of $3.1 billion, three of AT&T’s new
capacity. Both within and across industries, competitors (MCI, Sprint, and National
technological developments have had far- Telecommunications Network) would be
reaching impact. To give some examples, able to absorb the entire long distance
the widespread acceptance of radial tires switched service that was supplied by
(lasting three to five times longer than the AT&T in 1990 (Federal Communications
older bias ply technology and providing Commission (1991), p. 1140).
better gas mileage) caused excess capacity
in the tire industry; the personal computer
Organizational innovation
revolution forced contraction of the market
for mainframes; the advent of aluminum Overcapacity can be caused not only by
and plastic alternatives reduced demand changes in the physical technology, but
for steel and glass containers; and fiberop- also by changes in organizational prac-
tic, satellite, digital (ISDN), and new tices and management technology. The
compression technologies dramatically vast improvements in telecommunications,
increased capacity in telecommunication. including computer networks, electronic
Wireless personal communication such as mail, teleconferencing, and facsimile
cellular phones and their replacements transmission are changing the workplace in
promise to further extend this dramatic major ways that affect the manner in which
change. people work and interact. It is far less
The changes in computer technology, valuable for people to be in the same
including miniaturization, have not only geographical location to work together
revamped the computer industry, but also effectively, and this is encouraging smaller,
redefined the capabilities of countless other more efficient, entrepreneurial organizing
industries. Some estimates indicate the units that cooperate through technology.19
price of computing capacity fell by a factor This encourages even more fundamental
of 1,000 over the last decade.18 This means changes. Through competition “virtual
that computer production lines now organizations”—networked or transitory
THE MODERN INDUSTRIAL REVOLUTION 19
a Population statistics from Monthly Bulletin of Statistics (United Nations, 1993), 1991 data.
b Potential labor force estimated by applying the 40.4 percent labor force participation rate in the European
Economic Community to the 1991 population estimates, using the most recent employment estimates
(Statistical Yearbook, United Nations, 1992) for each member country.
c Unless otherwise noted, refers to 1991 earnings from the Monthly Bulletin of Statistics (United Nations,
1993) or earnings from Statistical Yearbook (United Nations, 1992) adjusted to 1991 levels using the
Consumer Price Index. Earnings for Poland were calculated using 1986 earnings and 1986 year-end exchange
rate, while earnings for Romania were calculated using 1985 earnings and 1985 exchange rate. An approxima-
tion for the former U.S.S.R. was made using 1987 data for daily earnings in the U.S.S.R. and the estimated
1991 exchange rate for the former U.S.S.R. from the Monthly Bulletin of Statistics.
d NA Not available. In the case of Yugoslavia, inflation and currency changes made estimates unreliable. For
Indonesia, Vietnam, Malaysia, and Taiwan data on earnings in manufacturing are unavailable.
e Average daily wage weighted according to projected labor force in each grouping.
THE MODERN INDUSTRIAL REVOLUTION 21
European Economic Community average Act, which will remove trade barriers
wage is about $80 per day with a total between Canada, the United States, and
labor force of about 130 million. The labor Mexico) is but one general example of
forces that have affected world trade conflicts that are also occurring in the
extensively in the last several decades total steel, automobile, computer chip, computer
only about 90 million (Hong Kong, Japan, screen, and textile industries. In addition it
Korea, Malaysia, Singapore, and would not be surprising to see a return to
Taiwan).26 demands for protection from even domestic
While the changes associated with bringing competition. This is currently underway in
a potential 1.2 billion low-cost laborers the deregulated airline industry, an industry
onto world markets will significantly that is faced with significant excess capacity.
increase average living standards through- We should not underestimate the strains
out the world, they will also bring massive this continuing change will place on
obsolescence of capital (manifested in the worldwide social and political systems. In
form of excess capacity) in Western both the First and Second Industrial
economies as the adjustments sweep Revolutions, the demands for protection
through the system. Western managers from competition and for redistribution of
cannot count on the backward nature of income became intense. It is conceivable
these economies to limit competition from that Western nations could face the modern
these new human resources. Experience in equivalent of the English Luddites who
China and elsewhere indicates the problems destroyed industrial machinery (primarily
associated with bringing relatively current knitting frames) in the period 1811 to
technology on line with labor forces in these 1816, and were eventually subdued by the
areas is possible with fewer difficulties than militia (Watson (1993)). In the United
one might anticipate.27 States during the early 1890s, large groups
One can confidently forecast that the of unemployed men (along with some
transition to open capitalist economies will vagrants and criminals), banding together
generate great conflict over international under different leaders in the West,
trade as special interests in individual Midwest, and East, wandered cross-country
countries try to insulate themselves from in a march on Congress. These “industrial
competition and the required exit.The tran- armies” formed to demand relief from “the
sition of these economies will require large evils of murderous competition; the sup-
redirection of Western labor and capital to planting of manual labor by machinery; the
activities where it has a comparative excessive Mongolian and pauper immigration;
advantage. While the opposition to global the curse of alien landlordism . . .”
competition will be strong, the forces are (McMurray (1929), p. 128). Although the
likely to be irresistible in this day of rapid armies received widespread attention and
and inexpensive communication, trans- enthusiasm at the onset, the groups were
portation, miniaturization, and migration. soon seen as implicit threats as they
The bottom line, of course, is that with roamed from town to town, often stealing
even more excess capacity and the require- trains and provisions as they went. Of the
ment for additional exit, the strains put on 100,000 men anticipated by Coxey, only
the internal control mechanisms of Western 1,000 actually arrived in Washington to
corporations are likely to worsen for protest on May 1, 1893. At the request of
decades to come. the local authorities, these protesters
In the 1980s managers and employees disbanded and dispersed after submitting a
demanded protection from the capital petition to Congress (McMurray (1929),
markets. Many are now demanding protec- pp. 253–262).
tion from international competition in the We need look no further than central and
product markets (often under the guise of eastern Europe or Asia to see the effects of
protecting jobs). The current dispute over policies that protect organizations from
the NAFTA (North American Free Trade foreign and domestic competition. Hundreds
22 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
of millions of people have been condemned this way, exit is significantly delayed at
to poverty as a result of governmental substantial cost of real resources to soci-
policies that protect firms from competition ety. The tire industry is an example.
in the product markets (both domestic and Widespread consumer acceptance of radial
foreign) and attempt to ensure prosperity tires meant that worldwide tire capacity
and jobs by protecting organizations had to shrink by two-thirds (because
against decline and exit. Such policies are radials last three to five times longer than
self-defeating, as employees of state-owned bias ply tires). Nonetheless, the response by
factories in these areas are now finding. the managers of individual companies was
Indeed, Porter (1990) finds that the most often equivalent to: “This business is going
successful economies are those blessed through some rough times. We have to
with intense internal competition that make major investments so that we will
forces efficiency through survival of the have a chair when the music stops.”
fittest. A. William Reynolds (1988), Chairman and
Our own experience in the 1980s demon- CEO of GenCorp (maker of General Tires),
strated that the capital markets can also illustrates this reaction in his testimony
play an important role—that capital before the Subcommittee on Oversight and
market pressures, while not perfect, can Investigations (February 18, 1988), U.S.
significantly increase efficiency by bringing House Committee on Energy and
about earlier adjustments. Earlier Commerce:
adjustments avoid much of the waste
generated when failure in the product The tire business was the largest piece of
markets forces exit. GenCorp, both in terms of annual rev-
enues and its asset base.Yet General Tire
was not GenCorp’s strongest performer.
IV. THE DIFFICULTY OF EXIT Its relatively poor earnings performance
was due in part to conditions affecting
The asymmetry between growth and all of the tire industry. . . . In 1985
decline worldwide tire manufacturing capacity
substantially exceeded demand. At the
Exit problems appear to be particularly same time, due to a series of technological
severe in companies that for long periods improvements in the design of tires and
enjoyed rapid growth, commanding market the materials used to make them, the
positions, and high cash flow and profits. In product life of tires had lengthened
these situations, the culture of the organi- significantly. General Tire, and its com-
zation and the mindset of managers seem petitors, faced an increasing imbalance
to make it extremely difficult for adjust- between supply and demand. The eco-
ment to take place until long after the nomic pressure on our tire business was
problems have become severe, and in some substantial. Because our unit volume
cases even unsolvable. In a fundamental was far below others in the industry, we
sense, there is an asymmetry between the had less competitive flexibility. . . . We
growth stage and the contraction made several moves to improve our
stage over the life of a firm. We have spent competitive position: We increased our
little time thinking about how to manage investment in research and development.
the contracting stage efficiently, or more We increased our involvement in the high
importantly how to manage the growth performance and light truck tire cate-
stage to avoid sowing the seeds of decline. gories, two market segments which
In industry after industry with excess offered faster growth opportunities. We
capacity, managers fail to recognize that developed new tire products for those
they themselves must downsize; instead segments and invested heavily in an
they leave the exit to others while they con- aggressive marketing program designed
tinue to invest. When all managers behave to enhance our presence in both markets.
THE MODERN INDUSTRIAL REVOLUTION 23
presumably, this is a major reason for their hostile tender offers in the 1970s. Prior to
existence. the 1970s capital market discipline took
Indeed the gradual abrogation of the place primarily through the proxy process.
legal notion of “at will” employment is (Pound (1993) analyzes the history of the
coming close to granting property rights in political model of corporate control.)
jobs to all employees.29 While casual The legal/political/regulatory system is
breach of implicit contracts will destroy far too blunt an instrument to handle the
trust in an organization and seriously problems of wasteful managerial behavior
reduce efficiency, all organizations must effectively. (The breakup and deregulation
evolve a way to change contracts that are of AT&T, however, is one of the court
no longer optimal. For example, bank- system’s outstanding successes. As we shall
ruptcy is essentially a state-supervised see below, it helped create over $125 billion
system for breaking (or more politely, of increased value between AT&T and the
rewriting) contracts that are mutually Baby Bells.)
inconsistent and therefore, unenforceable. While the product and factor markets are
All developed economies evolve such a slow to act as a control force, their
system. Yet, the problem is a very general discipline is inevitable—firms that do
one, given that the optimality of changing not supply the product that customers
contracts must be one of the major reasons desire at a competitive price cannot survive.
for leaving many of them implicit. Research Unfortunately, when product and factor
into the optimal breach of contracts, and the market disciplines take effect it can often
bonding against opportunistic behavior be too late to save much of the enterprise.
that must accompany it, is an important topic To avoid this waste of resources, it is impor-
that has received considerable attention tant for us to learn how to make the other
in the law and economics literature (see three organizational control forces more
Polinsky (1989) ) but is deserving of more expedient and efficient.
attention by organization theorists. Substantial data support the proposition
that the internal control systems of publicly
held corporations have generally failed to
V. THE ROLE OF THE MARKET FOR cause managers to maximize efficiency and
CORPORATE CONTROL value.30 More persuasive than the formal
statistical evidence is the fact that few
firms ever restructure themselves or
The four control forces operating on engage in a major strategic redirection
the corporation without a crisis either in the capital
There are only four control forces operating markets, the legal/political/regulatory
on the corporation to resolve the problems system, or the product/factor markets. But
caused by a divergence between managers’ there are firms that have proved to be flex-
decisions and those that are optimal from ible in their responses to changing market
society’s standpoint. They are the conditions in an evolutionary way. For
example, investment banking firms and
● capital markets, consulting firms seem to be better at
● legal/political/regulatory system, responding to changing market conditions.
● product and factor markets, and
● internal control system headed by Capital markets and the market for
the board of directors. corporate control
As explained elsewhere (Jensen (1989a, The capital markets provided one mecha-
1989b, 1991), Roe (1990, 1991)), the nism for accomplishing change before
capital markets were relatively constrained losses in the product markets generate a
by law and regulatory practice from about crisis. While the corporate control activity
1940 until their resurrection through of the 1980s has been widely criticized as
THE MODERN INDUSTRIAL REVOLUTION 25
took more than ten years to accomplish the generated cash flows. For example, consider
change leaves serious questions about the a firm that provides dividends plus capital
social costs of continuing the waste caused gains to its shareholders over a ten-year
by ineffective control. It appears that period that equal the cost of capital on the
internal control systems have two faults. beginning of period share value. Suppose,
They react too late, and they take too long however, that management spent $30 billion
to effect major change. Changes motivated of internally generated cash flow on R&D
by the capital market are generally accom- and capital expenditures that generated no
plished quickly—within one and a half to returns. In this case the firm’s shareholders
three years. As yet no one has demon- suffered an opportunity loss equal to the
strated the social benefit from relying on value that could have been created if the
purely internally motivated change that firm had paid the funds out to them and they
offsets the costs of the decade-long delay had invested it in equivalently risky projects.
exhibited by General Mills. The opportunity cost of R&D and capital
In summary, it appears that the infrequency expenditures thus can be thought of as the
with which large corporate organizations returns that would have been earned by an
restructure or redirect themselves solely investment in equivalent-risk assets over
on the basis of the internal control the same time period. We don’t know
mechanisms in the absence of crises in the exactly what the risk is, nor what the
product, factor, or capital markets or expected returns would be, but we can
the regulatory sector is strong testimony to make a range of assumptions. A simple
the inadequacy of these control mechanisms. measure of performance would be the dif-
ference between the total value of the R&D
plus capital and acquisition expenditures
VII. DIRECT EVIDENCE OF THE invested in a benchmark strategy and the
FAILURE OF INTERNAL CONTROL total value the firm actually created with
SYSTEMS its investment strategy. The benchmark
strategy can be thought of as the ending
value of a comparable-risk bank account
The productivity of R&D and capital
(with an expected return of 10 percent)
expenditures into which the R&D and capital expendi-
The control market, corporate restructurings, tures in excess of depreciation (hereafter
and financial distress provide substantial referred to as net capital expenditures) had
evidence on the failings of corporate internal been deposited instead of invested in real
control systems. My purpose in this section projects. For simplicity I call this the
is to provide another and more direct benchmark strategy. The technical details
estimate of the effectiveness of internal con- of the model are given in the Appendix.The
trol systems by measuring the productivity of calculation of the performance measure
corporate R&D and capital expenditures. takes account of all stock splits, stock
The results reaffirm that many corporate dividends, equity issues and repurchases,
control systems are not functioning well. dividends, debt issues and payments, and
While it is impossible to get an unambigu- interest.
ous measure of the productivity of R&D
and capital expenditures, by using a period Three measures of the productivity of
as long as a decade we can get some R&D and net capital expenditures
approximations. We cannot simply measure
the performance of a corporation by Measure 1
the change in its market value over time
(more precisely the returns to its share- Consider an alternative strategy which pays
holders) because this measure does not the same dividends and stock repurchases
take account of the efficiency with which as the firm actually paid (and raises the
the management team manages internally same outside capital) and puts the R&D
28 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
and capital and acquisition expenditures like a one-horse shay, the firm arrives at the
(in excess of depreciation) in marketable end of the period still generating cash
securities of the same risk as the R&D and returns, but then collapses with no addi-
capital expenditures, yielding expected tional cash payments to equityholders, and
returns equal to their cost of capital, i. equity value of zero as of the horizon date.
Under the assumption that the zero invest-
ment and R&D strategy yields a terminal
value of the firm equal to the ending debt Measure 3
plus the beginning value of equity (that is, To allow for the effects of the reduced
investment equal to depreciation is suffi- investment and R&D on intermediate cash
cient to maintain the original equity value flows my third measure assumes that all
of the firm), Measure 1 is the difference intermediate cash flows are reduced in the
between the actual ending total value of benchmark investment strategy by
the firm and the value of the benchmark. the amount paid out to shareholders in
The exact equation is given in the Appendix the form of dividends and net share
for this measure as well as the next two repurchases and that the original value of
measures of performance. the equity is maintained. This measure is
Unless capital and R&D expenditures likely to yield an upward biased estimate of
are completely unproductive, this first the productivity of R&D and capital
crude measure of the productivity of R&D expenditures.
and capital expenditures will be biased
downward. I define two additional meas-
ures that use different assumptions about The data and results
the effect of the reduced R&D and capital
expenditures on the ending value of the The data for this analysis consist of all 432
firm’s equity and on the ability of the firm firms on COMPUSTAT with 1989 sales of
to make the intermediate cash dividend and $250 million or more for which complete
stock repurchase payouts to shareholders. data on R&D, capital expenditures, depre-
If R&D is required to maintain a competitive ciation, dividends, and market value were
position in the industry, the ending value of available for the period December 31, 1979
the equity in the benchmark strategy is through December 31, 1990.The estimates
likely to be less than the initial value of of the productivity of R&D are likely to be
equity even though nominal depreciation of upward biased because the selection crite-
the capital stock is being replaced. ria use only firms that managed to survive
Moreover, with no R&D and maintenance through the period and eliminate those that
only of the nominal value of the capital failed. I have calculated results for various
stock, the annual cash flows from opera- rates of interest but report only those using
tions are also likely to be lower than those a 10 percent rate of return.This rate is prob-
actually realized (because organizational ably lower than the cost of capital for R&D
efficiency and product improvement will expenditures at the beginning of the period
lag competitors, and new product introduc- when interest rates were in the high teens,
tion will be lower). Therefore I use two and probably about right or on the high side
more conservative measures that will yield at the end of the period when the cost of
higher estimates of the productivity of capital was probably on the order of 8 to 10
these expenditures. percent. A low approximation of the cost of
capital appropriate to R&D and capital
expenditures will bias the performance
Measure 2 measures up, so I am reasonably comfort-
able with these conservative assumptions.
The second measure assumes that Because they are interesting in their own
replacement of depreciation and zero right,Table 1.2 presents the data on annual
expenditures on R&D are sufficient to R&D and capital expenditures of nine
maintain the intermediate cash flows but, selected Fortune 500 corporations and the
THE MODERN INDUSTRIAL REVOLUTION 29
Table 1.2 Total R&D and Capital Expenditures for Selected Companies and the Venture Capital
Industry, 1980–1990 ($ Billions)
Venture
Capital
Year GM IBM Xerox Kodak Intel GE Industry Merck AT&T
Total R&D Expenditures
1980 2.2 1.5 0.4 0.5 0.1 0.8 0.6 0.2 0.4
1981 2.2 1.6 0.5 0.6 0.1 0.8 1.2 0.3 0.5
1982 2.2 2.1 0.6 0.7 0.1 0.8 1.5 0.3 0.6
1983 2.6 2.5 0.6 0.7 0.1 0.9 2.6 0.4 0.9
1984 3.1 3.1 0.6 0.8 0.2 1.0 2.8 0.4 2.4
1985 4.0 3.5 0.6 1.0 0.2 1.1 2.7 0.4 2.2
1986 4.6 4.0 0.7 1.1 0.2 1.3 3.2 0.5 2.3
1987 4.8 4.0 0.7 1.0 0.3 1.2 4.0 0.6 2.5
1988 5.3 4.4 0.8 1.1 0.3 1.2 3.9 0.7 2.6
1989 5.8 5.2 0.8 1.3 0.4 1.3 3.4 0.8 2.7
1990 5.9 4.9 0.9 1.3 0.5 1.5 1.9 0.9 2.4
Total 42.7 36.8 7.1 10.1 2.5 11.9 27.8 5.4 19.3
Total Capital Expenditures
1980 7.8 6.6 1.3 0.9 0.2 2.0 NA 0.3 17.0
1981 9.7 6.8 1.4 1.2 0.2 2.0 NA 0.3 17.8
1982 6.2 6.7 1.2 1.5 0.1 1.6 NA 0.3 16.5
1983 4.0 4.9 1.1 0.9 0.1 1.7 NA 0.3 13.8
1984 6.0 5.5 1.3 1.0 0.4 2.5 NA 0.3 3.5
1985 9.2 6.4 1.0 1.5 0.2 2.0 NA 0.2 4.2
1986 11.7 4.7 1.0 1.4 0.2 2.0 NA 0.2 3.6
1987 7.1 4.3 0.3 1.7 0.3 1.8 NA 0.3 3.7
1988 6.6 5.4 0.5 1.9 0.5 3.7 NA 0.4 4.0
1989 9.1 6.4 0.4 2.1 0.4 5.5 NA 0.4 3.5
1990 10.1 6.5 0.4 2.0 0.7 2.1 NA 0.7 3.7
Total 87.5 64.2 9.9 16.1 3.3 27.0 NA 3.7 91.2
Net Capital Expenditures (Capital Expenditures less Depreciation)
1980 3.6 3.8 0.5 0.5 0.1 1.2 NA 0.2 9.9
1981 5.3 3.5 0.6 0.7 0.1 1.1 NA 0.2 9.9
1982 1.7 3.1 0.4 0.9 0.1 0.6 NA 0.2 7.7
1983 1.1 1.3 0.3 0.2 0.1 0.6 NA 0.1 3.9
1984 1.1 2.3 0.5 0.2 0.3 1.3 NA 0.1 0.7
1985 3.5 3.4 1.2 0.6 0.1 0.8 NA 0.07 0.9
1986 5.6 1.3 0.2 0.5 0.0 0.6 NA 0.04 3.2
1987 0.8 0.7 0.3 0.6 0.1 0.2 NA 0.07 0.6
1988 0.7 1.5 0.2 0.7 0.3 0.3 NA 0.2 5.9
1989 2.0 2.2 0.2 0.8 0.2 0.7 NA 0.2 1.1
1990 2.7 2.3 0.3 0.7 0.4 0.6 NA 0.4 0.3
Total 24.5 25.4 2.7 6.4 1.8 8.0 NA 1.8 24.7
Total Value of R&D plus Net Capital Expenditures
67.2 62.2 9.8 16.7 4.3 19.9 27.8 7.2 44.0
Ending Equity Value of the Company, 12/90
26.2 64.6 3.2 13.5 13.5 50.0 60 34.8 32.9
NA Not available. Source: Annual reports, COMPUSTAT, Business Week R&D Scoreboard, William
Sahlman. Venture Economics for total disbursements by industry. Capital expenditures for the venture capital
industry are included in the R&D expenditures which are the total actual disbursements by the industry.
30 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
total venture capital industry from January had done this (and not changed the companies
1, 1980 through December 31, 1990.Table in any way), GM would have owned two of
1.3 contains calculations that provide some the world’s low-cost, high-quality automobile
benchmarks for evaluating the productivity producers.
of these expenditures. As Table 1.3 shows, the difference
Total R&D expenditures over the eleven- between the value of GM’s actual strategy
year period range from $42.7 billion for and the value of the equivalent-risk bank
General Motors to $5.4 billion for Merck. account strategy amounts to $ –100.7
The individual R&D expenditures of GM billion by Measure 2 (which assumes the
and IBM were significantly greater than the ending value of the company given no R&D
$27.8 billion spent by the entire U.S. venture or net capital expenditures is zero in the
capital industry over the eleven-year period. benchmark strategy), $ –115.2 billion for
Because venture capital data include both Measure 1 (which assumes the original
the R&D component and capital expendi- value of the equity is maintained), and
tures, we must add in corporate capital $ –90 billion by Measure 3 (which assumes
expenditures to get a proper comparison to cash flows fall by the amount of all interme-
the venture industry figures. Total capital diate cash outflows to shareholders and
expenditures range from $91.2 billion for debtholders). I concentrate on Measure 2
AT&T and $87.5 billion for GM to $3.7 which I believe is the best measure of the
billion for Merck. Capital expenditures net three. By this measure, IBM lost over $11
of depreciation range from $25.4 billion for billion relative to the benchmark strategy
IBM to $1.8 billion for Merck. (and this is prior to the $50 billion decline in
It is clear that GM’s R&D and invest- its equity value in 1991 and 1992), while
ment program produced massive losses.The Xerox and Kodak were down $8.4 billion
company spent a total of $67.2 billion in and $4.6 billion respectively. GE and Merck
excess of depreciation in the period and were major success stories, with value
produced a firm with total ending value of creation in excess of the benchmark strategy
equity (including the E and H shares) of of $29.9 billion and $28 billion respectively.
$26.2 billion. Ironically, its expenditures AT&T gained $2.1 billion over the bench-
were more than enough to pay for the mark strategy, after having gone through the
entire equity value of Toyota and Honda, court-ordered breakup and deregulation of
which in 1985 totaled $21.5 billion. If it the Bell system in 1984. The value gains of
Table 1.3 Benefit-Cost Analysis of Corporate R&D and Investment Programs: Actual Total Value
of Company at 12/31/90 Less Total Value of the Benchmark Strategy r 10 percent
(billions of dollars)
Venture
Capital
GM IBM Xerox Kodak Intel GE Industry Merck AT&T
the seven Baby Bells totaled $125 billion by Table 1.5 provides summary statistics
Measure 2 (not shown in the table), making (including the minimum, mean, five frac-
the breakup and deregulation a nontrivial tiles of the distribution, maximum, and
success given that prices to consumers have standard deviation) on R&D expenditures,
generally fallen in the interim. net capital expenditures, and the three
The value created by the venture capital performance measures. The mean ten-year
industry is difficult to estimate. We would R&D and net capital expenditures are
like to have estimates of the 1990 total $1.296 billion and $1.367 billion respectively;
end-of-year value of all companies funded the medians are $146 million and $233
during the eleven-year period. This value is million. The average of Measure 2 over all
not available so I have relied on the $60 432 firms is slightly over $1 billion with a
billion estimate of the total value of all t-value of 3.0, indicating that on average
IPOs during the period. This overcounts this sample of firms created value above
those firms that were funded prior to 1980 that of the benchmark strategy. The
and counts as zero all those firms that had average for Measures 1 and 2 are $ –221
not yet come public as of 1990. Because of million and $1.086 billion respectively. All
the pattern of increasing investment over productivity measures are upward biased
the period from the mid-1970s, the over- because failed firms are omitted from the
counting problem is not likely to be as sample, and because the decade of the
severe as the undercounting problem. Thus, eighties was an historical outlier in stock
the value added by the industry over the market performance. The median perform-
bank account strategy is most probably ance measures are $24 million, $200
greater than $17 billion as shown in Table million, and $206 million respectively. The
1.3. Since the venture capital industry is in maximum performance measures range
Table 1.3 as another potential source of from $37.7 billion to $47 billion.
comparison, and since virtually its entire Although the average performance
value creation is reflected in its ending measures are positive, well-functioning
equity value, I have recorded its value cre- internal control systems would substan-
ation under each measure as the actual tially truncate the lower tail of the distri-
estimate of greater than $17 billion. bution. And given that the sample is subject
Because the extreme observations in the to survivorship bias,36 and that the period
distribution are the most interesting, Table was one in which stock prices performed
1.4 gives the three performance measures historically above average, the results
for the 35 companies at the bottom of the demonstrate major inefficiencies in the
list of 432 firms ranked in reverse order on capital expenditure and R&D spending
Measure 2 (Panel A), and on the 35 com- decisions of a substantial number of
panies at the top of the ranked list (Panel firms.37 I believe we can improve these
B), also in reverse order. As the tables control systems substantially, but to do so
show, GM ranked at the bottom of the we must attain a detailed understanding of
performance list, preceded by Ford, British how they work and the factors that lead to
Petroleum, Chevron, Du Pont, IBM, Unisys, their success or failure.
United Technologies, and Xerox. Obviously
many of the United States’ largest and
best-known companies appear on this list VIII. REVIVING INTERNAL
(including GD prior to its recent turn- CORPORATE CONTROL SYSTEMS
around), along with Japan’s Honda Motor
company. Panel B shows that Philip Morris Remaking the board as an effective
created the most value in excess of control mechanism
the benchmark strategy, followed by
Wal-Mart, Bristol Myers, GE, Loews, The problems with corporate internal
Merck, Bellsouth, Bell Atlantic, Procter & control systems start with the board of
Gamble, Ameritech, and Southwestern Bell.35 directors. The board, at the apex of the
32 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Table 1.4 Difference between Value of Benchmark Strategy for Investing R&D and Net Capital
Expenditure and Actual Strategy under Three Assumptions regarding Ending Value of
Equity and Intermediate Cash Flows for Benchmark Strategy (Performance Measures 1–3)
Panel A: Performance measures for the 35 companies at the bottom of the ranked list of 432
companies in the period 1980–1990 on performance measure 2. r 10 percent
Panel B: Performance measures for the 35 companies ranked at the top of the list of 432
companies in the period 1980–1990 on performance measure 2. r 10 percent.
Table 1.5 Summary Statistics on R&D, Capital Expenditures, and Performance Measures for 432
Firms with Sales Greater than $250 Million in the Period 1980–1990 r 10 percent
(millions of dollars)
internal control system, has the final and to provide high-level counsel. Few
responsibility for the functioning of the boards in the past decades have done this
firm. Most importantly, it sets the rules of job well in the absence of external crises.
the game for the CEO.The job of the board This is particularly unfortunate given that
is to hire, fire, and compensate the CEO, the very purpose of the internal control
34 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
mechanism is to provide an early warning officer.’ I should have known right there
system to put the organization back on that he wasn’t going to pay a goddamn
track before difficulties reach a crisis bit of attention to anything I said.” So it
stage. The reasons for the failure of the turned out, and after a year Hanley quit
board are not completely understood, but the board in disgust.
we are making progress toward under-
standing these complex issues. The The result is a continuing cycle of
available evidence does suggest that CEOs ineffectiveness: by rewarding consent and
are removed after poor performance,38 but discouraging conflicts, CEOs have the
the effect, while statistically significant, power to control the board, which in turn
seems too late and too small to meet the ultimately reduces the CEO’s and the
obligations of the board. I believe bad company’s performance. This downward
systems or rules, not bad people, underlie spiral makes the resulting difficulties
the general failings of boards of directors. likely to be a crisis rather than a series of
Some caution is advisable here because small problems met by a continuous
while resolving problems with boards can self-correcting mechanism. The culture of
cure the difficulties associated with a boards will not change simply in response
nonfunctioning court of last resort, this to calls for change from policy makers,
alone cannot solve all the problems with the press, or academics. It only will
defective internal control systems. I resist follow, or be associated with, general
the temptation in an already lengthy paper recognition that past practices have
to launch into a discussion of other organi- resulted in major failures and substantive
zational and strategic issues that must be changes in the rules and practices governing
attacked. A well-functioning board, however, the system.
is capable of providing the organizational
culture and supporting environment for a Information problems
continuing attack on these issues.
Serious information problems limit the
Board culture effectiveness of board members in the typical
large corporation. For example, the CEO
Board culture is an important component almost always determines the agenda and
of board failure. The great emphasis on the information given to the board. This
politeness and courtesy at the expense of limitation on information severely hinders
truth and frankness in boardrooms is both the ability of even highly talented board
a symptom and cause of failure in the members to contribute effectively to the
control system. CEOs have the same inse- monitoring and evaluation of the CEO and
curities and defense mechanisms as other the company’s strategy.
human beings; few will accept, much less Moreover, the board requires expertise
seek, the monitoring and criticism of an to provide input into the financial aspects
active and attentive board. Magnet (1992, of planning—especially in forming the
p. 86) gives an example of this environ- corporate objective and determining the
ment. John Hanley, retired Monsanto CEO, factors which affect corporate value.
accepted an invitation from a CEO Yet such financial expertise is generally
lacking on today’s boards. Consequently,
to join his board—subject, Hanley wrote, boards (and management) often fail to
to meeting with the company’s general understand why long-run market value
counsel and outside accountants as a maximization is generally the privately and
kind of directorial due diligence. Says socially optimal corporate objective, and
Hanley: “At the first board dinner the they often fail to understand how to
CEO got up and said, ‘I think Jack was a translate this objective into a feasible foun-
little bit confused whether we wanted him dation for corporate strategy and operating
to be a director or the chief executive policy.
THE MODERN INDUSTRIAL REVOLUTION 35
democracies (whether at the local, state, or process of hiring, firing, evaluating, and
federal level), or at their management of compensating the CEO. Clearly, the CEO
quasi-business organizations such as the cannot perform this function apart from his
Post Office, schools, or power generation or her personal interest. Without the direc-
entities such as the TVA. This does not mean, tion of an independent leader, it is much
however, that the current corporate system more difficult for the board to perform its
is satisfactory as it stands; indeed there is critical function.Therefore, for the board to
significant room for rethinking and revision. be effective, it is important to separate the
For example, proxy regulations by the CEO and chairman positions.41 The inde-
SEC make the current process far less pendent chairman should, at a minimum, be
efficient than it otherwise could be. given the rights to initiate board appoint-
Specifically, it has been illegal for any ments, board committee assignments, and
shareholder to discuss company matters (jointly with the CEO) the setting of the
with more than ten other shareholders board’s agenda. All these recommenda-
without prior filing with, and approval of, tions, of course, will be made conditional on
the SEC. The November 1992 relaxation the ratification of the board.
of this restriction allows an investor to An effective board will often evidence
communicate with an unlimited number of tension among its members as well as with
other stockholders provided the investor the CEO. But I hasten to add that I am not
owns less than 5 percent of the shares, has advocating continuous war in the boardroom.
no special interest in the issue being In fact, in well-functioning organizations the
discussed, and is not seeking proxy authority. board will generally be relatively inactive
These restrictions still have obvious short- and will exhibit little conflict. It becomes
comings that limit effective institutional important primarily when the rest of the
action by those shareholders most likely to internal control system is failing, and this
pursue an issue. should be a relatively rare event. The
As equity holdings become concentrated challenge is to create a system that will not
in institutional hands, it is easier to resolve fall into complacency and inactivity during
some of the free-rider problems that limit periods of prosperity and high-quality
the ability of thousands of individual share- management, and therefore be unable to rise
holders to engage in effective collective early to the challenge of correcting a failing
action. In principle such institutions can management system. This is a difficult task
therefore begin to exercise corporate because there are strong tendencies for
control rights more effectively. Legal and boards to evolve a culture and social norms
regulatory restrictions, however, have pre- that reflect optimal behavior under prosper-
vented financial institutions from playing a ity, and these norms make it extremely
major corporate monitoring role. (Roe difficult for the board to respond early to
(1990, 1991), Black (1990), and Pound failure in its top management team.42
(1991) provide an excellent historical
review of these restrictions.) Therefore, if
institutions are to aid in effective gover- Resurrecting active investors
nance, we must continue to dismantle the A major set of problems with internal
rules and regulations that have prevented control systems are associated with the
them and other large investors from curbing of what I call active investors
accomplishing this coordination. (Jensen (1989a, 1989b)). Active investors
are individuals or institutions that simulta-
The CEO as chairman of the board neously hold large debt and/or equity
positions in a company and actively partic-
It is common in U.S. corporations for the ipate in its strategic direction. Active
CEO to also hold the position of chairman investors are important to a well-functioning
of the board. The function of the chairman governance system because they have the
is to run board meetings and oversee the financial interest and independence to view
THE MODERN INDUSTRIAL REVOLUTION 37
firm has equity value of zero. This second 1980) conducts detailed firm and industry studies.
measure is: See also Hirschman’s (1970) work on exit.
2 Measured by multifactor productivity, U.S.
Department of Labor (USDL) (1990, Table 3).
VT VT Vn Bn (Kt Rt at) See Jensen (1991) for a summary. Multifactor
(1 i)n t (2) productivity showed no growth between 1973
and 1980 and grew at the rate of 1.9 per-
Alternatively, to allow for the effects of cent per year between 1950 and 1973.
the reduced investment and R&D on inter- Manufacturing labor productivity grew at an
mediate cash flows my third measure annual rate of 2.3 percent in the period 1950 to
assumes that all intermediate cash flows 1981 and at 3.8 percent in 1981 to 1990
(USDL, 1990, Table 3). Using data recently
are reduced in the benchmark investment
revised by the Bureau of Economic Analysis
strategy by the amount paid out to share- from 1977 to 1990, the growth rate in the
holders in the form of dividends and net earlier period was 2.2 and 3.0 percent in the
share repurchases and that the original 1981 to 1990 period (USDL, 1991, Table 1).
value of the equity is maintained. This Productivity growth in the nonfarm business
measure is likely to yield an upward biased sector fell from 1.9 percent in the 1950 to 1981
estimate of the productivity of R&D and period to 1.1 percent in the 1981 to 1990
capital expenditures. The measure is:58 period (USDL, 1990, Table 2). The reason for
the fall apparently lies in the relatively large
growth in the service sector relative to the
VT VT Vn dt (1 r)n t manufacturing sector and the low measured
S0 Bn (Kt Rt at) productivity growth in services.
(1 i)n t (3) There is considerable controversy over the
adequacy of the measurement of productivity in
the service sector. The USDL has no productiv-
ity measures for services employing nearly
NOTES 70 percent of service workers, including, among
others, health care, real estate, and securities
* Harvard Business School. Presidential brokerage. In addition, many believe that service
Address to the American Finance Association, sector productivity growth measures are downward
January 1993, Anaheim, California. I appreci- biased. Service sector price measurements, for
ate the research assistance of Chris Allen, Brian example, take no account of the improved
Barry, Susan Brumfield, Karin Monsler, and par- productivity and lower prices of discount outlet
ticularly Donna Feinberg, the support of the clubs such as Sam’s Club. The Commerce
Division of Research of the Harvard Business Department measures output of financial
School, and the comments of and discussions services as the value of labor used to produce it.
with George Baker, Carliss Baldwin, Joe Bower, Since labor productivity is defined as the value
Alfred Chandler, Harry and Linda DeAngelo, of total output divided by total labor inputs it is
Ben Esty, Takashi Hikino, Steve Kaplan, Nancy impossible for measured productivity to grow.
Koehn, Claudio Loderer, George Lodge, John Between 1973 and 1987 total equity shares
Long, Kevin Murphy, Malcolm Salter, René traded daily grew from 5.7 million to 63.8
Stulz, Richard Tedlow, and especially Richard million, while employment only doubled—implying
Hackman, Richard Hall, and Karen Wruck on considerably more productivity growth than
many of these ideas. that reflected in the statistics. Other factors,
1 In a rare finance study of exit, DeAngelo and however, contribute to potential overestimates
DeAngelo (1991) analyze the retrenchment of of productivity growth in the manufacturing
the U.S. steel industry in the 1980s. Ghemawat sector. See Malabre and Clark (1992) and
and Nalebuff (1985) have an interesting paper Richman (1993).
entitled “Exit,” and Anderson (1986) provides 3 Nominal and real hourly compensation,
a detailed comparison of U.S. and Japanese Economic Report of the President, Table B42
retrenchment in the 1970s and early 1980s and (1993).
their respective political and regulatory policies 4 USDL (1991).
toward the issues. Bower (1984, 1986) analyzes 5 USDL (1990). Trends in U.S. producivity
the private and political responses to decline have been controversial issues in academic and
in the petrochemical industry. Harrigan (1988, policy circles in the last decade. One reason,
THE MODERN INDUSTRIAL REVOLUTION 43
I believe, is that it takes time for these 15 Zellner (1992) discusses the difficulties
complicated changes to show up in the traditional retailers have in meeting Wal-Mart’s
aggregate statistics. In their recent book prices.
Baumol, Blackman, and Wolff (1989, pp. ix-x) 16 Vietor, Forthcoming.
changed their formerly pessimistic position. In 17 Source: COMPUSTAT.
their words: “This book is perhaps most easily 18 “In 1980 IBM’s top-of-the-line computers
summed up as a compendium of evidence provided 4.5 MIPS (millions of instructions per
demonstrating the error of our previous second) for $4.5 million. By 1990, the cost of a
ways. . . . The main change that was forced MIP on a personal computer had dropped to
upon our views by careful examination of the $1,000 . . .” (Keene (1991)), p. 110). By 1993
long-run data was abandonment of our earlier the price had dropped to under $100. The
gloomy assessment of American productivity technological progress in personal computers
performance. It has been replaced by the has itself been stunning. Intel’s Pentium (586)
guarded optimism that pervades this book. This chip, introduced in 1993, has a capacity of 100
does not mean that we believe retention of MIPS—100 times the capacity of its 286 chip
American leadership will be automatic or easy. introduced in 1982 (Brandt (1993)). In
Yet the statistical evidence did drive us to con- addition, the progress of storage, printing, and
clude that the many writers who have suggested other related technology has also been rapid
that the demise of America’s traditional posi- (Christensen (1993)).
tion has already occurred or was close at hand 19 The Journal of Financial Economics which
were, like the author of Mark Twain’s obituary, I have been editing with several others since
a bit premature. . . . It should, incidentially, be 1973 is an example. The JFE is now edited by
acknowledged that a number of distinguished seven faculty members with offices at three
economists have also been driven to a similar universities in different states and the main
evaluation . . . ” editorial administrative office is located in yet
6 As measured by the Wilshire 5,000 index of another state. North Holland, the publisher, is
all publicly held equities. located in Amsterdam, the printing is done in
7 Bureau of the Census (1991). India, and mailing and billing is executed in
8 Business Week Annual R&D Scoreboard. Switzerland. This “networked organization”
9 Annual premiums reported by Mergerstat would have been extremely inefficient two
Review (1991, fig. 5) weighted by value of decades ago without fax machines, high-speed
transaction in the year for this estimate. modems, electronic mail, and overnight delivery
10 I assume that all transactions without services.
publicly disclosed prices have a value equal to 20 Wruck and Jensen (1993) provide an
20 percent of the value of the average publicly analysis of the critical organizational innovations
disclosed transaction in the same year, and that that total quality management is bringing to the
they have average premiums equal to those for technology of management.
publicly disclosed transactions. 21 A collapse I predicted in Jensen (1989a).
11 In some cases buyers overpay, perhaps 22 See Neff, Holyoke, Gross, and Miller
because of mistakes or because of agency (1993).
problems with their own shareholders. Such 23 Steel industry employment is now down to
overpayment represents only a wealth transfer 160,000 from its peak of 600,000 in 1953
from the buying firm’s claimants to those of the (Fader (1993)).
selling firm and not an efficiency gain. 24 I am indebted to Steven Cheung for
12 Healy, Palepu, and Ruback (1992) esti- discussions on these issues.
mate the total gains to buying- and selling-firm 25 Although migration will play a role it will
shareholders in the 50 largest mergers in the be relatively small compared to the export of
period 1979 to 1984 at 9.1 percent. They also the labor in products and services. Swissair’s
find a strong positive cross-sectional relation 1987 transfer of part of its reservation system
between the value change and the cash flow to Bombay and its 1991 announcement of plans
changes resulting from the merger. to transfer 150 accounting jobs to the same city
13 See Kaplan (1989), Jensen, Kaplan, and are small examples (Economist Intelligence
Stiglin (1989), Pontiff, Shleifer, and Weisbach Unit (1991)).
(1990), Asquith and Wizman (1990), and 26 Thailand and China have played a role in
Rosett (1990). the world markets in the last decade, but since
14 Its high of $139.50 occurred on 2/19/91 it has been such a small part of their potential I
and it closed at $50.38 at the end of 1992. have left them in the potential entrant category.
44 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
27 In a recent article focusing on the (1990), Tedlow (1991), Tiemann (1990),
prospects for textile manufacturer investment in Wruck and Stephens (1992a, 1992b), Wruck
Central European countries, van Delden (1993, (1990, 1991, 1992), Wruck and Palepu
p. 43) reports: “When major French group (1992).
Rhone Poulenc’s fibres division started a 31 In May 1985, Uniroyal approved an LBO
discussion for a formal joint venture in 1991, proposal to block hostile advances by Carl
they discovered an example of astonishing Icahn. About the same time, BF Goodrich began
competitiveness. Workers—whose qualifications diversifying out of the tire business. In January
matched those normal in the West—cost only 1986, Goodrich and Uniroyal independently
8% of their West European counterparts, and spun off their tire divisions and together, in a
yet achieved productivity rates of between 60% 50–50 joint venture, formed the Uniroyal-
and 75% compared to EC level. Moreover, Goodrich Tire Company. By December 1987,
energy costs of the integrated power station are Goodrich had sold its interest in the venture to
50% below West German costs, and all of this Clayton and Dubilier; Uniroyal followed soon
is complemented by extremely competitive raw after. Similarly, General tire moved away from
material prices.” tires: the company, renamed GenCorp in 1984,
The textile industry illustrates the problems sold its tire division to Continental in 1987.
with chronic excess capacity brought on by a Other takeovers in the industry during this
situation where the worldwide demand for period include the sale of Firestone to
textiles grows fairly constantly, but growth in the Bridgestone and Pirelli’s purchase of the
productivity of textile machinery through tech- Armstrong Tire Company. By 1991, Goodyear
nological improvements is greater. Moreover, was the only remaining major American tire
additional capacity is being created because manufacturer. Yet it too faced challenges in the
new entrants to the global textile market must control market: in 1986, following three years
upgrade outdated (and less productive) weaving of unprofitable diversifying investments,
machinery with new technology to meet mini- Goodyear initiated a major leveraged stock
mum global quality standards. This means repurchase and restructuring to defend itself
excess capacity is likely to be a continuing from a hostile takeover from Sir James
problem in the industry and that adjustment will Goldsmith. Uniroyal-Goodrich was purchased by
have to occur through exit of capacity in Michelin in 1990. See Tedlow (1991).
high-cost Western textile mills. 32 In 1992 dollars, calculated from
28 Total quality management programs Mergerstat Review, 1991, p. 100f.
strongly encourage managers to benchmark 33 In 1992 dollars, Mergerstat Review,
their firm’s operations against the most 1991, figs. 29 and 38.
successful worldwide competitors, and good 34 Source: GE annual reports.
cost systems and competitive benchmarking are 35 Because of the sharp decline in Japanese
becoming more common in well-managed firms. stock prices, the Japanese firms ranked in the
29 Shleifer and Summers (1988) seem to top 35 firms would have performed less well if
take the position that all implicit contracts the period since 1990 had been included.
should be enforced rigidly and never be 36 I am in the process of creating a database
breached. that avoids the survivorship bias. Hall (1993a,
30 A partial list of references is: Dann and 1993b) in a large sample free of survivorship
DeAngelo (1988), Mann and Sicherman bias finds lower market valuation of R&D in the
(1991), Baker and Wruck (1989), Berger and 1980s and hypothesizes that this is due to a
Ofek (1993), Bhide (1993), Brickley, Jarrell, higher depreciation rate for R&D capital. The
and Netter (1988), Denis (1992), Donaldson Stern Stewart Performance 1000 (1992) ranks
(1990), DeAngelo and DeAngelo (1991), companies by a measure of the economic value
DeAngelo, DeAngelo, and Rice (1984), Esty added by management decisions that is an
(1992, 1993), Grundfest (1990), Holderness alternative to performance measures 1–3 sum-
and Sheehan (1991), Jensen (1986a, 1986b, marized in Table 1.4 GM also ranks at the
1988, 1989a, 1989b, 1991), Kaplan (1989a, bottom of this list.
1989b, 1992), Lang, Poulsen, and Stulz 37 Changes in market expectations about the
(1992), Lang, Stulz, and Walkling (1991), prospects for a firm (and therefore changes in
Lewellen, Loderer, and Martin (1987), its market value) obviously can affect the
Lichtenberg (1992), Lichtenberg and Siegel interpretation of the performance measures.
(1990), Ofek (1993), Palepu (1990), Pound Other than using a long period of time there is
(1988, 1991, 1992), Roe (1990, 1991), Smith no simple way to handle this problem. The large
THE MODERN INDUSTRIAL REVOLUTION 45
increase in stock prices in the 1980s would 1992b), Wruck and Stephens (1992a, 1992b),
indicate that expectations were generally being Jensen and Barry (1992), Jensen, Burkhardt,
revised upward. and Barry (1992), Jensen, Dial, and Barry
38 CEO turnover approximately doubles from (1992), Lang and Stultz (1992), Denis (1992).
3 to 6 percent after two years of poor perform- 46 See Jensen (1986), and the references in
ance (stock returns less than 50 percent below the previous footnote.
equivalent-risk market returns, Weisbach 47 Assets do change somewhat after an LBO
(1988)), or increases from 8.3 to 13.9 percent because such firms often engage in asset sales
from the highest to the lowest performing decile after the transaction to pay down debt and to
of firms (Warner, Watts, and Wruck (1988)). get rid of assets that are peripheral to the
See also DeAngelo (1888) and DeAngelo and organization’s core focus.
DeAngelo (1989). 48 See Palepu (1990) for a review of
39 See also Jensen and Murphy (1990a, research on LBOs, their governance changes,
1990b) for similar estimates based on earlier and their productivity effects. Kaplan and
data. Stein (1993) show similar results in more
40 In their excellent analysis of boards, recent data.
Lipton and Lorsch (1992) also criticize the 49 See references in footnote 45 above. In a
functioning of traditionally configured boards, counterexample, Healy and Palepu argue in
recommend limiting membership to seven or their study of CUC that the value increase fol-
eight people, and encourage equity ownership by lowing its recapitalization occurred not because
board members. Research supports the proposi- of incentive effects of the deal, but because of
tion that as groups increase in size they become the information the recapitalization provided to
less effective because the coordination and the capital markets about the nature of the
process problems overwhelm the advantages company’s business and profitability.
gained from having more people to draw on (see 50 Jensen (1989a, 1989b) and Sahlman
Steiner (1972) and Hackman (1990)). (1990) analyze the LBO association and
41 Lipton and Lorsch (1992) stop short of venture capital funds respectively.
recommending appointment of an independent 51 Counterexamples are Bower (1970),
chairman, recommending instead the appoint- Baldwin and Clark (1992), Baldwin (1982,
ment of a “lead director” whose functions 1988, 1991), Baldwin and Trigeorgis (1992),
would be to coordinate board activities. and Shleifer and Vishny (1989).
42 Gersick and Hackman (1990) and 52 See Harris and Raviv (1988, 1991), and
Hackman (1993) study a similar problem: the Stulz (1990).
issues associated with habitual behavior 53 Jensen (1989a, 1989b), Kester (1991),
routines in groups to understand how to create Sahlman (1990), Pound (1988, 1991, 1992).
more productive environments. They apply the 54 Jensen (1983), Jensen and Meckling
analysis to airline cockpit crews. (1992).
43 See Roe (1990, 1991), Black (1990), and 55 Examples of this work include Gilson,
Pound (1991). Lang, and John (1990), Wruck (1990, 1991),
44 See Porter (1992a, 1992b, 1992c). Hall Lang and Stulz (1992), Lang, and Poulsen, and
and Hall provide excellent empirical tests of the Stulz (1992) on bankruptcy and financial
myopic capital market hypothesis on which distress, Warner, Watts, and Wruck (1989),
much of debate on the functioning of U.S. Weisbach (1988), Jensen and Murphy (1990a,
capital markets rests. 1990b) and Gibbons and Murphy (1990) on
45 See Kaplan (1989a, 1989b, 1989c, executive turnover, compensation, and organiza-
1992), Smith (1990), Wruck (1990), tional performance, Esty (1992, 1993) on the
Lichtenberg (1992), Lichtenberg and Siegel effects of organizational form on thrift losses,
(1990), Healy, Palepu, and Ruback (1992), and Gilson and Kraakman (1991) on governance,
Ofek (1993). Brickley and Dark (1987) on franchising,
There have now been a number of detailed Boycko, Shleifer, and Vishny (1993), Kaplan
clinical and case studies of these transactions (1989a, 1989b, 1989c, 1992), Smith (1990),
that document the effects of the changes on Kaplan and Stein (1993), Palepu (1990), and
incentives and organizational effectiveness as Sahlman (1990) on leverage buyouts and
well as the risks of bankruptcy from overlever- venture capital organizations.
aging. See Baker and Wruck (1989), Wruck 56 The 1989 Financial Institutions Reform,
(1991, 1992a), Holderness and Sheehan Recovery, and Enforcement Act increased
(1988, 1991), Wruck and Keating (1992a, federal authority and sanctions by shifting
46 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
regulation of the S&L industry from the FDIC Baker, George, and Karen Wruck, 1989,
to the Treasury, and insurance of the industry Organizational changes and value creation in
from FSLIC to the Federal Home Loan Bank leveraged buyouts: The case of O. M. Scott
Board.The act banned thrift investment in high- and Sons Company, Journal of Financial
yield bonds, raised capital ratios and insurance Economics 25, 163–190.
premiums. The 1990 Comprehensive Thrift and
Baldwin, Carliss Y., 1982, Optimal sequential
Bank Fraud Prosecution and Tax Payer
Recovery Act increased criminal penalties for investment when capital is not readily
financial institution-related crimes. The 1991 reversible, Journal of Finance 37, 763–782.
FDIC Improvement Act tightened examination ——, 1988, Time inconsistency in capital
and auditing standards, recapitalized the Bank budgeting, Working paper, Harvard Business
Insurance Fund and limited foreign bank pow- School.
ers. The Truth in Banking Act of 1992 required ——, 1991, How capital budgeting deters
stricter disclosure of interest rates and fees on innovation—and what companies can do
deposit accounts and tightened advertising about it, Research-Technology Management,
guidelines. The National Association of November-December, 39–45.
Insurance Commissioners (NAIC) substantially
—— and Kim B. Clark, 1992. Capabilities and
restricted the ability of insurance companies to
invest in high-yield debt in 1990 to 1991. capital investment: New perspectives on cap-
57 See Jensen (1989b, 1991), Roe (1990, ital budgeting, Journal of Applied Corporate
1991), Grundfest (1990), Bhide (1993), Black Finance 5, 67–82.
(1990), Pound (1988, 1991, 1992), and —— and Lenos Trigeorgis, 1992, Toward
DeAngleo, DeAngelo, and Gilson (1993). remedying the underinvestment problem:
58 Most conservatively, we could assume that Competitiveness, real options, capabilities, and
the cutback in R&D and capital expenditures TQM,Working paper, Harvard Business School.
under the alternative strategy results in a reduc- Baumol, William, Sue Anne Beattey Blackman,
tion in intermediate cash flows by the amount of and Edward Wolff, 1989, Productivity and
the net cash paid to shareholders in the form of
American Leadership (MIT Press, Boston).
dividends and share repurchases and a final
equity value of zero. Berger, Philip, and Eli Ofek, 1993, Leverage and
value: The role of agency costs and taxes,
Unpublished manuscript,The Wharton School.
VT VT Vn dt (1 r)n t Bn
Bhide, Amar, 1993, The hidden costs of stock
(Kt Rt)(1 i)n t (4) market liquidity, Journal of Financial
Economics 34, 31–55.
I expect this measure provides an unreasonably Black, Bernard S., 1990, Shareholder passivity
high estimate of the productivity of R&D and
reexamined, Michigan Law Review 89,
investment expenditures and therefore do not
report it. 520–608.
Bower, Joseph L., 1970, Managing the Resource
Allocation Process: A Study of Corporate
Planning and Investment (Division of
REFERENCES Research, Graduate School of Business
Administration, Harvard University, Boston).
Anderson, Douglas, 1986, Managing retreat: ——, 1984, Restructuring petrochemicals: A
Disinvestment policy, in Thomas K. comparative study of business and govern-
McCraw, ed.: America versus Japan ment to deal with a declining sector of the
(Harvard Business School Press, Boston), economy, in Bruce R. Scott and George C.
pp. 337–372. Lodge, eds.: U.S. Competitiveness in the
Argyris, Chris, 1990, Overcoming Organizational World Economy (Harvard Business School
Defenses (Allyn and Bacon, Needham, Press, Boston).
Mass.). ——, 1986, When Markets Quake (Harvard
Asquith, Paul, and Thierry A. Wizman, 1990, Business School Press, Boston).
Event risk, convenants, and bondholder Boycko, Maxim, Andrei Shleifer, and Robert W.
returns in leveraged buyouts, Journal of Vishny, 1993, Voucher privatization,
Financial Economics 27, 195–213. Unpublished manuscript, Harvard University.
THE MODERN INDUSTRIAL REVOLUTION 47
Brandt, Richard, 1993, Tiny transistors and —— and Edward Rice, 1984, Going private:
gold pizza, Business Week, March 29. Minority freezeouts and stockholder wealth,
Brickley, James A., and Frederick H. Dark, Journal of Law and Economics 27, 367–401.
1987, The choice of organizational form: The —— and Stuart C. Gilson, 1993,The collapse of
case of franchising, Journal of Financial First Executive Corporation: Junk bonds,
Economics 18, 401–420. adverse publicity, and the “run on the bank”
——, Gregg A. Jarrell, and Jeffrey M. Netter, phenomenon, Unpublished manuscript,
1988, The market for corporate control: The University of Southern California.
empirical evidence since 1980, Journal of DeAngelo, Linda, 1988, Managerial competition,
Economic Perspectives 2, 49–68. information costs, and corporate governance:
——, Ronald C. Lease, and Clifford W. Smith, The use of accounting performance measures
Jr., 1988, Ownership structure and voting on in proxy contests, Journal of Accounting and
antitakeover amendments, Journal of Economics 10, January, 3–36.
Financial Economics 20, 267–291. Denis, David J., 1992, Organizational form and
Burnham, James D., 1993, Changes and the consequences of highly leveraged transac-
Challenges: The Transformation of the U.S. tions: Kroger’s recapitalization and
Steel Industry, Policy Study No. 115 (Center Safeway’s LBO, Unpublished manuscript,
for the Study of American Business, Virginia Polytechnic Institute.
Washington University, St. Louis, Mo.). Dobrzynski, Judith, 1993, Relationship investing,
Chandler, Alfred D., Jr., 1977, The Visible Business Week, March 15.
Hand: The Managerial Revolution in Donaldson, Gordon, 1990, Voluntary restructur-
American Business (Harvard University ing: The case of General Mills, Journal of
Press, Cambridge, Mass.). Financial Economics 27, 117–141.
——, 1990, Scale and Scope, The Dynamics of Economic Report of the President, 1993 (U.S.
Industrial Capitalism (The Belknapp Press of Government Printing Office,Washington, D.C.).
Harvard University Press, Cambridge, Economist Intelligence Unit, 1991, Switzerland
Mass.). Country Report, No. 3.
——, 1992, The emergence of managerial capi- Economist, 1990, Out of the ivory tower,
talism, Harvard Business School Case No. February 3.
9–384–081, revised by Thomas J. McCraw. Esty, Benjamin C., 1992, Organizational form,
Christensen, Clayton, 1993, The rigid disk drive leverage and incentives: A study of risk-
industry, 1956–1990: A history of commer- taking in the S&L industry, Unpublished
cial and technological turbulence, Working manuscript, Harvard Business School.
paper, Harvard Business School. ——, 1993, A tale of two thrifts, Unpublished
Comment, Robert, and G.William Schwert, 1993, manuscript, Harvard Business School.
Poison or placebo? Evidence on the deterrent Fader, Barnaby J., 1993, Struggle to survive in
and wealth effects of modern antitakeover the town that steel forgot, New York Times,
measures. Unpublished manuscript, Simon April 27.
School of Business, University of Rochester. Federal Communications Commission, 1991,
Dann, Larry Y., and Harry DeAngelo, 1988, Competition in the Interstate Interexchange
Corporate financial policy and corporate con- Marketplace, FCC 91–251, Sept. 16.
trol: A study of defensive adjustments in asset Gersick, Connie J., and J. Richard Hackman,
and ownership structure, Journal of Financial 1990, Habitual routines in task-performing
Economics 20, 87–127. teams, Organizational Behavior and Human
DeAngelo, Harry, and Linda DeAngelo, 1989, Decision Processes 47, 65–97.
Proxy contests and the governance of publicly Ghemawat, Pankaj, and Barry Nalebuff, 1985,
held corporations, Journal of Financial Exit, Rand Journal of Economics 16, 184–194.
Economics 23, 29–59. Gibbons, Robert, and Kevin J. Murphy, 1990,
——, 1991, Union negotiations and corporate Relative performance evaluation for chief
policy: A study of labor concessions in the executive officers, Industrial and Labor
domestic steel industry during the 1980s, Relations Review 43, 30S–51S (Special
Journal of Financial Economics 30, 3–43. issue, Cornell University).
48 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Gilson, Ronald, and Reinier Kraakman, 1991, Organizations and States (Harvard
Reinventing the outside director: An agenda University Press, Cambridge, Mass.).
for institutional investors, Stanford Law Holderness, Clifford G., and Dennis P. Sheehan,
Review 43, 863–900. 1988,The role of majority shareholders in pub-
Gilson, Stuart C., Kose John, and Larry H. P. licly held corporations: An exploratory analysis,
Lang, 1990, Trouble debt restructurings: An Journal of Financial Economics 20, 317–346.
empirical study of private reorganization of ——, 1991, Monitoring an owner: The case of
firms in default, Journal of Financial Turner Broadcasting, Journal of Financial
Economics 27, 315–353. Economics 30, 325–346.
Grossman, Sanford J., and Oliver D. Hart, 1988, Jensen, Michael C., 1983, Organization theory
One share/one vote and the market for corpo- and methodology, Accounting Review 58,
rate control, Journal of Financial Economics 319–339.
20, 175–202. ——, 1986a,The agency costs of free cash flow:
Grundfest, Joseph, 1990, Subordination of Corporate finance and takeovers, American
American capital, Journal of Financial Economic Review 76, 323–329.
Economics 27, 89–117. ——, 1986b, The takeover controversy: Analysis
Hackman, J. Richard, ed., 1990, Groups That and evidence, Midland Corporate Finance
Work (Jossey-Bass, San Francisco). Journal 4(2), 6–32.
——, 1993, Teams, leaders, and organizations: ——, 1988, Takeovers: Their causes and conse-
New directions for crew-oriented flight quences, Journal of Economic Perspectives
training, in E. L. Weiner, B. G. Kanki, and 2, 21–48.
R. L. Helmreich, eds.: Cockpit Resource ——, 1989a, Eclipse of the public corporation,
Management (Academic Press, Orlando, Fla.). Harvard Business Review 67(5), 61–74.
Hall, Bronwyn H., 1993a, The stock market’s ——, 1989b, Active investors, LBOs, and the pri-
valuation of R&D investment during the vatization of bankruptcy, Journal of Applied
1980’s, American Economic Review, 83, Corporate Finance 2, 35–44.
259–264. ——, 1991, Corporate control and the politics of
——, 1993b, New evidence on the impacts of finance, Journal of Applied Corporate
research and development, Unpublished Finance 4, 13–33.
manuscript, Stanford University. —— and Brian Barry, 1992, Gordon Cain and
—— and Robert E. Hall, 1993, The value and the Sterling Group (A) and (B), Harvard
performance of U.S. corporations, Business School Case Nos. 9–942–021 and
Unpublished manuscript, Stanford University. 9–942–022.
Harrigan, Kathryn R., 1980, Strategies for Jensen, Michael C.,Willy Burkhardt, and Brian K.
Declining Businesses (Lexington Books, Barry, 1992, Wisconsin Central Ltd. Railroad
Lexington, Mass.). and Berkshire Partners (A): Leverage buyouts
——, 1988, Managing Maturing Businesses: and financial distress, Harvard Business School
Restructuring Declining Industries and Case No. 9–190–062.
Revitalizing Troubled Operations (Lexington Jensen, Michael C., Jay Dial, and Brian K.
Books, Lexington, Mass.). Barry, 1992, Wisconsin Central Ltd. Railroad
Harris, Milton, and Artur Raviv, 1988, and Berkshire Partners (B): LBO associa-
Corporate governance: Voting rights and tions and corporate governance, Harvard
majority rules, Journal of Financial Business School Case No. 9–190–070.
Economics 20, 203–235. Jensen, Michael C., Steven Kaplan, and Laura
——, 1991, The theory of capital structure, Stiglin, 1989, Effects of LBOs on tax rev-
Journal of Finance 46, 297–355. enues of the U.S. Treasury, Tax Notes 42,
Healy, Paul M., Krishna G. Palepu, and Richard 727–733.
S. Ruback, 1992, Does corporate perform- Jensen, Michael C., and William H. Meckling,
ance improve after mergers?, Journal of 1992, Knowledge, control and organizational
Financial Economics 31, 135–175. structure, in Lars Werin and Hans Wijkander,
Hirschman, Albert, 1970, Exit, Voice and eds.: Current Economics (Blackwell, Oxford),
Loyalty: Responses to Decline in Firms, pp. 251–274.
THE MODERN INDUSTRIAL REVOLUTION 49
Jensen, Michael C., and Kevin J. Murphy, —— and Ralph A. Walkling, 1991, A test of the
1990a, Performance pay and top-management free cash flow hypothesis: The case of bidder
incentives, Journal of Political Economy 98, returns, Journal of Financial Economics 29,
225–264. 315–335.
——, 1990b, CEO incentives—It’s not how Lewellen, Wilbur, Claudio Loderer, and Kenneth
much you pay, but how, Harvard Business Martin, 1987, Executive compensation and
Review, 68(3), 138–153. executive incentive problems: An empirical
Jensen, Michael C., and Jerold B. Warner, 1988, analysis, Journal of Accounting and
The distribution of power among corporate Economics 9, 287–310.
managers, shareholders, and directors. Lichtenberg, Frank R., 1992, Corporate
Journal of Financial Economics 20, 3–24. Takeovers and Productivity (M.I.T. Press:
Kaplan, Steven N., 1989a, The effects of man- Cambridge, Mass.).
agement buyouts on operating performance —— and Donald Siegel, 1990, The effects of
and value, Journal of Financial Economics leveraged buyouts on productivity and related
24, 581–618. aspects of firm behavior, Journal of Financial
——, 1989b, Campeau’s acquisition of Economics 27, 165–194.
Federated: Value added or destroyed, Journal Lipton, Martin, 1989, Corporate governance:
of Financial Economics 25, 191–212. Major issues for the 1990’s, Address to the
——, 1989c, Management buyouts: Evidence on Third Annual Corporate Finance Forum at
taxes as a source of value, Journal of Finance the J. Ira Harris Center for the Study of
44, 611–632. Corporate Finance, University of Michigan
——, 1992, Campeau’s acquisition of School of Business, April 6.
Federated: A post-petition post-mortem, —— and Jay Lorsch, 1992, A modest proposal
working paper, University of Chicago. for improved corporate governance, Business
—— and Jeremy Stein, 1990, How risky is the Lawyer 48, 59–77.
debt in highly leveraged transactions?, Loomis, Carol J., 1993, Dinosaurs?, Fortune,
Journal of Financial Economics 27, June 15, 85–000.
215–245. McCraw, Thomas K., 1981, Rethinking the trust
——, 1993, The evolution of buyout pricing and question, in T. McCraw, ed.: Regulation in
financial structure in the 1980s, Quarterly Perspective (Division of Research, Harvard
Journal of Economics 108, 313–358. University, Graduate School of Business, dis-
Keene, Peter G., 1991, Every Manager’s Guide tributed by Harvard University Press,
to Information Technology (Harvard Boston), pp. 1–55.
University Press, Boston). ——, 1992, Antitrust: The perceptions and real-
Kester, Carl, 1991, Japanese Takeovers: The ity in coping with big business, Harvard
Global Contest for Corporate Control Business School Case No. 9–391–292.
(Harvard Business School Press, Cambridge, McMurray, Donald L., 1929, Coxey’s Army: A
Mass.). Study of the Industrial Army Movement of
Lamoreaux, Naomi, R., 1985, The Great 1894 (Little, Brown, Boston).
Merger Movement in American Business, Magnet, Myron, 1992, Directors, wake up!
1895–1904 (Cambridge University Press, Fortune, June 15, 85–92.
Cambridge). Malabre, Alfred L., Jr., and Lindley H. Clark,
Lang, Larry H. P., Annette Poulsen, and Rene Jr., 1992, Dubious figures: Productivity
M. Stulz, 1992, Asset sales, leverage, and the statistics for the service sector may
agency costs of managerial discretion, understate gains, Wall Street Journal,
Unpublished manuscript, Ohio State August 12.
University. Mann, Steven V., and Neil W. Sicherman, 1991,
Lang, Larry H. P., and Rene M. Stulz, 1992, The agency costs of free cash flow:
Contagion and competitive intra-industry Acquisition activity, and equity issues,
effects of bankruptcy announcements: An Journal of Business 64, 213–227.
empirical analysis, Journal of Financial Mergerstat Review, 1991 (Merrill Lynch,
Economics 32, 45–60. Schaumburg, Illinois).
50 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Munro, J. Richard, 1989, Takeovers: The myths Price, Walter W., 1933, We Have Recovered
behind the mystique, Vital Speeches, May 15. Before! (Harper & Brothers, New York).
Murphy, Kevin, 1992, Executive Compensation Reynolds, A.William, 1988, Chairman and Chief
in Corporate America, 1992 (United Executive Officer, GenCorp Inc., before
Shareholders Association, Washington, D.C.). the Subcommittee on Oversight and
—— and Jay Dial, 1992, Compensation and Investigations, U.S. House Committee on
strategy at General Dynamics (A) and (B), Energy and Commerce, February 8.
Harvard Business School Case Nos. Richman, Louis S., 1993, Why the economic
N9–493–032 and N9–493–033. data mislead us, Fortune, March 8.
Neff, Robert, Larry Holyoke, Neil Gross, and Roe, Mark J., 1990, Political and legal
Karen Lowry Miller, 1993, Fixing Japan, restraints on ownership and control of public
Business Week, March 29. companies, Journal of Financial Economics
Ofek, Eli, 1993, Capital structure and firm 27, 7–42.
response to poor performance: An empirical ——, 1991, a political theory of American
analysis, Journal of Financial Economics corporate finance, Columbia Law Review 91,
34, 3–30. 10–67.
Palepu, Krishna G., 1990, Consequences of Rosett, Joshua G., 1990, Do union wealth
leveraged buyouts, Journal of Financial concessions explain takeover premiums? The
Economics 27, 247–262. evidence on contract wages, Journal of
Polinsky, A. Mitchell, 1989, An Introduction to Financial Economics 27, 263–282.
Law and Economics (Little, Brown & Co., Sahlman, William A., 1990, The structure and
Boston). governance of venture-capital organizations.
Pontiff, Jeffrey, Andrei Shleifer, and Michael S. Journal of Financial Economics 27, 473–521.
Weisbach, 1990, Reversions of excess pension Sahlman, William A., and Howard H.
assets after takeovers, Rand Journal of Stevenson, 1985, Capital market myopia,
Economics 21, 600–613. Journal of Business Venturing 1, 7–30.
Porter, Michael E., 1990, Competitive Advantage Schumpeter, Joseph A., 1976, Capitalism
of Nations (Free Press, New York). Socialism, and Democracy (Harper & Row,
——, 1992a, Capital choices: Changing the way New York).
America invests in industry, Journal of Shleifer, Andrei, and Larry Summers, 1988,
Applied Corporate Finance 5, 4–16. Breach of trust in hostile takeovers, in
——, 1992b. Capital choices: Changing the way A. Auerbach, ed.: Corporate Takeovers:
America invests in industry, A Research Causes and Consequences (University of
Report presented to The Council on Chicago Press, Chicago), pp. 33–56.
Competitiveness and co-sponsored by Shleifer, Andrei, and Robert W. Vishny, 1989,
Harvard Business School. Management entrenchment: The case of
——, 1992c, Capital disadvantage: America’s manager-specific investments, Journal of
failing capital investment system, Harvard Financial Economics 25, 123–139.
Business Review 70(5), 65–82. Smith, Abbie J., 1990, Corporate ownership
Pound, John, 1988, Proxy contests and the structure and performance: The case of
efficiency of shareholder oversight, Journal of management buyouts, Journal of Financial
Financial Economics 20, 237–265. Economics 27, 143–164.
——, 1991, Proxy voting and the SEC: Investor Steiner, I. D., 1972, Group Process and
protection versus market efficiency, Journal Productivity (Academic Press, New York).
of Financial Economics 29, 241–285. Stern Stewart Performance 1000, 1992,
——, 1992, Beyond takeovers: Politics comes to Corporate Finance, October/November, 33–45.
corporate control, Harvard Business Review Stewart, G. Bennett, III, 1990, Remaking the
70(2), 83–93. public corporation from within, Harvard
——, 1993, The rise of the political model of Business Review 68(4), 126–137.
corporate governance and corporate control, Stulz, Rene M., 1990, Managerial discretion
Unpublished manuscript, Harvard University, and optimal financing policies, Journal of
Kennedy School of Government. Financial Economics 26, 3–00.
THE MODERN INDUSTRIAL REVOLUTION 51
Swartz, L. Mick, 1992, The impact of third Watson, Bruce, 1993, For a while the Luddites
generation antitakeover laws on security had a smashing success, Smithsonian, April,
value, Unpublished manuscript, University of pp. 140–154.
Manitoba. Weisbach, Michael S., 1988, Outside directors
Tedlow, Richard, 1991, Hitting the skids: Tires and CEO turnovers, Journal of Financial
and time horizons, Unpublished manuscript, Economics 20, 431–460.
Harvard Business School. Wruck, Karen H. 1990, Financial distress,
Tiemann, Jonathan, 1990,The economics of exit reorganization, and organizational efficiency,
and restructuring: The Pabst Brewing Journal of Financial Economics 27,
Company, Unpublished manuscript, Harvard 420–444.
Business School. ——, 1991, What really went wrong at Revco?,
United Nations, 1993, Monthly Bulletin of Journal of Applied Corporate Finance 4,
Statistics, February (United Nations, New 79–92.
York). ——, 1992, Changing the rules of the game:
United Nations, 1992, 1988/89 Statistical Sealed Air Corporation’s leveraged recapital-
yearbook, thirty-seventh issue (United ization, Working paper, Harvard Business
Nations, New York). School.
U.S. Bureau of the Census, 1987, Census of —— and Michael C. Jensen, 1993, Science,
Manufacturing (U.S. Government Printing specific knowledge, and total quality manage-
Office, Washington, D.C.) ment, Unpublished manuscript, Harvard
U.S. Bureau of the Census, Housing and Business School.
Household Economic Statistics Division, Wruck, Karen H., and A. Scott Keating, 1992a,
1991 (U.S. Government Printing Office, Sterling Chemicals, Inc., quality and produc-
Washington, D.C.). tivity improvement program, Harvard
U.S. Department of Labor, Bureau Business School Case No. 9–493–026.
of Labor Statistics, Office of Productivity ——, 1992b, Sterling Chemicals, Inc., Harvard
and Technology, 1990, Multifactor Business School Case No. 9–493–025.
Productivity Measures, Report No. USDL Wruck, Karen H., and Krishna Palepu, 1992,
91–412. Consequences of leveraged shareholder payouts:
——, 1991, International Comparisons of Defensive versus voluntary recapitalizations,
Manufacturing Productivity and Unit Labor Working paper, Harvard Business School.
Cost Trends, Report No. USDL 92–752. Wruck, Karen H. and Steve-Anna Stephens,
van Delden, Hendrik H., 1993, Czech and 1992a, Leveraged buyouts and restructuring:
Slovak Republics: Lands of opportunity, ITS The case of Safeway, Inc., Harvard Business
Textile Leader, Spring, 43–48. School Case No. 192–095.
Vietor, Richard, 1994, Contrived Competition ——, 1992b, Leveraged buyouts and restructur-
(Harvard University Press, Cambridge, ing: The case of Safeway, Inc.: Media
Mass.) Forthcoming. response, Harvard Business School Case No.
Warner, Jerold B., Ross L. Watts, and Karen H. 192–094.
Wruck, 1989, Stock prices and top manage- Zellner, Wendy, 1992, Not everybody loves
ment changes, Journal of Financial Wal-Mart’s low prices, Wall Street Journal,
Economics 20, 461–492. October 12.
Chapter 2
ABSTRACT
This article surveys research on corporate governance, with special attention to the importance
of legal protection of investors and of ownership concentration in corporate governance systems
around the world.
be relatively rare if the incentive component Similarly to Berle and Means, Jensen and
of pay is substantial. In this way, incentive Murphy interpret their findings as evidence
contracts can induce the manager to act in of inefficient compensation arrangements,
investors’ interest without encouraging although in their view these arrangements
blackmail, although such contracts may be are driven by politically motivated restric-
expensive if the personal benefits of control tions on extremely high levels of pay.
are high and there is a lower bound on the Kaplan (1994a,b) shows that the
manager’s compensation in the bad states sensitivity of pay (and dismissal) to per-
of the world. Typically, to make such formance is similar in the United States,
contracts feasible, some measure of per- Germany, and Japan, although average
formance that is highly correlated with the levels of pay are the highest in the United
quality of the manager’s decision must be States. The question is whether there is a
verifiable in court. In some cases, the cred- similar failure to pay for performance in all
ibility of an implicit threat or promise from countries, or, alternatively, the results found
the investors to take action based on an by Jensen and Murphy are not so counter-
observable, but not verifiable, signal may intuitive. In particular, even the sensitivity
also suffice. Incentive contracts can take a of pay to performance that Jensen and
variety of forms, including share ownership, Murphy find would generate enormous
stock options, or a threat of dismissal if swings in executive wealth, which require
income is low (Jensen and Meckling considerable risk tolerance. More sensitivity
(1976), Fama (1980)). The optimal may not be efficient for risk-averse execu-
incentive contract is determined by the tives (Haubrich (1994)).
manager’s risk aversion, the importance of The more serious problem with high pow-
his decisions, and his ability to pay for the ered incentive contracts is that they create
cash flow ownership up front (Ross enormous opportunities for self-dealing for
(1973), Stiglitz (1975), Mirrlees (1976), the managers, especially if these contracts
Holmstrom (1979, 1982)). are negotiated with poorly motivated
Incentive contracts are indeed common boards of directors rather than with large
in practice. A vast empirical literature on investors. Managers may negotiate for
incentive contracts in general and manage- themselves such contracts when they know
ment ownership in particular dates back at that earnings or stock price are likely to
least to Berle and Means (1932), who rise, or even manipulate accounting num-
argue that management ownership in large bers and investment policy to increase their
firms is too small to make managers inter- pay. For example, Yermack (1997) finds
ested in profit maximization. Some of the that managers receive stock option grants
early studies take issue with Berle and shortly before good news announcements
Means by documenting a positive relation- and delay such grants until after bad
ship between pay and performance, and news announcements. His results suggest
thus rejecting the extreme hypothesis of that options are often not so much an
complete separation of ownership and incentive device as a somewhat covert
control (Murphy (1985), Coughlan and mechanism of self-dealing.
Schmidt (1985), Benston (1985)). More Given the self-dealing opportunities in
recently, Jensen and Murphy (1990) look high powered incentive contracts, it is not
at the sensitivity of pay of American execu- surprising that courts and regulators have
tives to performance. In addition to looking looked at them with suspicion. After all,
at salary and bonuses, Jensen and Murphy the business judgment rule that governs the
also examine stock options and the effects attitude of American courts toward agency
on pay of potential dismissal after poor problems keeps the courts out of corporate
performance. Jensen and Murphy arrive at decisions except in the matters of executive
a striking number that executive pay rises pay and self-dealing. These legal and polit-
(and falls) by about $3 per every $1000 ical factors, which appear to be common in
change in the wealth of a firm’s shareholders. other countries as well as in the United
A SURVEY OF CORPORATE GOVERNANCE 59
standards, which in Russia are about one exchange for their funds, only the hope that
tenth of those in the West. Even controlling they will make money in the future.
for this difference, the Russian assets sold at Reputation-building is a very common
a 99 percent discount. Very similar evidence explanation for why people deliver on their
comes from the oil industry, where Russian agreements even if they cannot be forced to
companies were valued at under 5 cents per (see, for example, Kreps (1990)). In the
barrel of proven reserves, compared to financing context, the argument is that
typical $4 to $5 per barrel valuations for managers repay investors because they
Western oil firms. An important element of want to come to the capital market and
this 99 percent discount is surely the reality raise funds in the future, and hence need to
of government expropriation, regulation, and establish a reputation as good risks in
taxation. Poor management is probably also order to convince future investors to give
a part of the story. But equally important them money. This argument has been made
seems to be the ability of managers of initially in the context of sovereign borrow-
Russian firms to divert both profits and ing, where legal enforcement of contracts is
assets to themselves. The Russian evidence virtually nonexistent (Eaton and Gersovitz
suggests that an upper bound on agency (1981), Bulow and Rogoff (1989)).
costs in the regime of minimal protection of However, several recent articles have
investors is 99 percent of value. presented reputation-building models of
private financing. Diamond (1989, 1991)
shows how firms establish reputations as
II. FINANCING WITHOUT good borrowers by repaying their short
GOVERNANCE term loans, and Gomes (1996) shows how
dividend payments create reputations that
The previous section raised the main enable firms to raise equity.
question of corporate governance: why do There surely is much truth to the reputation
investors part with their money, and give it models, although they do have problems. As
to managers, when both the theory and the pointed out by Bulow and Rogoff (1989),
evidence suggests that managers have pure reputational stories run into a back-
enormous discretion about what is done ward recursion problem. Suppose that at
with that money, often to the point of being some point in the future (or in some future
able to expropriate much of it? The question states of the world), the future benefits to
is particularly intriguing in the case of the manager of being able to raise outside
investors because, unlike highly trained funds are lower than the costs of paying
employees and managers, the initial what he promised investors already. In this
investors have no special ability to help the case, he rationally defaults on his repay-
firm once they have parted with their ments. Of course, if investors expect that
money. Their investment is sunk and such a time or state is reached in the
nobody—especially the managers—needs future, they would not finance the firm in
them. Yet despite all these problems, the first place. Under some plausible
outside finance occurs in almost all market circumstances discussed by Bulow and
economies, and on an enormous scale in the Rogoff, the problem unravels and there is
developed ones. How does this happen? no possibility of external finance. While
In this section, we begin to discuss the reputation is surely an important reason
various answers to the puzzle of outside why firms are able to raise money, the
finance by first focusing on two explana- available research suggests that it is prob-
tions that do not rely on governance proper: ably not the whole explanation for external
the idea that firms and managers have financing. For example, in Diamond’s
reputations and the idea that investors are (1989) model of corporate borrowing,
gullible and get taken. Both of these reputation plays a role alongside other
approaches have the common element that protections of creditors that prevent managers
investors do not get any control rights in from removing assets from the firm.
62 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
An alternative theory of how investors suggesting that the high yield bonds that
give their money to companies without were used to finance takeovers in the
receiving control rights in exchange United States in the late 1980s were
appeals to excessive investor optimism. systematically overvalued by investors.
Investors get excited about companies, and Evidence from both the United States and
hence finance them without thinking much other countries also indicates that the
about getting their money back, simply shares of companies issuing equity in initial
counting on short run share appreciation. or secondary offerings are systematically
An extreme version of this story is a Ponzi overvalued (Ritter (1991), Loughran, Ritter,
scheme, in which promoters raise external and Rydqvist (1994), Pagano, Panetta, and
funds sequentially, and use the funds raised Zingales (1995), Teoh, Welch, and Wong
from later investors to pay off initial (1995)).This evidence points to concentration
investors, thereby creating an illusion of of new issues during times when stock prices
high returns. Even without Ponzi schemes, are high, to poor long run performance of
if investors are sufficiently optimistic about initial public offerings, to earnings manipu-
short term capital gains and are prepared lation prior to the issue, and to deterioration
to part with their money without regard for of profitability following the issue. In short,
how the firm will ultimately pay investors excessive investor optimism as an explana-
back, then external finance can be sus- tion of security issues appears to have at
tained without effective governance. least some explanatory power.
Delong, Shleifer, Summers, and Waldmann Still, we do not believe that investors as
(1989, 1990) provide early models of a general rule are prepared to pay good
external finance based on excessive money for securities that are actually
investor optimism. worthless because managers can steal
Pyramid schemes have been an essential everything. As the evidence on agency
element of all major financial markets, theory indicates, managers can expropriate
going back at least to the Louisiana and only limited wealth, and therefore the
the South Sea Bubbles (Kindleberger securities that investors buy do have some
(1978)). Most railroad booms in the world underlying value.To explain why these secu-
were financed by investors who had virtu- rities have value, we need theories that go
ally no protection, only hope. In the United beyond investor overoptimism.
States, such schemes were very common as
recently as the 1920s (Galbraith (1955)),
and still happen occasionally today. They III. LEGAL PROTECTION
also occur in many transition economies, as
Russia’s famous pyramid scheme, MMM, in The principal reason that investors provide
which millions of people subscribed to external financing to firms is that they
shares of a company that used the proceeds receive control rights in exchange. External
to advertise on television while running a financing is a contract between the firm as
Ponzi scheme, vividly illustrates. Nor is it a legal entity and the financiers, which
crazy to assume that enormous volumes of gives the financiers certain rights vis a vis
equity financing in the rapidly growing East the assets of the firm (Hart (1995), part
Asian economies are based in part on II). If firm managers violate the terms of
investor optimism about near-term appre- the contract, then the financiers have the
ciation, and overlook the weakness of right to appeal to the courts to enforce
mechanisms that can force managers to their rights. Much of the difference in
repay investors. corporate governance systems around the
In recent years, more systematic statisti- world stems from the differences in the
cal evidence has pointed to the importance nature of legal obligations that managers
of investor optimism for financing in at have to the financiers, as well as in the
least some markets. Kaplan and Stein differences in how courts interpret and
(1993), for example, present evidence enforce these obligations.
A SURVEY OF CORPORATE GOVERNANCE 63
The most important legal right systematic evidence is mixed. In the United
shareholders have is the right to vote on States, boards, especially those dominated
important corporate matters, such as by outside directors, sometimes remove top
mergers and liquidations, as well as in managers after poor performance
elections of boards of directors, which in (Weisbach (1988)). However, a true per-
turn have certain rights vis a vis the formance disaster is required before boards
management (Manne (1965), Easterbrook actually act (Warner, Watts, and Wruck
and Fischel (1983)). (We discuss voting (1988)). The evidence on Japan and
rights as the essential characteristic of Germany (Kaplan (1994a,b)) similarly
equity in Section VI.) Voting rights, how- indicates that boards are quite passive
ever, turn out to be expensive to exercise except in extreme circumstances. Mace
and to enforce. In many countries, share- (1971) and Jensen (1993) argue very
holders cannot vote by mail and actually strongly that, as a general rule, corporate
have to show up at the shareholder meeting boards in the United States are captured by
to vote—a requirement that virtually guar- the management.
antees nonvoting by small investors. In In many countries, shareholder voting
developed countries, courts can be relied on rights are supplemented by an affirmative
to ensure that voting takes place, but even duty of loyalty of the managers to share-
there managers often interfere in the voting holders. Loosely speaking, managers have a
process, and try to jawbone shareholders duty to act in shareholders’ interest.
into supporting them, conceal information Although the appropriateness of this duty
from their opponents, and so on (Pound is often challenged by those who believe
(1988), Grundfest (1990)). In countries that managers also ought to have a duty of
with weaker legal systems, shareholder loyalty to employees, communities, credi-
voting rights are violated more flagrantly. tors, the state, and so on (see the articles in
Russian managers sometimes threaten Hopt and Teubner, Eds. (1985)), the courts
employee-shareholders with layoffs unless in Organization for Economic Cooperation
these employees vote with the manage- and Development (OECD) countries have
ment, fail to notify shareholders about generally accepted the idea of managers’
annual meetings, try to prevent hostile duty of loyalty to shareholders. There is a
shareholders from voting based on techni- good reason for this. The investments by
calities, and so on. Besides, as Stalin noted, shareholders are largely sunk, and further
“it is important not how people vote, but investment in the firm is generally not
who counts the votes,” and managers count needed from them. This is much less the
shareholders’ votes. Still, even in Russia, case with employees, community members,
courts have protected a large shareholder and even creditors. The employees, for
when a firm’s management erased his example, get paid almost immediately for
name from the register of shareholders. In their efforts, and are generally in a much
sum, both the legal extent and the court better position to hold up the firm by
protection of shareholder voting rights threatening to quit than the shareholders
differ greatly across countries. are. Because their investment is sunk,
Even if shareholders elect the board, shareholders have fewer protections from
directors need not necessarily represent expropriation than the other stakeholders
their interests. The structure of corporate do. To induce them to invest in the first
boards varies greatly even across developed place, they need stronger protections, such
economies, ranging from two-tier supervi- as the duty of loyalty.
sory and management boards in Germany, Perhaps the most commonly accepted
to insider-dominated boards in Japan, to element of the duty of loyalty are the legal
mixed boards in the United States restrictions on managerial self-dealing, such
(Charkham (1994)).The question of board as outright theft from the firm, excessive
effectiveness in any of these countries has compensation, or issues of additional securi-
proved to be controversial. The available ties (such as equity) to the management and
64 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
its relatives. In some cases, the law explicitly Like shareholders, creditors have a
prohibits self-dealing; in other cases, courts variety of legal protections, which also vary
enforce corporate charters that prohibit it across countries. (Again, we say more
(see Easterbrook and Fischel (1991)). about this in the discussion of debt and
Some legal restrictions on managers bankruptcy in Section VI.) These may
constrain their actions, by for example include the right to grab assets that serve
demanding that managers consult the as collateral for the loans, the right to
board of directors before making major liquidate the company when it does not pay
decisions, or giving shareholders appraisal its debts, the right to vote in the decision to
remedies to stop asset sales at low reorganize the company, and the right to
prices. Other restrictions specify that remove managers in reorganization. Legal
minority shareholders be treated as well as protection of creditors is often more effec-
the insiders (Holderness and Sheehan tive than that of the shareholders, since
(1988a)). default is a reasonably straightforward
Although the duty of loyalty is accepted violation of a debt contract that a court
in principle in most OECD countries, the can verify. On the other hand, when the
strictness with which the courts enforce it bankruptcy procedure gives companies the
varies greatly. In the United States, courts right of automatic stay of the creditors,
would interfere in cases of management managers can keep creditors at bay even
theft and asset diversion, and they would after having defaulted. Repossessing assets
surely interfere if managers diluted existing in bankruptcy is often very hard even for
shareholders through an issue of equity to the secured creditors (White (1993)). With
themselves. Courts are less likely to inter- multiple, diverse creditors who have
fere in cases of excessive pay, especially if conflicting interests, the difficulties of col-
it takes the complex form of option con- lecting are even greater, and bankruptcy
tracts, and are very unlikely to second proceedings often take years to complete
guess managers’ business decisions, includ- (Baird and Jackson (1985), Gertner and
ing the decisions that hurt shareholders. Scharfstein (1991), Weiss (1990)). This,
Perhaps most importantly, shareholders in of course, makes debt a less attractive
the United States have the right to sue the financing instrument to begin with (Bolton
corporation, often using class action suits and Scharfstein (1996)). Still, while costly
that get around the free rider problem, if to the creditors, bankruptcy is very tough
they believe that the managers have vio- on the debtor firms as well, since their
lated the duty of loyalty. managers typically get fired, assets liqui-
The United States is generally viewed as dated, and debt kept largely in place (Baird
relatively tough on managers in interpret- (1995)). Creditors’ legal rights are thus
ing the duty of loyalty, although some, enforced in a costly and inefficient way, but
including Bebchuk (1985) and Brudney they are enforced.
and Chirelstein (1978), believe it is not Because bankruptcy procedures are so
tough enough. For example, in France the complicated, creditors often renegotiate
doctrine of corporate opportunities, which outside of formal bankruptcy proceedings
prohibits managers from personally profit- both in the United States (Gilson, John, and
ing from business opportunities that are Lang (1990), Asquith, Gertner, and
offered to the corporation, is not accepted Scharfstein (1994)) and in Europe (OECD
by courts (Tunc (1991)). Outside the (1995)). The situation is worse in developing
United States and Canada, class action countries, where courts are even less reliable
suits are not generally permitted and con- and bankruptcy laws are even less complete.
tingent fees are prohibited (Romano The inefficiency of existing bankruptcy pro-
(1993a)). Outside the OECD, the duty of cedures has prompted some economists
loyalty is a much weaker concept, at least (Bebchuk (1988), Aghion, Hart, and Moore
in part because courts have no capability (1992)) to propose new ones, which try to
or desire to interfere in business. avoid complicated negotiations by first
A SURVEY OF CORPORATE GOVERNANCE 65
converting all the claims of a bankrupt concentrate share holdings. This can mean
company into equity, and then allowing the that one or several investors in the firm
equity holders to decide what to do with the have substantial minority ownership
bankrupt firm. It is possible that in the long stakes, such as 10 or 20 percent. A sub-
run, these proposals will reduce the cost of stantial minority shareholder has the
enforcing creditor rights. incentive to collect information and moni-
In sum, the extent of legal protection of tor the management, thereby avoiding the
investors varies enormously around the traditional free rider problem. He also has
world. In some countries, such as the United enough voting control to put pressure on
States, Japan, and Germany, the law pro- the management in some cases, or perhaps
tects the rights of at least some investors even to oust the management through a
and the courts are relatively willing to proxy fight or a takeover (Shleifer and
enforce these laws. But even in these coun- Vishny (1986b)). In the more extreme
tries, the legal system leaves managers with cases, large shareholders have outright
considerable discretion. In most of the rest control of the firms and their management
of the world, the laws are less protective of with 51 or more percent ownership. Large
investors and courts function less well and shareholders thus address the agency
stop only the clearest violations of investor problem in that they both have a general
rights. As a result, legal protection alone interest in profit maximization, and enough
becomes insufficient to ensure that control over the assets of the firm to have
investors get their money back. their interests respected.
In the United States, large share holdings,
and especially majority ownership, are
IV. LARGE INVESTORS relatively uncommon—probably because of
legal restrictions on high ownership and
If legal protection does not give enough exercise of control by banks, mutual funds,
control rights to small investors to induce insurance companies, and other institutions
them to part with their money, then (Roe (1994)). Even in the United States,
perhaps investors can get more effective however, ownership is not completely
control rights by being large. When control dispersed, and concentrated holdings by
rights are concentrated in the hands of a families and wealthy investors are more
small number of investors with a collec- common than is often believed (Eisenberg
tively large cash flow stake, concerted (1976), Demsetz (1983), Shleifer and
action by investors is much easier than Vishny (1986b)). Holderness and Sheehan
when control rights, such as votes, are split (1988a,b) in fact found several hundred
among many of them. In particular, this cases of over 51 percent shareholders in
concerted action is possible with only public firms in the United States. One other
minimal help from the courts. In effect, country where the rule is broadly dispersed
concentration of ownership leverages up ownership by diversified shareholders is the
legal protection. There are several distinct United Kingdom (Black and Coffee
forms that concentration can take, includ- (1994)).
ing large shareholders, takeovers, and large In the rest of the world, large share hold-
creditors. In this section, we discuss these ings in some form are the norm. In
forms of concentrating ownership, and how Germany, large commercial banks through
they address the agency problem. In the proxy voting arrangements often control
following section, we discuss some costs of over a quarter of the votes in major
having large investors. companies, and also have smaller but
significant cash flow stakes as direct share-
A. Large shareholders holders or creditors (Franks and Mayer
(1994), OECD (1995)). In addition, one
The most direct way to align cash flow and study estimates that about 80 percent of
control rights of outside investors is to the large German companies have an over
66 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
when firms default or violate debt are not well established, bank governance is
covenants (Smith and Warner (1979)) and likely to be less effective (see Barca (1995)
in part because they typically lend short on Italy).
term, so borrowers have to come back at The need for at least some legal protec-
regular, short intervals for more funds. As a tion is shared by all large investors. Large
result of having a whole range of controls, shareholders need courts to enforce their
large creditors combine substantial cash voting rights, takeover artists need court-
flow rights with the ability to interfere in protected mechanisms for buying shares and
the major decisions of the firm. Moreover, changing boards of directors, and creditors
in many countries, banks end up holding need courts to enable them to repossess
equity as well as debt of the firms they collateral. The principal advantage of large
invest in, or alternatively vote the equity of investors (except in takeovers) is that they
other investors (OECD (1995)). As a rely on relatively simple legal interventions,
result, banks and other large creditors are which are suitable for even poorly informed
in many ways similar to the large share- and motivated courts. Large investors put a
holders. Diamond (1984) presents one of lighter burden on the legal system than the
the first models of monitoring by the large small investors might if they tried to
creditors. enforce their rights. For this reason,
Although there has been a great deal of perhaps, large investors are so prevalent in
theoretical discussion of governance by most countries in the world, where courts
large creditors, the empirical evidence of are less equipped to meddle in corporate
their role remains scarce. For Japan, affairs than they are in the United States.
Kaplan and Minton (1994) and Kang and
Shivdasani (1995) document the higher
incidence of management turnover in V. THE COSTS OF LARGE INVESTORS
response to poor performance in companies
that have a principal banking relationship The benefits of large investors are at least
relative to companies that do not. For theoretically clear: they have both the
Germany, Gorton and Schmid (1996) find interest in getting their money back and the
evidence of banks improving company per- power to demand it. But there may be costs
formance (to the extent they hold equity) of large investors as well.The most obvious
more so than other block holders do in of these costs, which is also the usual
1974, although this is not so in 1985. For argument for the benefits of dispersed
the United States, DeLong (1991) points ownership, is that large investors are not
to a significant governance role played diversified, and hence bear excessive risk (see,
by J. P. Morgan partners in the companies e.g., Demsetz and Lehn (1985)). However,
J. P. Morgan invested in in the early 20th the fact that ownership in companies is so
century. More recently, U.S. banks play a concentrated almost everywhere in the
major governance role in bankruptcies, world suggests that lack of diversification
when they change managers and directors is not as great a private cost for large
(Gilson 1990). investors to bear as relinquishing control.
The effectiveness of large creditors, like A more fundamental problem is that the
the effectiveness of large shareholders, large investors represent their own
depends on the legal rights they have. In interests, which need not coincide with the
Germany and Japan, the powers of the interests of other investors in the firm, or
banks vis a vis companies are very signifi- with the interests of employees and man-
cant because banks vote significant blocks agers. In the process of using his control
of shares, sit on boards of directors, play a rights to maximize his own welfare, the
dominant role in lending, and operate in a large investor can therefore redistribute
legal environment favorable to creditors. In wealth—in both efficient and inefficient
other countries, especially where proce- ways—from others. This cost of concen-
dures for turning control over to the banks trated ownership becomes particularly
A SURVEY OF CORPORATE GOVERNANCE 69
rent-extraction by the main banks. Even be large (Bhagat et al. (1990), Rosett
more telling is the finding of Hoshi, (1990), Pontiff et al. (1990)). Of course,
Kashyap, and Scharfstein (1993) that, the significant protection of investors and
when regulatory change enabled Japanese employees in the United States may give an
firms to borrow in public capital markets unrepresentative picture of expropriation
and not just from the banks, high net worth in other countries.
firms jumped at the opportunity. This evi- Expropriation by large investors can be
dence suggests that, for these firms, the detrimental to efficiency through adverse
costs of bank finance exceeded its benefits. effects on the incentives of managers and
Franks and Mayer (1994) present a few employees, who might reduce their firm-
cases of German banks resisting takeovers specific human capital investments when
of their customer companies, either they are closely monitored by financiers or
because they were captured by the man- may be easily dismissed with the conse-
agement or because they feared losing quent loss of rents. Schmidt (1996) and
profits from the banking relationship. On Cremer (1995) make the general point of
the other hand, Gorton and Schmid (1996) how a principal’s high powered incentives
find no evidence of rent extraction by the can reduce an agent’s effort. In the case of
German banks. large shareholders, a similar point is made
The problem of expropriation by large by Burkart, Grom, and Panunzi (1997), in
investors becomes potentially more signifi- the case of takeovers by Shleifer and
cant when other investors are of a different Summers (1988), and in the case of banks
type, i.e., have a different pattern of cash by Rajan (1992). In all these examples, the
flow claims in the company. For example, if idea is that a large investor cannot commit
the large investor is an equity holder, he himself not to extract rents from the man-
may have an incentive to force the firm to ager ex post, and this adversely affects
take on too much risk, since he shares in ex ante managerial and employee incentives.
the upside while the other investors, who When the targets of expropriation by
might be creditors, bear all the costs of large investors are other investors, the
failure (Jensen and Meckling (1976)). adverse incentive effect of such expropria-
Alternatively, if the large investor is a cred- tion is the decline of external finance. Many
itor, he might cause the company to forego countries, for example, do not do much to
good investment projects because he bears protect minority investor rights, yet have
some of the cost, while the benefits accrue large investors in the form of families or
to the shareholders (Myers (1977)). banks. While this governance structure may
Finally, large investors might have a control managers, it leaves potential
greater incentive to redistribute rents from minority investors unprotected and hence
the employees to themselves than the man- unwilling to invest. Perhaps for this reason,
agers do (Shleifer and Summers (1988)). countries in Continental Europe, such is
The available evidence of redistributions Italy, Germany, and France, have relatively
between different types of claim holders in small public equity markets. In this regard,
the firm comes largely from corporate con- the existence of a large equity market in
trol transactions. Several studies, for example, Japan despite the weak protection of
ask whether shareholders expropriate minority investors is puzzling. The puzzle
bondholders in leveraged buy outs or may be explained by the predominance of
leveraged recapitalizations. Typically, these low powered incentives within large
redistributions are relatively small Japanese institutions or in the workings of
(Asquith and Wizman (1990)). Another reputations and implicit contracts in
group of studies ask whether takeovers lead Japan.
to large redistributions of wealth from the The Japanese example brings up a very
employees in the form of wage reductions, different view of large investors, namely
layoffs, and pension cutbacks. Again, these that they are too soft rather than too
redistributions typically do not appear to tough. This can be so for several reasons.
A SURVEY OF CORPORATE GOVERNANCE 71
First, large investors, whether shareholders organizational form that, for reasons
or creditors, may be soft when they them- discussed in this article, is rarely conducive
selves are corporations with their own to efficiency.
agency problems. Charkham (1994) shows,
for example, that German banks virtually
control themselves. “At general meetings in A. The debt versus equity choice
recent years, Deutsche Bank held voting Recent years saw a veritable flood of
rights for 47.2 percent of its shares, research on the debt contract as a mecha-
Dresdner for 59.25 percent, and Commerz- nism for solving agency problems. In this
bank for 30.29 percent” (p. 36). Moreover, new work, unlike in the Modigliani-Miller
banks have no incentive to discipline man- (1958) framework, where debt is associ-
agers, and some incentive to cater to them ated only with a particular pattern of cash
to get more business, as long as the firm is flows, the defining feature of debt is the
far away from default (Harris and Raviv ability of creditors to exercise control.
(1990)). Edwards and Fischer (1994) Specifically, debt is a contract in which a
summarize evidence suggesting that borrower gets some funds from the lender,
German banks are not nearly as active in and promises to make a prespecified
corporate governance as might be expected stream of future payments to the lender. In
given their lending power and control over addition, the borrower typically promises
equity votes. Second, some recent articles not to violate a range of covenants (Smith
show that, even if they don’t suffer from and Warner (1979)), such as maintaining
their own agency problems, large investors the value of assets inside the firm. If the
such as banks may be too soft because they borrower violates any covenant, and
fail to terminate unprofitable projects they especially if he defaults on a payment, the
have invested in when continuation is pre- lender gets certain rights, such as the
ferred to liquidation (Dewatripont and ability to repossess some of the firm’s
Maskin (1995), Gertner, Scharfstein, and assets (collateral) or the opportunity to
Stein (1994)). Finally, a large investor may throw the firm into bankruptcy. An essential
be rich enough that he prefers to maximize feature of debt, then, is that a failure by the
private benefits of control rather than borrower to adhere to the contract triggers
wealth. Unless he owns the entire firm, he the transfer of some control rights from
will not internalize the cost of these control him to the lender.
benefits to the other investors. While these The literature on debt can be usefully
arguments suggest a different set of problems divided into that before Grossman-Hart
with large investors, they too point to fail- (1986), and that after. Townsend (1978)
ures of large investors to force managers to and Gale and Hellwig (1985) consider
maximize profits and pay them out. models in which the borrower can abscond
with the profits of the firm. However, if the
lender is not repaid, he has the right to
VI. SPECIFIC GOVERNANCE investigate the books of the firm, and grab
ARRANGEMENTS its cash before the borrower can steal it.
Thus failure to repay triggers the transfer
In the previous sections, we discussed the of control over the assets from the bor-
roles of legal protection and concentrated rower to the lender. Gale and Hellwig
ownership in assuring that investors can (1985) show that the optimal contract that
collect their returns from firms. We have minimizes the expected investigation costs
postponed the discussion of specific is a debt contract. Grossman and Hart
contractual mechanisms used to address (1982) and Jensen (1986) model the role
the agency problem until this section. In of debt in committing the payout of free
particular, we now focus on debt and equity cash flows to investors. In Grossman and
as instruments of finance. In addition, we Hart (1982), in particular, default enables
discuss state ownership—a particular creditors to deprive the manager of the
72 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
benefits of control. A final important early Several other articles model the costs
article, which is not cast in the agency con- and benefits of the debt contract. The
text, but contains a highly relevant idea, is benefit is usually the reduction in the
Myers and Majluf (1984). They show that, agency cost, such as preventing the man-
because management has superior informa- ager from investing in negative net present
tion, external finance is costly. Moreover, value projects, or forcing him to sell assets
they argue that this adverse selection prob- that are worth more in alternative use. The
lem is minimized by the issuance of the main costs of debt are that firms may be
“safest” security, i.e., the security whose prevented from undertaking good projects
pricing is least sensitive to the manager’s because debt covenants keep them from
private information. Thus highly rated debt raising additional funds, or else they
with a fairly certain payoff stream is issued may be forced by creditors to liquidate
before equity, since equity is difficult to when it is not efficient to do so. Stulz
price without knowing the precise value of (1990), Diamond (1991), Harris and
the firm’s assets in place and future growth Raviv (1990), and Hart and Moore (1995)
opportunities. Debt is particularly easy to present some of the main models incorpo-
value where there is abundant collateral, so rating these ideas, whereas Lang, Ofek, and
that investors need only concern themselves Stulz (1996) present evidence indicating
with the value of the collateral and not with that leverage indeed curtails investment by
the valuation of the entire firm, as equity firms with poor prospects. Williamson
investors would need to. (1988) and Shleifer and Vishny (1992)
The next generation of papers adopts the argue that liquidations might be particu-
incomplete contracts framework more larly costly when alternative use of the
explicitly, and focuses on the transfer of asset is limited or when the potential
control from managers to creditors. Aghion buyers of the asset cannot raise funds
and Bolton (1992) use incomplete themselves. Dewatripont and Tirole (1994)
contract theory to characterize debt as an derive the optimal amount of debt in a
instrument whose holders take control of model where the tough negotiating stance
the firm in a bad state of the world. They of debt holders after default deters mana-
show that if the managerial benefits of gerial shirking ex ante. The model explains
control are higher in good states of the how the cash flow structure of debt as
world, then it may be efficient for managers senior claimant with little upside potential
to have control of assets in good states, and makes debt holders tough on managers
for creditors to have it in bad states. Their after a default. This makes it optimal to
model does not incorporate the idea that combine the specific form of cash flow
control reverts to the creditors in the case rights of debt with contingent control of
of default as opposed to some general bad the firm in the bad state. Berglof and von
state. Bolton and Scharfstein (1990) present Thadden (1994) similarly show why short
a model in which upon default, creditors term debt holders—who are the tough
have enough power to exclude the firm financiers in their model—should have
from the capital market, and hence stop control in the bad states. Many of these
future financing altogether. Hart and articles take advantage of Myers’ (1977)
Moore (1989, 1994a) explicitly model insight that debt overhang might be an
the idea that debt is a contract that gives effective deterrent to new financing and
the creditor the right to repossess investment.
collateral in case of default. Fear of such Because the rights of creditors are
liquidation keeps money flowing from the clearer, and violations of those rights are
debtors to the creditors. Hart and Moore’s easier to verify in courts, the existing liter-
models of debt show exactly how the ature has anointed debt as providing better
schedule of debt repayments depends on protection to outside investors than equity.
what creditors can realize once they gain However, the focus on large investors sheds
control. new light on the relative powers of debt and
A SURVEY OF CORPORATE GOVERNANCE 73
equity. Specifically, debt and equity ought (1991), Bolton and Scharfstein (1996)).
be compared in terms of the combination of In contrast, it may be easier to renegotiate
legal protections and ease of ownership with a bank.The difficulty of renegotiation,
concentration that each typically provides. and the power of dispersed creditors, might
First, does debt promote concentrated explain why public debt is an extremely
ownership? By far the dominant form of uncommon financing instrument, used only
lending around the world is bank lending. in a few developed countries, and even
Banks are usually large investors, who gain there much less than bank debt (Mayer
numerous control rights in the firm at the (1990)).
time of or even before default. For example, Unlike creditors, individual shareholders
the main bank can often take physical are not promised any payments in return
control of the firm’s bank account—which for their financial investment in the firm,
resides at that very bank—if it misses a although often they receive dividends at the
payment, thereby assuring fairly complete discretion of the board of directors. Unlike
control of the firm by the bank without creditors, individual shareholders have no
much involvement of the courts. This claim to specific assets of the firm, and
control is often bolstered by direct equity have no right to pull the collateral (one
ownership in the firm, as well as a large commonly studied exception is mutual
degree of monopoly power over any future funds, in which individual equity holders
credit extended to the firm (OECD can force a liquidation of their pro rata
(1995)). In contrast, American, Canadian, share of the assets and a repayment of its
and British firms make more extensive use value). Unlike creditors, shareholders do
of syndicated bank lending and even of not even have a final date at which the firm is
public debt, in which creditors are fairly liquidated and the proceeds are distributed.
dispersed (Mayer (1990)). In principle, they may never get anything
But even where debt is not very concen- back at all.
trated, the effective legal protection In addition to some relatively weak legal
afforded creditors is likely to be greater protections, the principal right that equity
than that enjoyed by dispersed equity holders typically get is the right to vote for
holders.The crucial feature of the creditors’ the board of directors. Even this right is
legal rights is that concerted action by not universal, since many countries have
multiple creditors is not required to take multiple classes of common stock, and
action against a delinquent debtor. The hence equity holders with inferior voting
legal obligation of the firm is an obligation rights get proportionately fewer votes than
to each and every creditor, and any of these their financial investment in the company.
creditors can typically sue the firm for pay- Because concerted action by a large group
ment of what is owed or for sale of assets. of shareholders is required to take control
Of course, once action is taken by one cred- via the voting mechanism, voting rights are
itor, the other creditors and the courts will of limited value unless they are concen-
take action to ensure that the first creditor trated. Most small shareholders do not
does not grab a disproportionate share for even have an incentive to become informed
himself. In fact, this ability to unilaterally on how to vote. Contacting and persuading
initiate the grab for assets in a multiple a large group of small shareholders
creditor situation lends the theoretical jus- through the proxy mechanism is difficult
tification for bankruptcy protection. and expensive, especially when the
Unlike equity, debt in a peculiar way may management stands in the way (Dodd and
be tougher when it is not concentrated. If a Warner (1983)). In contrast, when votes
borrower defaults on debt held by a large are concentrated—either in a large share
number of creditors, renegotiating with holding block or through a takeover—they
these creditors may be extremely difficult, become extremely valuable, since the party
and the borrower might be forced into that controls the concentrated votes can
bankruptcy (Gertner and Scharfstein make virtually all corporate decisions.
74 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Concentrated equity in this respect is more This may be especially problematic when
powerful than concentrated debt. The value the firm’s value consists primarily of future
of individual shares comes from the fact growth opportunities, but the bank’s debt
that the votes attached to them are valu- claim and unwillingness to take equity
able to those trying to control the firm, and give it little interest in the upside and a
the protection of minority shareholders distorted incentive to liquidate (Diamond
assures that those who have control must (1991), Hart and Moore (1995),
share some of the benefits with the minor- Dewatripont and Tirole (1994)). Rather
ity (Grossman and Hart (1988), Harris than give away control to the bank, such
and Raviv (1988)). firms often have highly concentrated equity
Because the equity holders have voting ownership by the entrepreneur and a
power and legal protection of minority venture capitalist. This may pave the way
shareholders, they have the ability to for some dispersed outside equity owner-
extract some payments from the managers ship as long as minority rights are well
in the form of dividends. Easterbrook enough protected.
(1984) articulates the agency theory of In fact, we do observe equity financing
dividend payments, in which dividends are primarily for young, growing firms, as well
for equity what interest is for debt: pay out as for firms in rapidly growing economies,
by the managers supported by the control whereas mature economies and mature
rights of the financiers, except in the case firms typically use bank finance when they
of equity these control rights are the voting rely on external funds at all (see Mayer
rights. More recently, Fluck (1995) and (1990), Singh (1995)). In the same spirit,
Myers (1995) present agency-theoretic Titman and Wessels (1988) and Rajan and
models of dividends, based on the idea that Zingales (1995) show for the United
shareholders can threaten to vote to fire States and several OECD economies
managers or liquidate the firm, and there- respectively that debt finance is most
fore managers pay dividends to hold off the common for firms with tangible assets.
shareholders.These models do not explicitly This analysis of equity financing still
address the free rider problem between leaves an important question open: how
shareholders; namely, how do they manage can firms raise equity finance in countries
to organize themselves to pose a threat to with virtually no protection of minority
the management when they are small investors, even if these countries are rap-
and dispersed? Concentration of equity idly growing? Singh (1995) provides some
ownership, or at least the threat of such evidence on the importance of equity
concentration, must be important to get financing in LDCs, although some of his
companies to pay dividends. data on equity financing might include pri-
One of the fundamental questions that vatizations and equity exchanges within
the equity contracts raise is how—given industrial groups, both of which often take
the weakness of control rights without the form of sales of large blocks and hence
concentration—do firms manage to issue need not reflect any minority purchases.
equity in any substantial amounts at all? One possible explanation is that, during a
Equity is the most suitable financing tool period of rapid economic growth, reputa-
when debt contracts are difficult to tional effects and the prospects of coming
enforce, i.e., when no specific collateral can back soon to the capital market sustain
be used to back credit and when near-term good behavior until the requisite institu-
cash flows are insufficient to service debt tions and legal protections are put in place
payments. Young firms, and firms with (Gomes (1996)). Investors can thus count
intangible assets, may need to be equity on reputation in the short run, and legal
financed simply because their assets have protection in the longer run when the firm’s
little or no liquidation value. If they are needs for access to capital markets are
financed by debt, their managers effectively smaller. Also, in some rapidly growing
give full control to the bank from the start. countries, such as Korea, the rates of return
A SURVEY OF CORPORATE GOVERNANCE 75
on investment may exceed the rates of there is some evidence that the way in
appropriation by the insiders. However, which profits are increased has to do with
another possibility is that speculative lower agency costs. Many LBOs are tar-
bubbles and investor overoptimism are geted at highly diversified firms, which sell
playing an important role in equity financing off many of their noncore divisions shortly
in rapidly growing economies.The available after the LBO (Bhagat, Shleifer, and Vishny
evidence does not satisfactorily account for (1990)). If the agency problem expresses
the puzzle of external equity financing in itself in the form of excessive size and
countries with only minimal legal protec- diversification, then the effect of debt over-
tion of investors. hang and large shareholders is to reduce
agency costs.
At the same time, LBOs illustrate the
B. LBOs potential costs of heavily concentrated
A remarkable recent phenomenon in the ownership. Jensen (1989a) conjectures
United States that illustrates both the ben- that because LBOs are so efficient, they
efits and the costs of having large investors would become a predominant organizational
is leveraged buy outs. In these transactions, form in the United States. Rappaport
shareholders of a publicly owned company (1990) in contrast argues that the heavy
are bought out by a new group of investors, oversight from investors might prevent
that usually includes old managers, a spe- future investment and growth, and hence be
cialized buyout firm, banks and public debt unattractive to the management. Bhagat,
holders (Jensen (1989a, 1989b)). With Shleifer, and Vishny (1990) argue that the
fewer constraints on compensation principal purpose of LBOs in the 1980’s
arrangements than when the firm was was to serve as a temporary financing tool
public, managers typically sharply increase for implementation of drastic short-run
their percentage ownership of the new improvements, such as divestitures. Kaplan
company, even though they take out some (1991) looks empirically at the question of
of their money invested in the firm (Kaplan whether LBOs are permanent organizations,
and Stein (1993)). The buyout firm typi- or whether, alternatively, they eventually
cally buys enough equity to control the return to the public equity market. His
firm. Most of the financing, however, comes evidence suggests that, while LBOs are not
from banks and from buyers of subordi- very short lived organizations, the median
nated public debt, which in the 1980s firm sells equity to the public within five to
became known as junk bonds. In some six years. Although this suggests that LBOs
cases, the decisions of the dispersed holders are not permanent organizations, Kaplan
of junk debt were coordinated by its under- also finds that even those firms issuing
writers. In short, LBOs had concentrated equity to the public retain a very heavy
equity ownership by managers and LBO concentration of both debt and equity
funds, as well as debt ownership by banks, ownership. Large investors remain even
and, in effect, the holders of public debt. when the original financing structure is too
Consistent with the idea that large tough to be permanent.
investors reduce agency problems, the
available evidence indicates that LBOs are C. Cooperatives and state ownership
efficient organizations. First, like other
takeovers, LBOs usually buy out the old We have suggested that, in some situations,
shareholders at a substantial premium, concentrated ownership may not be optimal
meaning at least prima facie that they were because nonshareholder constituencies
going to increase profits (DeAngelo, such as managers, employees, and con-
DeAngelo, and Rice (1984)). Second, there sumers are left with too few rents, and too
is direct evidence from the sample of LBOs little incentive to make relationship-specific
that subsequently went public that they do investments. In these situations, coopera-
increase profits (Kaplan (1989)). Third, tives might be a more efficient ownership
76 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
structure (Hansmann (1988), Hart and The view of corporate governance taken
Moore (1994b)). For example, private firms in this article helps explain the principal
with large investors might under-provide elements of the behavior of state firms.
quality or otherwise shortchange the firm’s While in theory these firms are controlled
stakeholders because of their single- by the public, the de facto control rights
minded focus on profits.This logic has been belong to the bureaucrats. These bureau-
used to explain why health care, child care, crats can be thought of as having
and even retailing are sometimes best pro- extremely concentrated control rights, but
vided by cooperatives, including consumer no significant cash flow rights because the
cooperatives. By voting on prices and qual- cash flow ownership of state firms is effec-
ity, stakeholders achieve a better outcome tively dispersed amongst the taxpayers of
than would a profit-maximizing owner. the country. Moreover, the bureaucrats
A similar argument has been used to typically have goals that are very different
justify state ownership of firms. Where from social welfare, and are dictated by
monopoly power, externalities, or distribu- their political interests (Shapiro and
tional issues raise concerns, private profit- Willig (1990), Boycko et al. (1996),
maximizing firms may fail to address these Shleifer and Vishny (1994)). For example,
concerns. A publicly spirited politician can they often cater to special interest groups
then improve efficiency by controlling the that help them win elections, such as
decisions of firms. Such social welfare public employee trade unions, which not
arguments underlie the traditional case for surprisingly typically strongly support
state ownership of railroads, electricity, state ownership (Lopez-de-Silanes,
prisons, schools, health care, and many Shleifer, and Vishny (1997)). In sum, the
other activities (Laffont and Tirole (1993), bureaucrats controlling state firms have
Sappington and Stiglitz (1987)). Versions at best only an indirect concern about
of this argument are used to justify state profits (because profits flow into the
ownership of industrial firms as well. government budget), and have objectives
With a few exceptions of activities where that are very different from the social
the argument for state ownership carries interest. Nonetheless, they have virtually
the day, such as police and prisons (Hart, complete power over these firms, and can
Shleifer, and Vishny (1997)), the reality of direct them to pursue any political
state ownership is broadly inconsistent objective. State ownership is then an
with this efficiency argument. First, state example of concentrated control with no
firms do not appear to serve the public cash flow rights and socially harmful
interest better than private firms do. For objectives. Viewed from this perspective,
example, in many countries state enterprises the inefficiency of state firms is not at all
are much worse polluters than private surprising.
firms. Indeed, the pollution problems are The recognition of enormous inefficiency
most severe in the former communist coun- of state firms, and the pressures on public
tries that were dominated by state firms budgets, have created a common response
(Grossman and Krueger (1993)). Second, around the world in the last few years,
contrary to the theory, state firms are typi- namely privatization. In most cases,
cally extremely inefficient, and their losses privatization replaces political control with
result in huge drains on their countries’ private control by outside investors. At the
treasuries (Kikeri, Nellis, and Shirley same time, privatization in most countries
(1992) and Boycko, Shleifer, and Vishny creates concentrated private cash flow
(1995) survey the relevant evidence). In ownership to go along with control. The
their frequent disregard of social objectives, result of the switch to these relatively more
as well as in their extreme inefficiency, the efficient ownership structures is typically a
behavior of state firms is inconsistent significant improvement in performance of
with the efficiency justification for their privatized firms (Megginson et al. (1994),
existence. Lopez-de-Silanes (1994)).
A SURVEY OF CORPORATE GOVERNANCE 77
The cases where privatization does not VII. WHICH SYSTEM IS THE BEST?
work as well as intended can also be under-
stood from the corporate governance Corporate governance mechanisms vary a
perspective. For example, when firms are great deal around the world. Firms in the
privatized without the creation of large United States and the United Kingdom
investors, agency costs of managerial substantially rely on legal protection of
control may rise even when the costs of investors. Large investors are less preva-
political control fall. In the United lent, except that ownership is concentrated
Kingdom, managers of privatized firms sporadically in the takeover process. In
such as water utilities receive large wage much of Continental Europe as well as in
increases (Wolfram (1995)). This outcome Japan, there is less reliance on elaborate
is not surprising, given that the controlling legal protections, and more reliance on
outside shareholders no longer exist in large investors and banks. Finally, in the
these firms, leaving managers with more rest of the world, ownership is typically
discretion. At the same time, we doubt that heavily concentrated in families, with a few
the problems of managerial discretion in large outside investors and banks. Legal
these companies are nearly as serious as the protection of investors is considerably
prior problems of political control. weaker than in Japan and Germany, let
Another example of postprivatization alone in Britain and the United States.This
difficulties with corporate governance is diversity of systems raises the obvious
Russia (Boycko et al. (1995)). For political question: what arrangement is the best
reasons, the Russian privatization has led from the viewpoint of attracting external
to controlling ownership by the manage- funds to firms? In this section, we attempt
ment of many companies. The management to deal with this question.
has almost complete control and substan-
tial cash flow rights, which can in principle A. Legal protection and large
lead to dramatically improved incentives.
investors
However, there are two problems—both of
which could have been predicted from the Our analysis leads us to conclude that both
theory. First, the virtual absence of protec- the legal protection of investors and some
tion of minority shareholders makes it form of concentrated ownership are
attractive for managers to divert resources essential elements of a good corporate gov-
from the firms despite their large personal ernance system. Large investors appear to
cash flow stakes, since in this way they do be necessary to force managers to distrib-
not need to share with outside investors at ute profits. These investors require at least
all. Second, managers in many cases are some basic legal rights, such as the voting
not competent to restructure the privatized rights or the power to pull collateral, to
firms, yet in virtue of their control rights exercise their power over the management.
remain on the job and “consume” the ben- If small investors are to be attracted to the
efits of control. In fact, some of the most business of financing companies, they as
successful privatizations in Russia have well require some legal protection against
been the ones where outside investors have expropriation by both the managers and
accumulated enough shares to either the large investors. Legal protection and
replace or otherwise control the manage- large investors are complementary in an
ment. Such outside investors have typically effective corporate governance system.
been less capable of diverting the profits Indeed, the successful corporate gover-
for themselves than the managers, as well nance systems, such as those of the United
as better capable of maximizing these profits. States, Germany, and Japan, rely on some
The example of the Russian privatization combination of concentrated ownership
vividly illustrates both the benefits and the and legal protection of investors. In the
costs of concentrated ownership without United States; both small and large share-
legal protection of minority investors. holders are protected through an extensive
78 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
system of rules that protects minority went public in Italy, compared to several
rights, allows for easy transfer of shares, thousand in the United States. Barca
keeps elections of directors relatively (1995) suggests that bank finance is also
uninhibited by managers, and gives share- difficult to obtain. Although Mayer (1990)
holders extensive powers to sue directors reports a significant amount of bank
for violations of fiduciary duty, including financing in Italy, most of it comes from
through class-action suits. Because of state bank financing of state firms. In Italy,
extensive bankruptcy protection of compa- most large firms not supported by the
nies, however, creditors in the United States government are family controlled and
have relatively fewer rights than do credi- internally financed.
tors in Germany and Japan. These legal Although there is little systematic evi-
rules support a system of active public dence available, most of the world appears
participation in the stock market, concen- to be more like Italy than like the United
tration of ownership through takeovers, but States, Germany, or Japan. A recent study
little governance by banks. of India, for example, shows that large
In Germany, creditors have stronger firms tend to be family controlled, and to
rights than they do in the United States, rely almost entirely on internal financing
but shareholder rights are weaker. Germany except when they get money from the
then has a system of governance by both government (Khanna and Palepu (1996)).
permanent large shareholders, for whom Latin American firms also face little
the existing legal rules suffice to exercise external corporate governance, and financing
their power, and by banks, but has virtually tends to be either internal or from
no participation by small investors in the government-controlled banks.The conclusion
market. Japan falls between the United we draw is simple: corporate governance
States and Germany in the degree of systems of the United States, Germany, and
protection of both shareholder and creditor Japan have more in common than is
rights, and as a result has powerful banks typically thought, namely a combination of
and powerful long term shareholders, large investors and a legal system that
although neither is evidently as powerful as protects investor rights. Corporate governance
they are in Germany. In addition, the systems elsewhere are less effective because
Japanese governance system has succeeded they lack the necessary legal protections.
in attracting small investors into the stock
market. Because both Germany and Japan B. Evolution of governance systems
have a system of permanent large investors,
hostile takeovers are rare in both countries. The above discussion does not address the
Although we compare the merits of the three question that has interested many people,
systems below, it is essential to remember namely which of the developed corporate
that all of them have effective legal protec- governance systems works the best? One
tion of at least some types of investors. could argue that, since all these systems
In much of the rest of the world, legal survived and the economies prospered, the
protection of investors is less substantial, governance systems of the United States,
either because laws are bad or because Japan, and Germany must be about equally
courts do not enforce these laws. As a con- good. However, recent research has shown
sequence, firms remain family-controlled that, historically, political pressures are as
and, even in some of the richest countries, important in the evolution of corporate
have difficulty raising outside funds, and governance systems as the economic ones.
finance most of their investment internally In a much-discussed recent book, Roe
(Mayer 1990). Pagano, Panetta, and (1994) argues that politics rather than
Zingales (1995) report the extraordinary economic efficiency shaped American
difficulties that firms face raising outside corporate law, at least at the Federal level.
funds in Italy. Over an 11-year period Roe provides a detailed account on how the
between 1982 and 1992, only 123 firms American political system systematically
A SURVEY OF CORPORATE GOVERNANCE 79
discouraged large investors. Banks, insurance have focused on is: what type of large
companies, mutual funds, and pension investors are best? How do U.S.-style
funds were all prevented from becoming takeovers compare to more permanent
influential in corporate affairs. The hostile large shareholders and creditors in West
political response to the 1980s takeovers Germany and Japan? We do not believe
can be viewed as a continuation of the that the available research provides a firm
promanagement and antilarge-shareholder answer to this question.
policies (Grundfest (1990), Jensen (1993)). Not surprisingly, the most enthusiastic
Roe does not explain whether the extremely assessments of American corporate gover-
fine development of the legal protection of nance system come from those who put
small shareholders in the United States is greater emphasis on the role of legal pro-
in part a response to the suppression of tection than on that of large investors
large investors, but this conclusion is (Easterbrook and Fischel (1991), Romano
actually suggested by some other work (1993a)). Romano (1993a) argues that
(e.g., Douglas (1940), Coffee (1991), competition between U.S. states has
Bhide (1993)). Roe’s conclusion is caused the State of Delaware, where many
nonetheless that the American system is far large companies are incorporated, to adopt
from efficient because of its discouragement corporate laws that effectively serve the
of the large investors. interests of shareholders, and thus secure
The trouble is, the argument that the effective corporate governance. Romano
political process accommodates the power- (1993a) even argues that Delaware
ful interests in the economy rather than adopted the most benign antitakeover
maximizing social welfare applies to legislation of all the states, thereby not pre-
Germany and Japan as well. Both countries cluding a future role for hostile takeovers.
have shaped their systems of powerful Easterbrook and Fischel (1991) do not
banks at the end of the 19th century, dur- discuss the role of large shareholders at all.
ing the period of rapid economic growth, Romano (1993b) believes that the fre-
and with strong support from the state quently mentioned hopes that institutional
(Gerschenkron (1962)). In both countries, investors in the United States become
the United States attempted to destroy the active, value-maximizing shareholders
powerful financial institutions during the (e.g., Black (1990)) are exaggerated. She
occupation after World War II (Adler is also skeptical about the potential gover-
(1949)), and in both countries it failed. nance role of banks. In short, the bet
Moreover, once German banks became suf- among these scholars is on the legal
ficiently powerful, they discouraged the protection of investors. To the extent that
introduction of disclosure rules, prohibitions takeovers complement this legal protection,
on insider trading, and other protections of they are viewed as sufficient.
minority shareholders—thus making sure In contrast, advocates of the German
that these investors never became a signif- and Japanese corporate governance system
icant economic or political force to protect point to the benefits of permanent long
their rights. Through this political channel, term investors relative to those of
the legal system has developed to accom- takeovers. Hoshi, Kashyap, and Scharfstein
modate the prevailing economic power, (1990, 1991) show that firms with a main
which happened to be the banks. banking relationship in Japan go through
Evolutionary arguments evidently do not financial distress with less economic dis-
adjudicate the question of which system is tress and better access to financing. In
more efficient. addition, a large theoretical and anecdotal
literature argues that the American corpo-
C. What kind of large investors? rate governance system, particularly
takeovers, imposes short horizons on the
The question that many of the comparisons behavior of corporate managers, and hence
of the United States, Japan, and Germany reduces the efficiency of investment
80 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
(Stein (1988, 1989), Shleifer and Vishny small investors from participating in financial
(1990)). The theories and the arguments markets. In sum, despite a great deal of
(Porter (1992)) in this area are remarkably controversy, we do not believe that either
short of any empirical support (see the theory or the evidence tells us which of
Poterba and Summers (1995)). Still, the the three principal corporate governance
superior performance of the Japanese and systems is the best. In this regard, we are
German economies, at least until the not surprised to see political and economic
1990s, has caused many to prefer their pressures for the three systems to move
governance systems to the American one toward each other, as exemplified by the
(see Aoki (1990), Roe (1993), and growing popularity of large shareholders in
Charkham (1994)). the United States, the emergence of public
We do not think that these debates have debt markets in Japan, and the increasing
been conclusive. True, American takeovers bank-bashing in Germany.
are a crude governance mechanism. But the At the same time, in thinking about the
U.S. economy has produced mechanisms of evolution of governance in transition
this kind repeatedly during the 20th economies, it is difficult to believe that
century, including mergers, proxy fights, either significant legal protection of
LBOs, and more recently vulture funds. investors or takeovers are likely to play a
Although many of these mechanisms run key role. In all likelihood, then, unless
into political trouble, new ones keep being Eastern Europe is stuck with insider
invented. The end of 1980s hostile domination and no private external finance
takeovers probably does not spell the end of at all (a risk in Russia), it will move toward
active large investors. Moreover, partly as a governance by banks and large shareholders.
result of takeovers, the American economy The early evidence from the Czech
in the 1980s went through a more radical, Republic (van Wijnbergen and Mancini
and possibly effective, restructuring than (1995)) and Russia (Blasi and Shleifer
the economies of Japan and Western (1996)) indeed suggests that large share-
Europe. Finally, because of extensive legal holders, which in the Czech Republic are
protection of small investors, young often bank-controlled mutual funds, play a
American firms are able to raise capital in central role in corporate governance. It
the stock market better than firms else- would be extremely fortunate if transition
where in the world. It is difficult to dismiss economies managed to approach the
the U.S. corporate governance system in corporate governance systems of
light of these basic facts. Germany and Japan, particularly in the
On the other hand, permanent large dimension of the legal protection of
shareholders and banks, such as those dom- investors. But this does not imply that the
inating corporate governance in Japan and United States should move in the same
Germany, obviously have some advantages, direction as well.
such as the ability to influence corporate
management by patient, informed investors.
These investors may be better able to help VIII. CONCLUSION
distressed firms as well. Still, there are
serious questions about the effectiveness of In the course of surveying the research on
these investors, largely because their corporate governance, we try to convey a
toughness is in doubt. As Charkham particular structure of this field. Corporate
(1994) has shown, German banks are large governance deals with the agency problem:
public institutions that effectively control the separation of management and
themselves. There is little evidence from finance. The fundamental question of
either Japan or Germany that banks are corporate governance is how to assure
very tough in corporate governance. financiers that they get a return on their
Finally, at least in Germany, large-investor- financial investment. We begin this survey
oriented governance system discourages by showing that the agency problem is
A SURVEY OF CORPORATE GOVERNANCE 81
Blasi, Joseph, and Andrei Shleifer, 1996, Comment, Robert, and Gregg Jarrell, 1995,
Corporate governance in Russia: An initial Corporate focus and stock returns, Journal of
look, in Roman Frydman, Cheryl W. Gray, and Financial Economics 37, 67–87.
Andrzej Rapaczynski, Eds.: Corporate Comment, Robert, and G. William Schwert,
Governance in Central Europe and Russia: 1995, Poison or placebo? Evidence on the
Vol. 2 Insiders and the State (Central deterrent and wealth effects of modern anti-
European University Press, Budapest). takeover measures, Journal of Financial
Bolton, Patrick, and David Scharfstein, 1990, Economics 39, 3–44.
A theory of predation based on agency Coughlan, Anne, and Ronald Schmidt, 1985,
problems in financial contracting, American Executive compensation, management
Economic Review 80, 94–106. turnover, and firm performance: An empirical
Bolton, Patrick, and David Scharfstein, 1996, investigation, Journal of Accounting and
Optimal debt structure and the number of cred- Economics 7, 43–66.
itors, Journal of Political Economy 104, 1–25. Cremer, Jacques, 1995, Arm’s length relation-
Boycko, Maxim, Andrei Shleifer, and Robert W. ships, Quarterly Journal of Economics 110,
Vishny, 1993, Privatizing Russia, Brookings 275–296.
Papers on Economic Activity, 139–192. Dann, Larry, and Harry DeAngelo, 1983,
Boycko, Maxim, Andrei Shleifer, and Robert W. Standstill agreements, privately negotiated
Vishny, 1995, Privatizing Russia (M.I.T. stock repurchases, and the market for corpo-
Press, Cambridge, Mass.). rate control, Journal of Financial Economics
Boycko, Maxim, Andrei Shleifer, and Robert W. 11, 275–300.
Vishny, 1996, A theory of privatization, Paish DeAngelo, Harry, and Linda DeAngelo, 1985,
lecture, Economic Journal 106, 309–319. Managerial ownership of voting rights,
Brudney, Victor, and Marvin A. Chirelstein, Journal of Financial Economics 14, 33–69.
1978, A restatement of corporate freeze- DeAngelo, Harry, Linda DeAngelo, and Edward
outs, Yale Law Journal 87, 1354–1375. Rice, 1984, Going private: Minority freeze-
Bulow, Jeremy, and Kenneth Rogoff, 1989, outs and stockholder wealth, Journal of Law
A constant recontracting model of sovereign and Economics 27, 367–401.
debt, Journal of Political Economy 97, DeAngelo, Harry, and Edward Rice, 1983,
155–178. Antitakeover amendments and stockholder
Burkart, M., Denis Gromb, and Fausto Panunzi, wealth, Journal of Financial Economics 11,
1997, Large shareholders, monitoring, and 329–360.
fiduciary duty, Quarterly Journal of De Long, J. Bradford, 1991, Did Morgan’s Men
Economics 112. Add Value? An Economist’s Perspective on
Burrough, Bryan, and John Helyar, 1990, Financial Capitalism, in Peter Temin, Ed.:
Barbarians at the Gate: The Fall of RJR Inside the Business Enterprise: Historical
Nabisco, (Harper & Row, New York). Perspectives on the Use of Information
Charkham, Jonathan, 1994, Keeping Good (University Press, Chicago).
Company: A Study of Corporate Governance De Long, J. Bradford, Andrei Shleifer, Lawrence
in Five Countries (Clarendon Press, Oxford). Summers, and Robert Waldmann, 1989, The
Clark, Robert, 1985, Agency costs versus fiduci- size and incidence of the losses from noise
ary duties, in John Pratt and Richard trading, Journal of Finance 44, 681–696.
Zeckhauser, Eds.: Principals and Agents: The De Long, J. Bradford, Andrei Shleifer,
Structure of Business (Harvard Business Lawrence Summers, and Robert Waldmann,
School Press, Cambridge, Mass.). 1990, Noise trader risk in financial mar-
Coase, Ronald, 1937, The nature of the firm, kets, Journal of Political Economy 98,
Economica 4, 386–405. 703–738.
Coase, Ronald, 1960, The problem of social Demsetz, Harold, 1983, The structure of owner-
cost, Journal of Law and Economics 3, 1–44. ship and the theory of the firm, Journal of
Coffee, John, 1991, Liquidity versus control:The Law and Economics 26, 375–390.
institutional investor as corporate monitor, Demsetz, Harold, and Kenneth Lehn, 1985, The
Columbia Law Review 91, 1277–1368. structure of corporate ownership: Causes and
84 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
consequences, Journal of Political Economy Eisenberg, Melvin, 1976, The Structure of the
93, 1155–1177. Corporation: A Legal Analysis (Little, Brown
Denis, David, and Jan Serrano, 1996, Active and Co., Boston).
investors and management turnover following Fama, Eugene, 1980, Agency problems and the
unsuccessful control contests, Journal of theory of the firm, Journal of Political
Financial Economics 40, 239–266. Economy 88, 288–307.
Dewatripont, Mathias, and Eric Maskin, 1995, Fama, Eugene, Lawrence Fisher, Michael
Credit and efficiency in centralized and Jensen, and Richard Roll, 1969, The adjust-
decentralized economies, Review of ment of stock prices to new information,
Economic Studies 62, 541–556. International Economic Review 10, 1–21.
Dewatripont, Mathias, and Jean Tirole, 1994, Fama, Eugene, and Michael Jensen, 1983a,
A theory of debt and equity: Diversity of Separation of ownership and control, Journal
securities and manager-shareholder congru- of Law and Economics 26, 301–325.
ence, Quarterly Journal of Economics 109, Fama, Eugene, and Michael Jensen, 1983b,
1027–1054. Agency problems and residuals claims,
Dewatripont, Mathias, and Jean Tirole, 1995, Journal of Law and Economics 26, 327–349.
The Prudential Regulation of Banks (MIT Fluck, Zsussanna, 1995, The optimality of debt
Press, Cambridge). versus outside equity, manuscript, New York
Diamond, Douglas, 1984, Financial intermedia- University.
tion and delegated monitoring, Review of Franks, Julian, and Colin Mayer, 1990,
Economic Studies 51, 393–414. Takeovers: Capital markets and corporate
Diamond, Douglas, 1989, Reputation acquisi- control: A study of France, Germany, and the
tion in debt markets, Journal of Political UK, Economic Policy: A European Forum
Economy 97, 828–862. 10, 189–231.
Diamond, Douglas, 1991, Debt maturity struc- Franks, Julian, and Colin Mayer, 1994,The own-
ture and liquidity risk, Quarterly Journal of ership and control of German corporations,
Economics 106, 1027–1054. manuscript, London Business School.
Dodd, Peter, and Jerold Warner, 1983, On cor- Galbraith, John Kenneth, 1955, The Great
porate governance: A study of proxy contests, Crash, 1929 (Houghton Mifflin, Boston).
Journal of Financial Economics 11, Gale, Douglas, and Martin Hellwig, 1985,
401–439. Incentive-compatible debt contracts:The one-
Douglas, William O., 1940, Democracy and period problem, Review of Economic Studies
Finance (Yale University Press, New Haven). 52, 647–663.
Downs, Anthony, 1957, An Economic Theory of Gerschenkron, Alexander, 1962, Economic
Democracy (Harper, New York). Backwardness in Historical Perspective
Easterbrook, Frank, 1984, Two agency-cost (Harvard University Press, Cambridge MA).
explanations of dividends, American Gertner, Robert, and David Scharfstein, 1991,
Economic Review 74, 650–659. A theory of workouts and the effects of
Easterbrook, Frank, and Daniel Fischel, 1983, reorganization law, Journal of Finance 46,
Voting in corporate law, Journal of Law and 1189–1222.
Economics 26, 395–427. Gertner, Robert, David Scharfstein, and Jeremy
Easterbrook, Frank, and Daniel Fischel, 1991, Stein, 1994, Internal versus external capital
The Economic Structure of Corporate Law markets, Quarterly Journal of Economics
(Harvard University Press, Cambridge, Mass.). 109, 1211–1230.
Eaton, Jonathan, and Mark Gersovitz, 1981, Gilson, Stuart, 1990, Bankruptcy, boards,
Debt with potential repudiation: Theoretical banks, and block holders, Journal of
and empirical analysis, Review of Economic Financial Economics 27, 355–387.
Studies 48, 289–309. Gilson, Stuart, Kose John, and Larry Lang,
Edwards, Jeremy, and Klaus Fischer, 1994, 1990,Troubled debt restructurings: an empir-
Banks, Finance and Investment in West ical study of private reorganization of firms in
Germany Since 1970 (Cambridge University default, Journal of Financial Economics 27,
Press, Cambridge). 315–355.
A SURVEY OF CORPORATE GOVERNANCE 85
Gomes, Armando, 1996, Dynamics of stock Hart, Oliver, and John Moore, 1989, Default and
prices, manager ownership, and private bene- renegotiation: A dynamic model of debt, man-
fits of control, manuscript, Harvard University. uscript, Harvard University.
Gorton, Gary, and Frank Schmid, 1996, Hart, Oliver, and John Moore, 1990, Property
Universal banking and the performance of rights and the nature of the firm, Journal of
German firms, Working Paper 5453, National Political Economy 98, 1119–1158.
Bureau of Economic Research, Cambridge, MA. Hart, Oliver, and John Moore, 1994a, A theory
Grossman, Sanford, and Oliver Hart, 1980, of debt based on the inalienability of human
Takeover bids, the free-rider problem, and the capital, Quarterly Journal of Economics 109,
theory of the corporation, Bell Journal of 841–879.
Economics 11, 42–64. Hart, Oliver, and John Moore, 1994b,The gover-
Grossman, Sanford, and Oliver Hart, 1982, nance of exchanges: Members’ cooperatives
Corporate financial structure and managerial versus outside ownership, manuscript,
incentives, in J. J. McCall Ed.: The Economics Harvard University.
of Information and Uncertainty (University of Hart, Oliver, and John Moore, 1995, Debt and
Chicago Press, Chicago), 123–155. seniority: An analysis of the role of hard
Grossman, Sanford, and Oliver Hart, 1986, The claims in constraining management,
costs and benefits of ownership: A theory of American Economic Review 85, 567–585.
vertical and lateral integration, Journal of Hart, Oliver, Andrei Shleifer, and Robert Vishny,
Political Economy 94, 691–719. 1997, The proper scope of government:
Grossman, Sanford, and Oliver Hart, 1988, One Theory and an application to prisons,
share-one vote and the market for corporate Quarterly Journal of Economics 112.
control, Journal of Financial Economics 20, Haubrich, Joseph, 1994, Risk aversion, perform-
175–202. ance pay, and the principal-agent problem,
Grossman, Gene, and Alan Krueger, 1993, Journal of Political Economy 102, 258–276.
Environmental impact of a North-American Hellwig, Martin, 1994, Banking and finance at the
free trade agreement, in The U.S.-Mexico end of the twentieth century, Wirtschaftswis-
Free Trade Agreement, P. Garber, Ed.: (MIT senschaftliches Zentrum Discussion Paper No.
Press, Cambridge, Mass.). 9426, University of Basel.
Grundfest, Joseph, 1990, Subordination of Holderness, Clifford, and Dennis Sheehan,
American capital, Journal of Financial 1988a, What constrains managers who own
Economics 27, 89–114. large blocks of stock? Managerial Economics
Hansmann, Henry, 1988, Ownership of the firm, Research Center Working Paper 88–07,
Journal of Law, Economics, and University of Rochester.
Organization 4, 267–304. Holderness, Clifford, and Dennis Sheehan,
Hansmann, Henry, 1996, The Ownership of 1988b, The role of majority shareholders in
Enterprise (Harvard University Press, publicly held corporations: An exploratory
Cambridge, MA). analysis, Journal of Financial Economics 20,
Harris, Milton, and Artur Raviv, 1988, 317–346.
Corporate governance: Voting rights and Holmstrom, Bengt, 1979, Moral hazard and
majority rules, Journal of Financial observability, Bell Journal of Economics 10,
Economics 20, 203–235. 4–29.
Harris, Milton, and Artur Raviv, 1990, Capital Holmstrom, Bengt, 1982, Managerial incentive
structure and the informational role of debt, problems: A dynamic perspective, in Essays in
Journal of Finance 45, 321–350. Economics and Management in Honor of
Hart, Oliver, 1995, Firms, Contracts, and Lars Wahlbeck (Swedish School of
Financial Structure (Oxford University Economics, Helsinki), 210–235.
Press, London). Holmstrom, Bengt, and Jean Tirole, 1989, The
Hart, Oliver, and Bengt Holmstrom, 1987, The theory of the firm, in R. Schmalensee and
Theory of Contracts, in T. Bewley, Ed.: R. D. Willig, Eds.: Handbook of Industrial
Advances in Economic Theory (University Organization, Vol. 1, (North-Holland,
Press, Cambridge, U.K.), 294–351. Amsterdam), 61–133.
86 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Hopt, Klaus, and Gunther Teubner, Eds., 1985, agency costs, and ownership structure,
Corporate Governance and Directors’ Journal of Financial Economics 3, 305–360.
Liabilities, (de Gruyter, Berlin). Jensen, Michael, and Kevin Murphy, 1990,
Horner, M., 1988, The value of the corporate Performance pay and top management incen-
voting right, Journal of Banking and Finance tives, Journal of Political Economy 98,
12, 69–83. 225–263.
Hoshi, Takeo, Anil Kashyap, and David Jensen, Michael, and Richard Ruback, 1983,
Scharfstein, 1990,The role of banks in reduc- The market for corporate control: The scien-
ing the costs of financial distress in Japan, tific evidence, Journal of Financial
Journal of Financial Economics 27, 67–88. Economics 11, 5–50.
Hoshi, Takeo, Anil Kashyap, and David Johnson, W. Bruce, Robert Magee, Nandu
Scharfstein, 1991, Corporate structure, liq- Nagarajan, and Harry Newman, 1985, An
uidity, and investment: Evidence from analysis of the stock price reaction to sudden
Japanese industrial groups, Quarterly Journal executive deaths: Implications for the man-
of Economics 106, 33–60. agement labor market, Journal of Accounting
Hoshi, Takeo, Anil Kashyap, and David & Economics 7, 151–174.
Scharfstein, 1993, The choice between public Kang, Jun-Koo, and Anil Shivdasani, 1995, Firm
and private debt: An analysis of post- performance, corporate governance, and top
deregulation corporate finance in Japan, executive turnover in Japan, Journal of
National Bureau of Economic Research Financial Economics 38, 29–58.
Working Paper 4421, Cambridge, Mass. Kaplan, Steven, 1989, The effects of manage-
Hunt, Bishop Carleton, 1936, The Development ment buyouts on operating performance and
of the Business Corporation in England, value, Journal of Financial Economics 24,
1800–1867, (Harvard University Press, 217–254.
Cambridge, Mass.). Kaplan, Steven, 1991, The staying power of
Jarrell, Gregg, and Annette Poulsen, 1988a, leveraged buyouts, Journal of Financial
Shark repellents and stock prices: The effects Economics 29, 287–313.
of antitakeover amendments since 1980, Kaplan, Steven, 1994a,Top executives, turnover,
Journal of Financial Economics 19, and firm performance in Germany, Journal of
127–168. Law, Economics, and Organization 10,
Jarrell, Gregg, and Annette Poulsen, 1988b, 142–159.
Dual-class recapitalizations as antitakeover Kaplan, Steven, 1994b, Top executive rewards
mechanisms: The recent evidence, Journal of and firm performance: A comparison of
Financial Economics 20, 129–152. Japan and the United States, Journal of
Jensen, Michael, 1986, Agency costs of free Political Economy 102, 510–546.
cash flow, corporate finance, and takeovers, Kaplan, Steven, and Bernadette Minton, 1994,
American Economic Review 76, 323–329. Appointments of outsiders to Japanese
Jensen, Michael, 1988, Takeovers: Their causes boards: Determinants and implications for
and consequences, Journal of Economic managers, Journal of Financial Economics
Perspectives 2, 21–48. 36, 225–257.
Jensen, Michael, 1989a, Eclipse of the public Kaplan, Steven, and Jeremy Stein, 1993, The
corporation, Harvard Business Review 67, Evolution of Buy out Pricing and Financial
60–70. Structure in the 1980s, Quarterly Journal of
Jensen, Michael, 1989b, Active investors, LBOs Economics 108, 313–357.
and the privatization of bankruptcy, Journal Kaplan, Steven, and Michael Weisbach, 1992,
of Applied Corporate Finance 2, 35–44. The success of acquisitions: Evidence from
Jensen, Michael, 1993, The modern industrial divestitures, Journal of Finance 47, 107–138.
revolution, exit, and the failure of internal Khanna, Tarun, and Krishna Palepu, 1996,
control systems, Journal of Finance 48, Corporate scope and severe market imperfec-
831–880. tions: An empirical analysis of diversified
Jensen, Michael, and William Meckling, 1976, business groups in India, manuscript, Harvard
Theory of the firm: Managerial behavior, University.
A SURVEY OF CORPORATE GOVERNANCE 87
Kikeri, Sunita, John Nellis, and Mary Shirley, Mace, Myles, 1971, Directors, Myth, and
1992, Privatization: The Lessons of Experience, Reality (Harvard Business School Press,
(The World Bank, Washington, D.C.). Boston).
Kindelberger, Charles P., 1978, Manias, Panics, Malatesta, Paul, and Ralph Walkling, 1988,
and Crashes, (Basic Books, New York). Poison Pill Securities: Stockholder Wealth,
Kreps, David, 1990, Corporate Culture and Profitability, and Ownership Structure, Journal
Economic Theory, in J. Alt and K. Shepsle, of Financial Economics 20, 347–376.
Eds.: Perspectives on Positive Political Manne, Henry, 1965, Mergers and the market
Economy (University Press, Cambridge). for corporate control, Journal of Political
Laffont, Jean-Jacques, and Jean Tirole, 1993, Economy 75, 110–126.
A Theory of Incentives in Regulation and Marris, Robin, 1964, The Economic Theory of
Procurement (MIT Press, Cambridge, Mass.). Managerial Capitalism (Free Press of
Lambert, Richard, and David Larcker, 1985, Glencoe, Illinois).
Golden parachutes, executive decision- Martin, Kenneth, and John McConnell, 1991,
making, and shareholder wealth, Journal of Corporate performance, corporate takeovers,
Accounting & Economics 7, 179–203. and management turnover, Journal of
Lang, Larry, and René Stulz, 1994, Tobin’s Q, Finance 46, 671–688.
corporate diversification, and firm perform- Mayer, Colin, 1990, Financial systems, corpo-
ance, Journal of Political Economy 102, rate finance, and economic development, in
1248–1280. R. G. Hubbard, Ed.: Asymmetric Information,
Lang, Larry, Eli Ofek, and René Stulz, 1996, Corporate Finance, and Investment
Leverage, investment, and firm growth, (University of Chicago Press, Chicago).
Journal of Financial Economics 40, 3–30. McConnell, John, and Chris Muscarella, 1986,
Lang, Larry, René Stulz, and Ralph Walkling, Corporate capital expenditure decisions and
1991, A test of the free cash flow hypothesis: the market value of the firm, Journal of
The case of bidder returns, Journal of Financial Economics 14, 399–422.
Financial Economics 29, 315–336. McConnell, John, and Henri Servaes, 1990,
Loughran, Timothy, Jay Ritter, and Kristian Additional evidence on equity ownership and
Rydqvist, 1994, Initial public offerings: corporate value, Journal of Financial
International insights, Pacific-Basin Finance Economics 27, 595–612.
Journal 2, 165–199. Megginson, William, Robert Nash, and Matthias
Lease, Ronald, John McConnell, and Wayne van Randenborgh, 1994, The financial and
Mikkelson, 1983,The market value of control operating performance of newly privatized
in publicly traded corporations, Journal of firms, Journal of Finance 49, 403–452.
Financial Economics 11, 439–471. Mirrlees, James, 1976, The optimal structure of
Lease, Ronald, John McConnell, and Wayne incentives and authority within an organiza-
Mikkelson, 1984, The market value of differ- tion, Bell Journal of Economics 7, 105–131.
ential voting rights in closely held corpora- Modigliani, Franco, and Merton Miller, 1958,
tions, Journal of Business 57, 443–467. The cost of capital, corporation finance, and
Levy, Haim, 1982, Economic valuation of voting the theory of investment, American Economic
power of common stock, Journal of Finance Review 48, 261–297.
38, 79–93. Morck, Randall, Andrei Shleifer, and Robert
Lewellen, Wilbur, Claudio Loderer, and Ahron Vishny, 1988a, Characteristics of targets of
Rosenfeld, 1985, Merger decisions and execu- hostile and friendly takeovers, in A. Auerbach,
tive stock ownership in acquiring firms, Journal Ed.: Corporate Takeovers: Causes and
of Accounting & Economics 7, 209–231. Consequences (University of Chicago Press,
Lopez-de-Silanes, Florencio, 1994, Chicago).
Determinants of privatization prices, manu- Morck, Randall, Andrei Shleifer, and Robert
script, Harvard University. Vishny, 1988b, Management ownership and
Lopez-de-Silanes, Florencio, Andrei Shleifer, and market valuation: An empirical analysis,
Robert W. Vishny, 1997, Privatization in the Journal of Financial Economics 20,
United States, Rand Journal of Economics 28. 293–315.
88 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Morck, Randall, Andrei Shleifer, and Robert Pound, John, 1988, Proxy contests and the
Vishny, 1989, Alternative Mechanisms of efficiency of shareholder oversight, Journal of
Corporate Control, American Economic Financial Economics 20, 237–265.
Review 79, 842–852. Prowse, Stephen, 1990, Institutional investment
Morck, Randall, Andrei Shleifer, and Robert patterns and corporate financial behavior in
Vishny, 1990, Do managerial objectives drive the United States and Japan, Journal of
bad acquisitions? Journal of Finance 45, Financial Economics 27, 43–66.
31–48. Rajan, Raghuram, 1992, Insiders and Outsiders:
Murphy, Kevin J., 1985, Corporate performance The choice between relationship and arms-
and managerial remuneration: An empirical length debt, Journal of Finance 47,
analysis, Journal of Accounting & Economics 1367–1400.
7, 11–41. Rajan, Raghuram, and Luigi Zingales, 1995,
Myers, Stuart, 1977, Determinants of corporate What do we know about capital structure?
borrowing, Journal of Financial Economics Some evidence from international data,
5, 147–175. Journal of Finance 50, 1421–1460.
Myers, Stuart, 1995, Inside and outside equity Rappaport, Alfred, 1990, The staying power of
financing, manuscript, Massachusetts the public corporation, Harvard Business
Institute of Technology. Review 1, 96–104.
Myers, Stuart, and Nicholas Majluf, 1984, Ritter, Jay, 1991, The long run performance of
Corporate financing and investment decisions IPOs, Journal of Finance 46, 3–27.
when firms have information that investors do Roe, Mark, 1993, Some differences in corporate
not have, Journal of Financial Economics 13, structure in Germany, Japan, and the United
187–221. States, The Yale Law Review 102,
New York Times, 1986, Tax trial only one of 1927–2003.
Posner problems, June 23, Section D, 3. Roe, Mark, 1994, Strong Managers Weak
Organization for Economic Cooperation and Owners: The Political Roots of American
Development, 1995, Corporate governance Corporate Finance (University Press,
environments in OECD countries, February 1. Princeton, N.J.).
Owen, Thomas, 1991, The Corporation under Roll, Richard, 1986, The hubris hypothesis of
Russian Law, 1800–1917: A Study in corporate takeovers, Journal of Business 59,
Tsarist Economic Policy (Cambridge 197–216.
University Press, New York). Ross, Stephen, 1973, The economic theory of
Pagano, Marco, F. Panetta, and Luigi agency: The principal’s problem, American
Zingales, 1995, Why do companies go Economic Review 63, 134–139.
public? An empirical analysis, manuscript, Rossett, Joshua, G., 1990, Do union wealth
Graduate School of Business, University of concessions explain takeover premiums? The
Chicago. evidence on contract wages, Journal of
Palepu, Krishna, 1986, Predicting takeover Financial Economics 27, 263–282.
targets: A methodological and empirical Romano, Roberta, 1993a, The Genius of
analysis, Journal of Accounting and American Corporate Law (American
Economics 8, 3–35. Enterprise Institute Press, Washington, D.C.).
Pontiff, Jeffrey, Andrei Shleifer, and Michael Romano, Roberta, 1993c, Public pension fund
Weisbach, 1990, Reversions of excess pension activism in corporate governance reconsid-
assets after takeovers, Rand Journal of ered, Columbia Law Review 93, 795–853.
Economics 21, 600–613. Rydqvist, Kristian, 1987, Empirical investiga-
Porter, Michael, 1992, Capital disadvantage: tion of the voting premium, Working Paper
America’s failing capital investment system, 35, Northwestern University.
Harvard Business Review, Sept.–Oct., 65–83. Ryngaert, Michael, 1988, The effect of poison
Poterba, James M., and Lawrence H. Summers, pill securities on shareholder wealth, Journal
1995, A survey of U.S. companies, time hori- of Financial Economics 20, 377–417.
zons, and hurdle rates, Sloan Management Sappington, David, and Joseph Stiglitz, 1987,
Review 37, 43–53. Privatization, information, and incentives,
A SURVEY OF CORPORATE GOVERNANCE 89
Journal of Policy Analysis and Management Singh, Ajit, 1995, Corporate financial patterns
6, 567–582. in industrializing economies: A comparative
Scharfstein, David, 1988, The disciplinary role international study, Technical Paper 2,
of takeovers, Review of Economic Studies 55, April, Washington, DC: World Bank and
185–199. International Finance Corporation.
Schmidt, Klaus, 1996, The costs and benefits of Smith, Clifford, and Jerold Warner, 1979, On
privatization, The Journal of Law, Economics financial contracting: An analysis of bond
& Organization 12, 1–24. covenants, Journal of Financial Economics 7,
Shapiro, Carl, and Robert D. Willig, 1990, 117–161.
Economic rationales for the scope of privati- Stein, Jeremy, 1988, Takeover threats and man-
zation, in E. N. Suleiman and J. Waterbury, agerial myopia, Journal of Political Economy
Eds.: The Political Economy of Public Sector 96, 61–80.
Reform and Privatization (Westview Press, Stein, Jeremy, 1989, Efficient capital markets,
London), 55–87. inefficient firms: A model of myopic corpo-
Shivdasani, Anil, 1993, Board composition, rate behavior, Quarterly Journal of
ownership structure, and hostile takeovers, Economics 104, 655–669.
Journal of Accounting and Economics 16, Stigler, George, 1958, The economies of scale,
167–198. Journal of Law and Economics 1, 54–71.
Shleifer, Andrei, and Lawrence Summers, 1988, Stiglitz, Joseph, 1975, Incentives, risk, and
Breach of trust in hostile takeovers, in A. J. information: Notes toward a theory of hierar-
Auerbach, Ed.: Corporate Takeovers: Causes chy, Bell Journal of Economics 6, 552–579.
and Consequence (University of Chicago Stulz, René, 1988, Managerial control of voting
Press, Chicago), 65–88. rights, Journal of Financial Economics 20,
Shleifer, Andrei, and Robert Vishny, 1986a, 25–59.
Greenmail, white knights, and shareholder’s Stulz, René, 1990, Managerial discretion and
interest, Rand Journal of Economics 17, optimal financing policies, Journal of
293–309. Financial Economics 26, 3–27.
Shleifer, Andrei, and Robert Vishny, 1986b, The Economist, June 1995, A Survey of Korea.
Large shareholders and corporate control, Teoh, S. H., Ivo Welch, and T. J. Wong, 1995,
Journal of Political Economy 94, 461–488. Earnings management and post-issue
Shleifer, Andrei, and Robert Vishny, 1988, under performance of seasoned equity
Value maximization and the acquisition offerings, manuscript, University of
process, Journal of Economic Perspectives California.
2, 7–20. Tirole, Jean, 1994, Incomplete contracts:Where
Shleifer, Andrei, and Robert Vishny, 1989, do we stand? manuscript, IDEI, Toulouse,
Management entrenchment: The case of France.
manager-specific investments, Journal of Titman, Sheridan, and Roberto Wessels, 1988,
Financial Economics 25, 123–140. The determinants of capital structure choice,
Shleifer, Andrei, and Robert Vishny, 1990, Journal of Finance 43, 1–19.
Equilibrium short horizons of investors and Townsend, Robert, 1978, Optimal contracts and
firms, American Economic Review Papers competitive markets with costly state verifi-
and Proceedings 80, 148–153. cation, Journal of Economic Theory 21,
Shleifer, Andrei, and Robert Vishny, 1992, 265–293.
Liquidation values and debt capacity: A mar- Tunc, Andre, 1991, Corporate Law, in R.
ket equilibrium approach, Journal of Finance Buxbaum, G. Hertig, A. Hirsch, and K. Hopt,
47, 1343–1366. Eds.: European Business Law: Legal and
Shleifer, Andrei, and Robert Vishny, 1993, Economic Analysis of Integration and
Corruption, Quarterly Journal of Economics Harmonization (de Gruyter, Berlin).
108, 599–618. Walkling, Ralph, and Michael Long, 1984,
Shleifer, Andrei, and Robert Vishny, 1994, Agency theory, managerial welfare, and
Politicians and firms, Quarterly Journal of takeover bid resistance, Rand Journal of
Economics 109, 995–1025. Economics 15, 54–68.
90 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Warner, Jerold, Ron Watts, and Karen Wruck, Williamson, Oliver, 1985, The Economic
1988, Stock prices and top management Institutions of Capitalism (Free Press, New
changes, Journal of Financial Economics 20, York).
461–492. Williamson, Oliver, 1988, Corporate finance and
Weinstein, David E., and Yishay Yafeh, 1994, On corporate governance, Journal of Finance 43,
the costs of a bank-centered financial system: 567–592.
Evidence from the changing main bank rela- Wolfram, Catherine, 1995, Increases in
tions in Japan, manuscript, Harvard University. executive pay following privatization, manu-
Weisbach, Michael, 1988, Outside directors and script, MIT.
CEO turnover, Journal of Financial Wruck, Karen, 1989, Equity ownership concen-
Economics 20, 431–460. tration and firm value, Journal of Financial
Weiss, Lawrence, 1990, Bankruptcy resolution: Economics 23, 3–28.
Direct costs and violation of priority of Yafeh, Yishay, and Oved Yosha, 1996, Large
claims, Journal of Financial Economics 27, shareholders and banks: Who monitors and
285–314. how? manuscript, Hebrew University,
White, Michelle, 1993, The costs of corporate Jerusalem, Israel.
bankruptcy: a U.S.-European comparison, Yermack, David, 1997, Good timing: CEO stock
manuscript, University of Michigan. option awards and company news announce-
van Wijnbergen, Sweder, and Anton Mancini, ments, Journal of Finance 52, 449–476.
1995, Voucher privatization, corporate control, Zingales, Luigi, 1994, The value of the voting
and the cost of capital: An analysis of the right: A study of the Milan stock exchange
Czech privatization program, manuscript, LSE. experience, The Review of Financial Studies
Williamson, Oliver, 1964, The Economics of 7, 125–148.
Discretionary Behavior: Managerial Zingales, Luigi, 1995, What determines the
Objectives in a Theory of the Firm (Prentice value of corporate votes? Quarterly Journal
Hall, Englewood Cliffs, N.J.). of Economics 110, 1075–1110.
Chapter 3
ABSTRACT
Recent research has documented large differences among countries in ownership concentration
in publicly traded firms, in the breadth and depth of capital markets, in dividend policies, and
in the access of firms to external finance. A common element to the explanations of these
differences is how well investors, both shareholders and creditors, are protected by law from
expropriation by the managers and controlling shareholders of firms. We describe the differences
in laws and the effectiveness of their enforcement across countries, discuss the possible origins
of these differences, summarize their consequences, and assess potential strategies of corporate
governance reform. We argue that the legal approach is a more fruitful way to understand
corporate governance and its reform than the conventional distinction between bank-centered
and market-centered financial systems.
pricing, asset stripping, and investor may use these resources for their own
dilution, though often legal, have largely benefit. Jensen and Meckling view financial
the same effect as theft. In still other claims as contracts that give outside
instances, expropriation takes the form of investors, such as shareholders and credi-
diversion of corporate opportunities from tors, claims to the cash flows. In their
the firm, installing possibly unqualified model, the limitation on expropriation is
family members in managerial positions, the residual equity ownership by entrepre-
or overpaying executives. In general, neurs that enhances their interest in
expropriation is related to the agency dividends relative to perquisites.
problem described by Jensen and Research by Grossman, Hart, and
Meckling (1976), who focus on the Moore, summarized in Hart (1995), makes
consumption of “perquisites” by man- a further key advance by focusing squarely
agers and other types of empire building. on investor power vis a vis the insiders, and
It means that the insiders use the profits distinguishing between the contractual and
of the firm to benefit themselves rather residual control rights that investors have.
than return the money to the outside Economists have used this idea to model
investors. financial instruments not in terms of their
If extensive expropriation undermines cash flows, but in terms of the rights they
the functioning of a financial system, how allocate to their holders. In this frame-
can it be limited? The legal approach to work, investors get cash only because they
corporate governance holds that the have power. This can be the power to
key mechanism is the protection of change directors, to force dividend
outside investors – whether shareholders or payments, to stop a project or a scheme
creditors – through the legal system, meaning that benefits the insiders at the expense of
both laws and their enforcement. Although outside investors, to sue directors and get
reputations and bubbles can help raise compensation, or to liquidate the firm and
funds, variations in law and its enforcement receive the proceeds. Unlike in the
are central to understanding why firms Modigliani-Miller world, changing the
raise more funds in some countries than in capital structure of the firm changes
others. To a large extent, potential share- the allocation of power between the insiders
holders and creditors finance firms because and the outside investors, and thus
their rights are protected by the law. These almost surely changes the firm’s investment
outside investors are more vulnerable to policy.
expropriation, and more dependent on the In both the contractual framework of
law, than either the employees or the Jensen and Meckling and the residual con-
suppliers, who remain continually useful to trol rights framework of Grossman, Hart,
the firm and are thus at a lesser risk of and Moore, the rights of the investors are
being mistreated. protected and sometimes even specified by
The legal approach to corporate gover- the legal system. For example, contract law
nance is a natural continuation of the field deals with privately negotiated arrange-
as it has developed over the last 40 years. ments, whereas company, bankruptcy, and
Modigliani and Miller (1958) think of securities laws specifically describe some
firms as collections of investment projects of the rights of corporate insiders and out-
and the cash flows these projects create, side investors. These laws, and the quality
and hence naturally interpret securities of their enforcement by the regulators and
such as debt and equity as claims to these courts, are essential elements of corporate
cash flows. They do not explain why the governance and finance (La Porta et al.,
managers would return the cash flows to 1997, 1998). When investor rights such as
investors. Jensen and Meckling (1976) the voting rights of the shareholders and
point out that the return of the cash flows the reorganization and liquidation rights
from projects to investors cannot be taken of the creditors are extensive and well
for granted, and that the insiders of firms enforced by regulators or courts, investors
INVESTOR PROTECTION AND CORPORATE GOVERNANCE 93
are willing to finance firms. In contrast, law also shapes the opportunities for
when the legal system does not protect out- external finance.
side investors, corporate governance and The legal approach to corporate gover-
external finance do not work well. nance has emerged as a fruitful way to
Jensen and Meckling (1976) recognize think about a number of questions in
the role of the legal system when they finance. In Section 2, we discuss the differ-
write: ences in legal investor protection among
countries and the possible judicial,
This view of the firm points up the political, and historical origins of these
important role which the legal system differences. In Section 3, we summarize the
and the law play in social organizations, research on the economic consequences of
especially, the organization of economic investor protection. In Section 4, we compare
activity. Statutory law sets bounds on the the legal approach to corporate governance
kinds of contracts into which individuals to the more standard focus on the relative
and organizations may enter without importance of banks and stock markets
risking criminal prosecution. The police as ways to explain country differences. In
powers of the state are available and Section 5, we discuss both the difficulties
used to enforce performance of contracts and the opportunities for corporate gover-
or to enforce the collection of damages nance reform. Section 6 concludes.
for non-performance. The courts adjudi-
cate contracts between contracting
parties and establish precedents which 2. INVESTOR PROTECTION
form the body of common law. All of
these government activities affect both When investors finance firms, they typically
the kinds of contracts executed and the obtain certain rights or powers that are
extent to which contracting is relied generally protected through the enforce-
upon (p. 311). ment of regulations and laws. Some of
these rights include disclosure and
One way to think about legal protection accounting rules, which provide investors
of outside investors is that it makes the with the information they need to exercise
expropriation technology less efficient. At other rights. Protected shareholder rights
the extreme of no investor protection, the include those to receive dividends on
insiders can steal a firm’s profits perfectly pro-rata terms, to vote for directors, to
efficiently. Without a strong reputation, no participate in shareholders’ meetings, to
rational outsider would finance such a firm. subscribe to new issues of securities on the
As investor protection improves, the insiders same terms as the insiders, to sue directors
must engage in more distorted and waste- or the majority for suspected expropriation,
ful diversion practices such as setting up to call extraordinary shareholders’ meetings,
intermediary companies into which they etc. Laws protecting creditors largely deal
channel profits. Yet these mechanisms are with bankruptcy and reorganization
still efficient enough for the insiders to procedures, and include measures that
choose to divert extensively. When investor enable creditors to repossess collateral, to
protection is very good, the most the insiders protect their seniority, and to make it
can do is overpay themselves, put relatives harder for firms to seek court protection in
in management, and undertake some reorganization.
wasteful projects. After a point, it may be In different jurisdictions, rules protect-
better just to pay dividends. As the diver- ing investors come from different sources,
sion technology becomes less efficient, the including company, security, bankruptcy,
insiders expropriate less, and their private takeover, and competition laws, but also
benefits of control diminish. Firms then from stock exchange regulations and
obtain outside finance on better terms. By accounting standards. Enforcement of laws
shaping the expropriation technology, the is as crucial as their contents. In most
94 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Table 3.1 presents the percentage of distinguish between two broad kinds of
countries in each legal family that give answers: the “judicial” explanations that
investors the rights discussed by La Porta, account for the differences in the legal
Lopez-de-Silanes, Shleifer, and Vishny, as philosophies using the organization of the
well as the mean for that family antidirector legal system, and the “political” explana-
and creditor rights scores. How well legal tions that account for these differences
rules protect outside investors varies using political history.
systematically across legal origins. The “judicial” explanation of why com-
Common law countries have the strongest mon law protects investors better than civil
protection of outside investors – both law has been most recently articulated by
shareholders and creditors – whereas Coffee (2000) and Johnson et al. (2000b).
French civil law countries have the weakest Legal rules in the common law system are
protection. German civil law and usually made by judges, based on prece-
Scandinavian countries fall in between, dents and inspired by general principles
although comparatively speaking they have such as fiduciary duty or fairness. Judges
stronger protection of creditors, especially are expected to rule on new situations by
secured creditors. In general, differences applying these general principles even when
among legal origins are best described by specific conduct has not yet been described
the proposition that some countries protect or prohibited in the statutes. In the area of
all outside investors better than others, and investor expropriation, also known as self-
not by the proposition that some countries dealing, the judges apply what Coffee calls
protect shareholders while other countries a “smell test,” and try to sniff out whether
protect creditors. even unprecedented conduct by the insiders
Table 3.1 also points to significant is unfair to outside investors.The expansion
differences among countries in the quality of legal precedents to additional violations
of law enforcement as measured by the of fiduciary duty, and the fear of such
efficiency of the judiciary, (lack of) corrup- expansion, limit the expropriation by the
tion, and the quality of accounting stan- insiders in common law countries. In con-
dards. Unlike legal rules, which do not trast, laws in civil law systems are made by
appear to depend on the level of economic legislatures, and judges are not supposed to
development, the quality of enforcement is go beyond the statutes and apply “smell
higher in richer countries. In particular, the tests” or fairness opinions. As a conse-
generally richer Scandinavian and German quence, a corporate insider who finds a way
legal origin countries receive the best not explicitly forbidden by the statutes to
scores on the efficiency of the judicial sys- expropriate outside investors can proceed
tem. The French legal origin countries have without fear of an adverse judicial ruling.
the worst quality of law enforcement of the Moreover, in civil law countries, courts do
four legal traditions, even controlling for not intervene in self-dealing transactions as
per capita income. long as these have a plausible business
Because legal origins are highly corre- purpose.The vague fiduciary duty principles
lated with the content of the law, and of the common law are more protective of
because legal families originated before investors than the bright line rules of the
financial markets had developed, it is civil law, which can often be circumvented
unlikely that laws were written primarily in by sufficiently imaginative insiders.
response to market pressures. Rather, the The judicial perspective on the differ-
legal families appear to shape the legal ences is fascinating and possibly correct,
rules, which in turn influence financial but it is incomplete. It requires a further
markets. But what is special about legal assumption that the judges have an inclina-
families? Why, in particular, is common law tion to protect the outside investors rather
more protective of investors than civil law? than the insiders. In principle, it is easy to
These questions do not have accepted imagine that the judges would use their dis-
answers. However, it may be useful here to cretion in common law countries to narrow
Table 3.1 Legal origin and investors rights
This table presents data on measures of investor protection for 49 countries classified by their legal origin.The source of the data is La Porta et al. (1998).
Panel A shows the measures of shareholder protection across legal origins. The “antidirector rights index” is a summary measure of shareholder protec-
tion.This index ranges from zero to six and is formed by adding one when: the country allows shareholders to mail their proxy vote to the firm; shareholders
are not required to deposit their shares prior to the General Shareholders’. Meeting; cumulative voting or proportional representation of minorities in the
board of directors is allowed; an oppressed minorities mechanism is in place; the minimum percentage of share capital that entitles a shareholder to call
for an Extraordinary Shareholders’. Meeting is less than or equal to 10 percent (the sample median); shareholders have preemptive rights that can only
be waved by a shareholders’, vote. The rest of the rows in Panel A show the percentage of countries within each legal origin for which each component of
the “antidirector rights index” is provided by the law. Panel B shows the measures of creditor protection across legal origins. The “creditor rights index”
is a summary measure of creditor protection.This index ranges from zero to four and is formed by adding one when: the country imposes restrictions, such
as creditors’ consent or minimum dividends to file for reorganization; secured creditors are able to gain possession of their security once the reorganiza-
tion petition has been approved (no automatic stay); secured creditors are ranked first in the distribution of the proceeds that result from the disposition
of the assets of a bankrupt firm; the debtor does not retain the administration of its property pending the resolution of the reorganization. The rest of the
rows in Panel B show the percentage of countries within each legal origin for which each component of the “creditor rights index” is provided by the law.
Panel C shows measures of legal enforcement. “Efficiency of the judicial system” is an index ranging from zero to ten representing the average of investors’
assessments of conditions of the judicial system in each country between 1980–1983 (lower scores represent lower efficiency levels). “Corruption” is an
index ranging from zero to ten representing the average of investors’ assessments of corruption in government in each country between 1982 and 1995
(lower scores represent higher corruption). “Accounting standards” is an index created by examining and rating companies’ 1990 annual reports on their
inclusion or omission of 90 items falling in the categories of general information, income statements, balance sheets, funds flow statement, accounting stan-
dards, stock data, and special items.
Legal origin
96 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
the interpretation of fiduciary duty and to Recent research supports the proposition
sanction expropriation rather than prohibit that civil law is associated with greater
it. Common law judges could also in princi- government intervention in economic
ple use their discretion to serve political activity and weaker protection of private
interests, especially when the outside property than common law. La Porta et al.
investors obstruct the government’s goals. (1999a) examine the determinants of
To explain investor protection, it is not government performance in a large number
enough to focus on judicial power; a political of countries. To measure government
and historical analysis of judicial objectives interventionism, they consider proxies for
is required. From this perspective, impor- the amount and quality of regulation, the
tant political and historical differences prevalence of corruption and of red tape,
between mother countries shape their laws. and bureaucratic delays. As a general rule,
This is not to say that laws never change they find that civil law countries, particu-
(in Section 5 we focus specifically on legal larly French civil law countries, are more
reform) but rather to suggest that history interventionist than common law countries.
has persistent effects. The inferior protection of the rights of out-
La Porta et al. (1999a) argue that an side investors in civil law countries may be
important historical factor shaping laws is one manifestation of this general phenome-
that the state has a relatively greater role non. This evidence provides some support
in regulating business in civil law countries for interpreting the differences in legal
than in common law ones. One element of families based on political history.3
this view, suggested by Finer (1997) and
other historians, points to the differences in
the relative power of the king and the prop- 3. CONSEQUENCES OF INVESTOR
erty owners across European states. In PROTECTION
England from the seventeenth century on,
the crown partially lost control of the Three broad areas in which investor
courts, which came under the influence of protection has been shown to matter are
the parliament and the property owners the ownership of firms, the development of
who dominated it. As a consequence, financial markets, and the allocation of
common law evolved to protect private real resources.
property against the crown. Over time,
courts extended this protection of property
owners to investors. In France and 3.1. Patterns of ownership and
Germany, by contrast, parliamentary power control
was weaker. Commercial Codes were
adopted only in the nineteenth century by The focus on expropriation of investors and
the two great state builders, Napoleon and its prevention has a number of implications
Bismarck, to enable the state to better for the ownership structures of firms.
regulate economic activity. Over time, the Consider first the concentration of control
state maintained political control over rights in firms (as opposed to the dividend
firms and resisted the surrender of that or cash flow rights). At the most basic
power to financiers.2 Perhaps as impor- level, when investor rights are poorly
tantly, the state in civil law countries did protected and expropriation is feasible on a
not surrender its power over economic substantial scale, control acquires enormous
decisions to courts, and hence maintained value because it gives the insiders the
the statutory approach to commercial opportunity to expropriate efficiently.
laws. As we noted above, however, fairness When the insiders actually do expropriate,
assessments of self-dealing transactions, the so called private benefits of control
for which judicial power and discretion become a substantial share of the firm’s
are essential, may be central to limiting value. This observation raises a question:
expropriation. will control in such an environment be
INVESTOR PROTECTION AND CORPORATE GOVERNANCE 99
concentrated in the hands of an entrepreneur outsiders and still retain control by holding
or dispersed among many investors? on to the shares with superior voting rights.
The research in this area originates in He can also use a pyramidal structure, in
the work of Grossman and Hart (1988) which a holding company he controls sells
and Harris and Raviv (1988), who examine shares in a subsidiary that it itself controls.
the optimal allocation of voting and cash Wolfenzon (1999) shows that an entrepre-
flow rights in a firm. The specific question neur can then control the subsidiary
of how control is likely to be allocated has without owning a substantial fraction of its
not received a clear answer. For several cash flow rights, and that such schemes are
reasons, entrepreneurs may wish to keep more attractive when the protection of
control of their firms when investor protec- outside investors is weaker. An entrepreneur
tion is poor. La Porta et al. (1999) note can also keep control through cross-
that if expropriation of investors requires shareholdings among firms, which make it
secrecy, sharing control may restrain the harder for outsiders to gain control of one
entrepreneur beyond his wishes. Zingales group firm without buying all of them.
(1995), La Porta et al. (1999), and What about the distribution of cash flow
Bebchuk (1999) argue that if entrepreneurs rights between investors as opposed to
disperse control between many investors, control? If an entrepreneur retains control
they give up the “private benefits” premium of a firm, how can he raise any external
in a takeover. In Bebchuk’s (1999) model, funds from outside investors – for financing
diffuse control structures are unstable or for diversification – who expect to be
when investors can concentrate control expropriated? Jensen and Meckling (1976)
without fully paying for it. Finally, an entre- would suggest that cash flow ownership by
preneur or his family may need to retain an entrepreneur reduces incentives for
control of the firm because the family’s expropriation and raises incentives to pay
reputation is needed to raise external funds out dividends. La Porta et al. (1999b)
when the legal protection of outside show that this need for higher cash flow
investors is poor. For all these reasons, ownership as a commitment to limit expro-
firms in countries with poor investor priation is higher in countries with inferior
protection may need concentrated control. shareholder protection.
Bennedsen and Wolfenzon (2000) make The available evidence on corporate
a countervailing argument. When investor ownership patterns around the world
protection is poor, dissipating control supports the importance of investor protec-
among several large investors – none of tion. This evidence was obtained for a
whom can control the decisions of the firm number of individual countries, including
without agreeing with the others – may Germany (Edwards and Fischer, 1994;
serve as a commitment to limit expropria- Gorton and Schmid, 2000), Italy (Barca,
tion. When there is no single controlling 1995), and seven Organization for
shareholder, and the agreement of several Economic Cooperation and Development
large investors (the board) is needed for countries (European Corporate Governance
major corporate actions, these investors Network, 1997). La Porta et al. (1998)
might together hold enough cash flow describe ownership concentration in their
rights to choose to limit expropriation of sample of 49 countries, while La Porta
the remaining shareholders and pay the et al. (1999) examine patterns of control in
profits out as efficient dividends. When the largest firms from each of 27 wealthy
the dissipation of control reduces ineffi- economies. The data show that countries
cient expropriation, it may emerge as an with poor investor protection typically
optimal policy for a wealth-maximizing exhibit more concentrated control of firms
entrepreneur. than do countries with good investor pro-
An entrepreneur has a number of ways tection. In the former, even the largest
to retain control of a firm. He can firms are usually controlled either by the
sell shares with limited voting rights to the state or by the families that founded or
100 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
acquired these firms. In the latter countries, Creditor rights encourage the development
the Berle and Means corporation – with of lending, and the exact structure of these
dispersed shareholders and professional rights may alternatively favor bank lending
managers in control – is more common.4 or market lending. Shareholder rights
Claessens et al. (2000) examine a encourage the development of equity
sample of nearly 3,000 firms from 9 East markets, as measured by the valuation of
Asian economies. Except in Japan, which firms, the number of listed firms (market
has fairly good shareholder protection, they breadth), and the rate at which firms go
find a predominance of family control and public. For both shareholders and credi-
family management of the corporations in tors, protection includes not only the rights
their sample, with some state control as written into the laws and regulations but
well.They also present remarkable evidence also the effectiveness of their enforcement.
of “crony capitalism” in Asia: outside Consistent with these predictions, La Porta
Japan, the top 10 families in each of the et al. (1997) show that countries that
remaining 8 countries studied control protect shareholders have more valuable
between 18 and 58 percent of the aggre- stock markets, larger numbers of listed
gate value of listed equities. securities per capita, and a higher rate of
In sum, the evidence has proved to be IPO (initial public offering) activity than
broadly consistent with the proposition do the unprotective countries. Countries
that the legal environment shapes the value that protect creditors better have larger
of the private benefits of control and credit markets.
thereby determines the equilibrium owner- Several recent studies have also estab-
ship structures. Perhaps the main implica- lished a link between investor protection,
tions of this evidence for the study of insider ownership of cash flows, and corpo-
corporate governance are the relative rate valuation.5 Gorton and Schmid
irrelevance of the Berle and Means corpo- (2000) show that higher ownership by the
ration in most countries in the world and large shareholders is associated with
the centrality of family control. Indeed, higher valuation of corporate assets in
La Porta et al. (1999) and Claessens et al. Germany. Claessens et al. (1999) use a
(2000) find that family-controlled firms sample of East Asian firms to show that
are typically managed by family members greater insider cash flow ownership is asso-
so that the managers appear to be kept on ciated with higher valuation of corporate
a tighter leash than what Berle and Means assets, whereas greater insider control of
describe. As Shleifer and Vishny (1997) voting rights is associated with lower valu-
have argued, in large corporations of most ation of corporate assets. Using a sample
countries, the fundamental agency problem of firms from 27 wealthy economies,
is not the Berle and Means conflict La Porta et al. (1999b) find that firms in
between outside investors and managers, countries with better shareholder protection
but rather that between outside investors have higher Tobin’s Q than do firms in
and controlling shareholders who have countries with inferior protection.They also
nearly full control over the managers. find that higher insider cash flow ownership
is (weakly) associated with higher corporate
3.2. Financial markets valuation, and that this effect is greater
in countries with inferior shareholder
The most basic prediction of the legal protection. These results support the roles
approach is that investor protection of investor protection and cash flow
encourages the development of financial ownership by the insiders in limiting
markets. When investors are protected expropriation.
from expropriation, they pay more for Johnson et al. (2000a) draw an ingenious
securities, making it more attractive for connection between investor protection and
entrepreneurs to issue these securities.This financial crises. In countries with poor
applies to both creditors and shareholders. protection, the insiders might treat outside
INVESTOR PROTECTION AND CORPORATE GOVERNANCE 101
investors well as long as future prospects papers show that an exogenous component
are bright and they are interested in con- of financial market development, obtained
tinued external financing. When future by using legal origin as an instrument,
prospects deteriorate, however, the insiders predicts economic growth.
step up expropriation, and the outside More recent research distinguishes the
investors, whether shareholders or credi- three channels through which finance can
tors, are unable to do anything about it. contribute to growth: saving, factor accu-
This escalation of expropriation renders mulation, and efficiency improvements.
security price declines especially deep in Beck et al. (2000) find that banking sector
countries with poor investor protection. To development exerts a large impact on total
test this hypothesis, Johnson et al. (2000a) factor productivity growth and a less
examine the depreciation of currencies and obvious impact on private savings and capital
the decline of the stock markets in 25 accumulation. Moreover, this influence
countries during the Asian crisis of continues to hold when an exogenous
1997–1998. They find that governance component of banking sector development,
variables, such as investor protection obtained by using legal origin as an instru-
indices and the quality of law enforcement, ment, is taken as a predictor. Wurgler
are powerful predictors of the extent of (2000) finds that financially developed
market declines during the crisis. These countries allocate investment across indus-
variables explain the cross-section of tries more in line with the variation in
declines better than do the macroeconomic growth opportunities than do financially
variables that have been the focus of the underdeveloped countries. Morck et al.
initial policy debate. (2000) find that stock markets in more
developed countries incorporate firm-
specific information better, helping to
3.3. Real consequences allocate investment more effectively. This
Through its effect on financial markets, research suggests that financial development
investor protection influences the real improves resource allocation. Through this
economy. According to Beck et al. (2000), channel, investor protection may benefit
financial development can accelerate eco- the growth of productivity and output.
nomic growth in three ways. First, it can
enhance savings. Second, it can channel
these savings into real investment and 4. BANK AND MARKET CENTERED
thereby foster capital accumulation. Third, GOVERNANCE
to the extent that the financiers exercise
some control over the investment decisions Traditional comparisons of corporate
of the entrepreneurs, financial development governance systems focus on the institu-
allows capital to flow toward the more pro- tions financing firms rather than on the
ductive uses, and thus improves the effi- legal protection of investors. Bank-centered
ciency of resource allocation. All three corporate governance systems, such as
channels can in principle have large effects those of Germany and Japan, are
on economic growth. compared to market-centered systems,
A large body of research links financial such as those of the United States and the
development to economic growth. King and United Kingdom (see, e.g., Allen and Gale,
Levine (1993) initiate the modern incarna- 2000). Relatedly, relationship-based
tion of this literature by showing that coun- corporate governance, in which a main
tries with larger initial capital markets bank provides a significant share of
grow faster in the future. Demirguc-Kunt finance and governance to each firm, is
and Maksimovic (1998), Levine and contrasted with market-based governance,
Zervos (1998), Rajan and Zingales in which finance is provided by large num-
(1998), and Carlin and Mayer (1999) bers of investors and in which takeovers
extend these findings. Several of these play a key governance role.
102 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
have strong legal protection of creditors, (1931) and others have argued that
particularly of secured creditors. Without bubbles play an important and positive role
such rights German banks would have in stimulating investment.
much less power. The United Kingdom also To summarize, bank-versus market
has a large banking and public debt sector, centeredness is not an especially useful way
again because creditors have extensive to distinguish financial systems. Investor
rights, as well as a large equity market. rights work better to explain differences
Italy and Belgium, by contrast, have among countries, and in fact are often
developed neither debt nor equity markets necessary for financial intermediaries to
because no outside investors are protected develop. Moreover, even if some countries
there.8 The point here is simple: all outside go through monopoly banking in their
investors, be they large or small, creditors development process, this stage has little to
or shareholders, need rights to get their recommend it other than as a stepping
money back. Investor rights are a stone toward more developed markets. And
more primitive determinant of financial to get to more developed markets, it is
development than is the size of particular essential to improve the rights of outside
institutions. investors.
Despite the difficulty of classifying
financial systems into bank- and market
centered, economists at least since 5. POSSIBILITIES FOR REFORM
Gerschenkron (1962) have engaged in a
lively debate as to which one is superior, In the last decade, the reform of corporate
focusing on the hypothesis that bank- governance has attracted interest in
centered systems are particularly suitable for Western and Eastern Europe, Latin
developing economies.This is not a place to America, and Asia. The discussions have
review this debate. Rather, our concern is intensified since the Asian financial crisis,
that the interest in monopoly bank lending and took on the flavor of reforming “the
distracts attention from the important role global financial architecture”. To discuss
that stock markets play in external finance. any reform, it is important to start with its
Equity financing is essential for the expan- goals. Our analysis suggests that one objec-
sion of new firms whose main asset are the tive of corporate governance reform is to
growth opportunities. In principle, firms protect the rights of outside investors,
could utilize private equity financing, but it including both shareholders and creditors.
has many of the same problems of exces- As the evidence described in Section 3
sive investor power suppressing entrepre- shows, the benefits of such reform would be
neurial initiative as does monopoly banking to expand financial markets, to facilitate
(see, e.g., Myers, 1977; Burkart et al., external financing of new firms, to move
1997). Public equity financing, for which a away from concentrated ownership, to
developed stock market is needed, has improve the efficiency of investment alloca-
other advantages over private equity tion, and to facilitate private restructuring
financing. It allows the buyers of equity to of financial claims in a crisis.
diversify. It offers the initial equity holders, So what, if anything, can be done to
such as venture capitalists, an attractive achieve these goals, and what are the
exit option through the public equity mar- obstacles? To organize this discussion, we
kets. Last but not least, it allows firms to follow Coffee (1999) and Gilson (2000) in
time their equity issues to take advantage drawing a distinction between legal and
of favorable investor sentiment toward functional convergence. Legal convergence
their industry, or toward the market as a refers to the changes in rules and enforce-
whole. Such sentiment may play a benefi- ment mechanisms toward some successful
cial role when shareholders are skeptical standard. To converge to effective investor
about the likelihood of getting back a protection in this way, most countries
return on their money. Indeed, Keynes require extensive legal, regulatory, and
104 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
have been wasted, in part because of a lack intermediaries. Thus the accounting
of appreciation of the need to protect profession, once it recognized the increased
investors. Recent research points to some demand for its services, became an
crucial principles of investor protection independent private force in assuring the
that reforms need to focus on. compliance with disclosure regulations. As
The first such principle is that legal rules a consequence, a small Commission was
do matter. It is not just the stance of the able to regulate a huge market with
law or the political sentiment of the day relatively few resources. The principle of
that shapes financial markets. One illustra- recruiting private intermediaries into the
tion of this principle, described by Johnson enforcement of securities regulations has
(1999), is the Neuer Markt in Germany, a since been followed by a number of
segment of the Frankfurt Stock Exchange countries, including Germany and Poland.
created especially for listing new firms. A third and related principle of
Because the Neuer Markt operates in successful reform, stressed by Glaeser et al.
Germany, the corporate law, the securities (2001), is that government regulation of
law, and other basic laws and regulations financial markets may be useful when court
that are applied to the companies listing enforcement of private contracts or laws
there are the general German rules. The cannot be relied upon. An example of how
politics are German as well. As part of a regulation can work when judicial enforce-
private contract with firms wishing to list ment is limited comes from the securities
on the Neuer Markt, the Deutsche Bourse – law reform in Poland and the Czech
which operates the Frankfurt Stock Republic, two transition economies whose
Exchange – has mandated that these firms judiciaries in the early 1990s were gener-
must comply with international accounting ally viewed as ineffective. At that time the
standards and agree to greater disclosure Polish government introduced a tough
than that required of already listed firms. securities law focused on shareholder
The new listing venue, with its greater protection. Like the U.S. securities law, the
restrictions on the entrepreneurs, has Polish regulations focused on significant
sharply accelerated the pace of initial pub- disclosure by new issuers and already listed
lic offerings in Germany. At the same time, firms, as well as on licensing and close
the captains of German industry have administrative oversight of intermediaries.
accepted it because their firms were not The law also provided for a creation of a
directly affected.This points to one possible powerful SEC with significant enforcement
strategy of overcoming political opposition powers that did not require reliance on
to reform. courts. This reform was followed by a
A second principle is that good legal remarkable development of the Polish
rules are the ones that a country can stock market, with both new and already
enforce. The strategy for reform is not to listed companies raising equity in the
create an ideal set of rules and then see market.
how well they can be enforced, but rather to By contrast, the Czech government chose
enact the rules that can be enforced within neither to introduce tough securities laws
the existing structure. One example of the nor to create a powerful market regulator
success of such a policy is the U.S. securi- at the time of privatization. Perhaps as a
ties legislation of 1933–1934, described by consequence, the Czech markets have been
Landis (1938) and McCraw (1984). This plagued by massive expropriation of minority
legislation placed much of the responsibil- shareholders – the so-called “tunneling” of
ity for accurate corporate accounting and assets from both firms and mutual funds.
disclosure on intermediaries, and focused In contrast to the Polish market, the Czech
the regulatory oversight by the Securities market stagnated, with hundreds of compa-
and Exchange Commission (SEC) on these nies getting delisted and virtually no public
relatively few intermediaries. The SEC equity financing by firms (see Coffee,
also emphasized self-regulation by the 1999; Pistor, 1999; Glaeser et al., 2001).
106 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Financial markets need some protection of the law remains a crucial channel through which
outside investors, whether by courts, gov- politics affects corporate governance.
ernment agencies, or market participants 4 The evidence also reveals that control is
themselves. Improving such protection is a valued, and specifically that voting premiums
increase as shareholder protection deteriorates
difficult task. In part, the nature of investor
(see, for example, Modigliani and Perotti, 1998;
protection, and more generally of regula- Nenova, 1999; Zingales, 1994).
tion of financial markets, is deeply rooted 5 In addition, La Porta et al. (2000) show
in the legal structure of each country and that better minority shareholder protection is
in the origin of its laws. Marginal reform associated with higher dividend pay-outs in a
may not successfully achieve the reformer’s cross-section of firms from around the world.
goals. In part, the existing corporate 6 Jensen (1989) expresses some early skepti-
governance arrangements benefit both the cism about the Japanese financial system.
politicians and the entrenched economic 7 Hellwig (1999) doubts that banks are so
interests, including the families that man- powerful, even in the case of Germany.
8 Levine et al. (2000) find that the La Porta
age the largest firms in most countries in
et al. (1998) measure of creditor rights is
the world. Corporate governance reform correlated with measures of financial intermedi-
must circumvent the opposition by these aries development across countries, while their
interests. Despite these difficulties, reform measure of shareholder rights is correlated with
of investor protection is politically feasible stock market development.
in some circumstances, and can bring
significant benefits. It can take the form of
opting into more protective legal regimes REFERENCES
or introducing more radical changes in the
legal structure. The integration of world Allen, F., Gale, D., 2000. Comparing Financial
capital markets makes such reforms Systems. MIT Press, Cambridge, MA.
more likely today than they have been in Aoki, M., Patrick, H., 1993. The Japanese Main
decades. Bank System: Its Relevance for Developing
and Transforming Economies. Oxford
University Press, New York.
NOTES Barca, F., 1995. On corporate governance in
Italy: issues, facts, and agency. Unpublished
* We are grateful to Nicholas Barberis, working paper. Bank of Italy, Rome.
Simeon Djankov, Oliver Hart, Michael Jensen, Bebchuk, L., 1999.The rent protection theory of
Simon Johnson, Ross Levine, and Daniel corporate ownership and control.
Wolfenzon for helpful comments, and also to the Unpublished working paper. Harvard Law
NSF for financial support of this research. School, Cambridge, MA.
† Corresponding author. Tel.: 617–495–5046;
Beck, T., Levine, R., Loayza, N., 2000. Finance
fax: 617–496–1708. E-mail address: ashleifer@
harvard.edu (A. Shleifer). and the sources of growth. Journal of
1 Socialist countries had a legal tradition Financial Economics 58, 261–300.
based on Soviet law, but because the laws of Bennedsen, M., Wolfenzon, D., 2000. The bal-
these countries are changing rapidly during the ance of power in closely held corporations.
transition out of socialism, La Porta et al. Journal of Financial Economics 58, 113–139.
(1998) do not consider them. Berglof, E., von Thadden, L., 1999. The changing
2 According to Cameron (1961), France had corporate governance paradigm: implications
a lively stock market in the nineteenth century. for transition and developing countries.
Nearly all firms listed on it, however, benefitted Unpublished working paper. Stockholm
from government concessions, investment, own-
Institute of Transition Economics,
ership, subsidies, protection, and often outright
guarantees to investors. Stockholm, Sweden.
3 Berglof and von Thadden (1999) and Rajan Burkart, M., Gromb, D., Panunzi, F., 1997.
and Zingales (1999) argue that political factors Large Shareholders, monitoring, and fiduci-
affect corporate governance through channels ary duty. Quarterly Journal of Economics
other than the law itself. This may be true, but 112, 693–728.
108 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Cameron, R., 1961. France and the Economic Unpublished working paper. Stanford
Development of Europe. Princeton University University, Stanford, CA.
Press, Princeton, NJ. Glaeser, E., Johnson, S., Shleifer, A., 2001.
Carlin, W., Mayer, C., 1999. Finance, investment Coase versus the Coasians. Quarterly Journal
and growth. Unpublished working paper. of Economics, forthcoming.
University College, London. Gorton, G., Schmid, F., 2000. Universal banking
Claessens, S., Djankov, S., Lang, L., 2000. The and the performance of German firms.
separation of ownership and control in East Journal of Financial Economics 58, 29–80.
Asian corporations. Journal of Financial Grossman, S., Hart, O., 1988. One-share-one-
Economics 58, 81–112. vote and the market for corporate control.
Claessens, S., Djankov, S., Fan, J., Lang, L., Journal of Financial Economics 20,
1999. Expropriation of minority shareholders 175–202.
in East Asia. Unpublished working paper. The Harris, M., Raviv, A., 1988. Corporate gover-
World Bank, Washington, DC. nance: voting rights and majority rules.
Coase, R., 1961. The problem of social cost. Journal of Financial Economics 20,
Journal of Law and Economics 3, 1–44. 203–236.
Coffee, J., 1999. The future as history: the Hart, O., 1995. Firms, Contracts, and Financial
prospects for global convergence in corporate Structure. Oxford University Press, London.
governance and its implications. Hellwig, M., 1999. On the economics and
Northwestern Law Review 93, 631–707. politics of corporate finance and corporate
Coffee, J., 2000. Privatization and corporate control. Unpublished working paper.
governance: the lessons from securities mar- University of Mannheim, Mannheim.
ket failure. Working paper no. 158. Columbia Henry, P., 2000. Do stock market liberalizations
Law School, New York. cause investment booms? Journal of
David, R., Brierley, J., 1985. Major Legal Financial Economics 58, 301–334.
Systems in the World Today. Stevens and Hirschman, A., 1963. Journeys Toward
Sons, London. Progress. The Twentieth Century Fund, New
Demirguc-Kunt, A., Maksimovic, V., 1998. Law, York.
finance, and firm growth. Journal of Finance Hoshi, T., Kashyap, A., Scharfstein, D., 1991.
53, 2107–2139. Corporate structure, liquidity, and invest-
Easterbrook, F., Fischel, D., 1991. The ment: evidence from Japanese industrial
Economic Structure of Corporate Law. groups. Quarterly Journal of Economics 106,
Harvard University Press, Cambridge, MA. 33–60.
Edwards, J., Fischer, K., 1994. Banks, Finance Jensen, M., 1989. Eclipse of the modern
and Investment in West Germany Since 1970. corporation. Harvard Business Review 67,
Cambridge University Press, Cambridge, UK. 61–74.
European Corporate Governance Network Jensen, M., Meckling, W., 1976. Theory of the
(ECGN). 1997. The Separation of Ownership firm: managerial behavior, agency costs, and
and Control: A Survey of 7 European ownership structure. Journal of Financial
Countries Preliminary Report to the Economics 3, 305–360.
European Commission. Volumes 1–4. Johnson, S., 1999. Does investor protection
Brussels, European Corporate Governance matter? Evidence from Germany’s Neuer
Network. Markt. Unpublished working paper. MIT
Finer, S., 1997. The History of Government, Press, Cambridge, MA.
Vol. III. Cambridge University Press, Johnson, S., Boone, P., Breach, A., Friedman, E.,
Cambridge, UK. 2000a. Corporate governance in the Asian
Gerschenkron, A., 1962. Economic financial crisis, Journal of Financial
Backwardness in Historical Perspective. Economics 58, 141–186.
Belknap Press of Harvard University Press, Johnson, S., La Porta, R., Lopez-de-Silanes, F.,
Cambridge, MA. Shleifer, A., 2000b. Tunneling. American
Gilson, R., 2000. Globalizing corporate Economic Review Papers and Proceedings
governance: convergence of form or function. 90, 22–27.
INVESTOR PROTECTION AND CORPORATE GOVERNANCE 109
Kang, J., Stulz, R., 1998. Do banking shocks Modigliani, F., Perotti, E., 1998. Security versus
affect borrowing firm performance? An bank finance: the importance of a proper
analysis of the Japanese experience. enforcement of legal rules. Unpublished
Unpublished working paper. Ohio State working paper. MIT Press, Cambridge, MA.
University, Columbus, OH. Morck, R., Nakamura, M., 1999. Banks and
Keynes, J., 1931. An economic analysis of corporate control in Japan. Journal of
unemployment. In Collected Writings,Vol. XII. Finance 54, 319–340.
Macmillan, London. Morck, R., Yeung, B., Yu, W., 2000. The
King, R., Levine, R., 1993. Finance and growth: information content of stock markets: why do
Schumpeter might be right. Quarterly Journal emerging markets have synchronous price
of Economics 108, 717–738. movements? Journal of Financial Economics
Landis, J., 1938. The Administrative Process. 58, 215–260.
Yale University Press, New Haven, CT. Myers, S., 1977. Determinants of corporate
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., borrowing. Journal of Financial Economics 5,
1999. Corporate ownership around the world. 147–175.
Journal of Finance 54, 471–517. Nenova, T., 1999. The value of a corporate vote
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., and private benefits: a cross-country analysis.
Vishny, R., 1997. Legal determinants of exter- Unpublished working paper. Harvard
nal finance. Journal of Finance 52, University, Cambridge, MA.
1131–1150. Pistor, K., 1999. Law as a determinant of equity
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., market development. Unpublished working
Vishny, R., 1998. Law and finance. Journal of paper. Harvard University, Cambridge, MA.
Political Economy 106, 1113–1155. Porter, M., 1992. Capital disadvantage:
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., America’s falling capital investment system.
Vishny, R., 1999a. The quality of government. Harvard Business Review 46, 65–72.
Journal of Law, Economics and Organization Rajan, R., Zingales, L., 1998. Financial depend-
15, 222–279. ence and growth. American Economic Review
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., 88, 559–586.
Vishny, R., 1999b. Investor protection and Rajan, R., Zingales, L., 1999. The politics of
corporate valuation. NBER Working Paper financial development. Unpublished working
7403. National Bureau of Economic paper. University of Chicago, Chicago, IL.
Research, Cambridge, MA. Ramseyer, M., Nakazato, M., 1999. Japanese
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., Law: an Economic Approach. University of
Vishny, R., 2000. Agency problems and divi- Chicago Press, Chicago, IL.
dend policies around the world. Journal of Sapienza, P., 1999. The effects of bank mergers
Finance 55, 1–33. on loan contracts. Unpublished working
Levine, R., Zervos, S., 1998. Stock markets, paper. Northwestern University, Chicago, IL.
banks and economic growth. American Shleifer, A.,Treisman, D., 2000. Without a Map:
Economic Review 88, 537–558. Political Tactics and Economic Reform in
Levine, R., Loayza, N., Beck,T., 2000. Financial Russia. MIT Press, Cambridge, MA.
intermediary development and economic Shleifer, A., Vishny, R., 1997. A survey of cor-
growth: causality and causes. Journal of porate governance. Journal of Finance 52,
Monetary Economics, 46, 31–77. 737–783.
Mayer, C., 1988. New issues in corporate Stigler, G., 1964. Public regulation of the securi-
finance. European Economic Review 32, ties market. Journal of Business 37, 117–142.
1167–1188. Stulz, R., 1999. International Portfolio Flows
McCraw, T.K., 1984. Prophets of Regulation. and Securities Markets.Working Paper 99–3,
The Belknap Press of Harvard University Fisher College of Business. The Ohio State
Press, Cambridge, MA. University, Columbus, OH.
Modigliani, F., Miller, M., 1958.The cost of cap- Weinstein, D., Yafeh, Y., 1998. On the costs of a
ital, corporation finance, and the theory of bank-centered financial system: evidence
investment. American Economic Review 48, from the main bank relations in Japan.
261–297. Journal of Finance 53, 635–672.
110 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Wolfenzon, D., 1999. A theory of pyramidal Zingales, L., 1994. The value of the voting
structures. Unpublished working paper. right: a study of the Milan stock exchange.
Harvard University, Cambridge, MA. The Review of Financial Studies 7, 125–148.
Wurgler, J., 2000. Financial markets and the Zingales, L., 1995. Inside ownership and the
allocation of capital. Journal of Financial decision to go public. Review of Economic
Economics 58, this issue. Studies 62, 425–448.
Chapter 4
ABSTRACT
We provide measures of absolute and relative equity agency costs for corporations under
different ownership and management structures. Our base case is Jensen and Meckling’s
(1976) zero agency-cost firm, where the manager is the firm’s sole shareholder. We utilize a
sample of 1,708 small corporations from the FRB/NSSBF database and find that agency costs
(i) are significantly higher when an outsider rather than an insider manages the firm; (ii) are
inversely related to the manager’s ownership share; (iii) increase with the number of nonmanager
shareholders, and (iv) to a lesser extent, are lower with greater monitoring by banks.
THE SOCIAL AND PRIVATE COSTS OF AN and Meckling agency theory, the zero
AGENT’S ACTIONS due to incomplete align- agency-cost base case is, by definition, the
ment of the agent’s and owner’s interests firm owned solely by a single owner-manager.
were brought to attention by the seminal When management owns less than 100
contributions of Jensen and Meckling percent of the firm’s equity, shareholders
(1976) on agency costs. Agency theory has incur agency costs resulting from manage-
also brought the roles of managerial ment’s shirking and perquisite consumption.
decision rights and various external and Because of limitations imposed by personal
internal monitoring and bonding mechanisms wealth constraints, exchange regulations
to the forefront of theoretical discussions on the minimum numbers of shareholders,
and empirical research. Great strides have and other considerations, no publicly traded
been made in demonstrating empirically firm is entirely owned by management.
the role of agency costs in financial Thus, Jensen and Meckling’s zero agency
decisions, such as in explaining the choices cost base case cannot be found among the
of capital structure, maturity structure, usual sample of publicly traded firms for
dividend policy, and executive compensation. which information is readily available. The
However, the actual measurement of the absence of information about sole owner-
principal variable of interest, agency costs, manager firms explains why agency costs
in both absolute and relative terms, has are often inferred but not directly meas-
lagged behind. ured in the empirical finance literature.
To measure absolute agency costs, a zero No-agency-cost base case firms, how-
agency-cost base case must be observed to ever, can be found among non–publicly
serve as the reference point of comparison traded firms. Until recently, data on
for all other cases of ownership and man- non–publicly traded firms, which tend to be
agement structures. In the original Jensen much smaller than their publicly traded
112 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
counterparts, have been sparse. In 1997, specific operational knowledge on the part
the Federal Reserve Board released its of nonmanaging shareholders, and the lack
National Survey of Small Business of an external market for shares, however,
Finances (NSSBF), which collected data may offset the presence of dominant
from a nationally representative sample of shareholders. Additionally, heavy reliance
small businesses. Data from the NSSBF of the non–publicly traded firms on bank
enable us to analyze the relationship financing could give banks a special role in
between agency costs and ownership struc- delegated monitoring on behalf of other
ture because the survey provides financial shareholders. Thus, it would seem that
data on a group of firms whose manage- determination of the size of agency costs
ment owns 100 percent of equity. These for these firms is an empirical issue.
firms enable us to estimate the expected Our results provide direct confirmation
expense for the no-outside-equity agency-cost of the predictions made by Jensen and
base case. Furthermore, the database Meckling (1976). Agency costs are indeed
includes firms with a wide range of ownership higher among firms that are not 100 percent
and manager/owner structures, including owned by their managers, and these costs
firms owned by two individuals as well as increase as the equity share of the owner-
firms managed by outsiders with no equity manager declines. Hence, agency costs
stake. As a consequence, small firms increase with a reduction in managerial
appear well suited for a study of equity- ownership, as predicted by Jensen and
related agency costs. Meckling.These results hold true after con-
We use two alternative measures of trolling for differences across industries,
agency costs. The first is direct agency the effects of economies of scale, and dif-
costs, calculated as the difference in dollar ferences in capital structure. We also find
expenses between a firm with a certain some evidence that delegated monitoring of
ownership and management structure and small firms by banks reduces agency costs.
the no-agency-cost base case firm. This The paper is organized as follows. In
measure captures excessive expenses Section I, we discuss the nature of equity
including perk consumption. To facilitate agency costs in various ownership struc-
cross-sectional comparisons, we standard- tures and explain the broad outline of our
ize expenses by annual sales. Our second empirical model. In Section II, we provide
measure of agency costs is a proxy for the a description of the data. We present results
loss in revenues attributable to inefficient and analysis in Section III, followed by a
asset utilization, which can result from summary and conclusions in Section IV.
poor investment decisions (e.g., investing in
negative net-present-value assets) or from
management’s shirking (e.g., exerting too I. AGENCY COSTS AMONG SMALL
little effort to help generate revenue). This BUSINESSES
second measure of agency costs is calcu-
lated as the ratio of annual sales to total When compared to publicly traded firms,
assets, an efficiency ratio. We can then small businesses come closest to the type
measure agency costs as the difference in of firms depicted in the stylized theoretical
the efficiency ratio, or, equivalently, the dol- model of agency costs developed by Jensen
lar revenues lost, between a firm whose and Meckling (1976). At one extreme of
manager is the sole equity owner and a ownership and management structures are
firm whose manager owns less than 100 firms whose managers own 100 percent of
percent of equity. the firm. These firms, by their definition,
Monitoring of managers’ expenditures on have no agency costs. At the other extreme
perquisites and other personal consumption are firms whose managers are paid
relies on the vigilance of the nonmanaging employees with no equity in the firm. In
shareholders and/or related third parties, between are firms where the managers own
such as the company’s bankers. The lack of some, but not all, of their firm’s equity.
AGENCY COSTS AND OWNERSHIP STRUCTURE 113
Agency costs arise when the interests of perks, but only percent of each dollar in
the firm’s managers are not aligned with firm profit, the manager who owns less
those of the firm’s owner(s), and take the than 100 percent of the firm has the
form of preference for on-the-job perks, incentive to consume perks rather than to
shirking, and making self-interested and maximize the value of the firm to all share-
entrenched decisions that reduce share- holders. At the extreme is the manager with
holder wealth.The magnitude of these costs zero ownership ( 0), who gains 100
is limited by how well the owners and dele- percent of perquisite consumption, but zero
gated third parties, such as banks, monitor percent of firm profits (in the case when
the actions of the outside managers. salary is independent of firm performance).
To illustrate, consider those firms where Aggregate expenditure on monitoring by
a single owner controls 100 percent of the the nonmanaging shareholders decreases
stock but hires an outsider to manage the as their individual ownership shares
business. On the one hand, agency costs decline. This is due to the well-known free-
may be small because the sole owner can rider problem in spending for quasi-public
internalize all monitoring costs and has the goods, such as monitoring effort. Each
right to hire and fire the manager. More monitoring shareholder, with ownership i
specifically, such an owner incurs 100 percent must incur 100 percent of the monitoring
of the monitoring costs and receives 100 costs, but realizes only i percent of the
percent of the resulting benefits. On the monitoring benefits (in the form of reduced
other hand, the sole owner may not be able agency costs). A nonmonitoring share-
to monitor perfectly for the same reasons holder, however, enjoys the full benefits of a
that he or she hired an outside manager, monitoring shareholder’s activity without
such as lack of time or ability. Owners incurring any monitoring cost. Thus, as the
of small firms typically lack financial number of non-manager shareholders
sophistication, and may not be capable of increases, aggregate expenditure on
performing random audits or fully under- monitoring declines, and the magnitude of
standing the operating or financial results. owner-manager agency-cost problems
Consequently, these firms incur residual increases. Offsetting this relationship are
agency costs. If these costs are significant, concerns among shareholders about an
they must reflect a failure of the owner’s increase in the probability that the firm will
monitoring activities. Potential explana- be unable to pay off bank debt or secure
tions for this failure are lax monitoring by future financing from the same or new
the owners and the lack of an adequate investors, which may produce some
monitoring technology available for the restraint in agency behavior. However, as
owners. In this case, the separation of the noted by Williams (1987), these counter-
management function (initiation and vailing forces to agency behavior are
implementation) versus the control func- expected to decline in effectiveness when
tion by nonmanaging owners/shareholders the firm is not in imminent danger of
(ratification and monitoring), as suggested insolvency.
by Fama and Jensen (1983a, 1983b), To summarize, against the null hypothe-
may not be complete or effective. Thus, sis that agency costs are independent of the
residual agency costs are still expected in a ownership and control structure,1 we
sole owner firm when the manager is an postulate the following hypotheses derived
outsider. from agency theory when compared to the
Agency costs attributable to the diver- base case: (i) agency costs are higher at
gence of interests vary inversely with the firms whose managers own none of the
manager’s ownership stake. As the number firm’s equity, (ii) agency costs are an
of shareholders increases from one, the inverse function of the managers’ owner-
ownership of the owner/manager falls to , ship stake, and (iii) agency costs are an
where 0 1. Because the manager increasing function of the number of
gains 100 percent of each dollar spent on nonmanager shareholders.
114 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
error include differences in the accounting family controls the firm, the controlling
methods chosen with respect to the recog- family also fulfills the monitoring role that
nition and timing of revenues and costs, large blockholders perform at publicly
poor record-keeping typical of small busi- traded corporations. Due to more diffused
nesses, and the tendency of small-business ownership among older businesses with
owners to exercise flexibility with respect to larger families, however, monitoring by
certain cost items. For example, owners family members whose interests may not
may raise/lower expenses, including their always be aligned should be less effective
own pay, when profits are high/low. than monitoring by a sole owner.
Fortunately, these items are sources of Agency costs should increase with the
random measurement errors that may be number of nonmanager shareholders. As
reduced with a larger sample across firms the number of shareholders increases, the
in different industries and age. free-rider problem reduces the incentives
for limited-liability shareholders to moni-
tor. With less monitoring, agency costs
B. Ownership structure increase. Hence, we hypothesize that the
The corporate form of organization, with expense and asset-utilization ratios
the limited-liability provision that makes it should be positively and negatively related
more efficient for risk-sharing than propri- to the natural logarithm of one plus the
etorships or partnerships, allows the firm number of nonmanaging shareholders,
to expand and raise funds from a large respectively.7
number of investors.5 Thus, it has a richer Finally, agency costs should be higher at
set of ownership and management struc- firms managed by an outsider. This rela-
tures. The NSSBF provides four variables tionship follows directly from the agency
that we use to capture various aspects of the theory of Jensen and Meckling (1976). As
ownership structure of small-business noted above, this is the extreme case where
corporations: (i) the ownership share of the the manager gains 100 percent of
primary owner, (ii) an indicator for firms perquisite consumption, but little of the
where a single family controls more than firm’s profits.
50 percent of the firm’s shares, (iii) the
number of nonmanager shareholders,6 and, C. External monitoring by banks
(iv) an indicator for firms managed by a
shareholder rather than an outsider. Banks play a pivotal role in small business
According to theory, agency costs should financing because they are the major
be inversely related to the ownership share source of external funds for such firms.
of the primary owner. For a primary owner Cole, Wolken, and Woodburn (1996) report
who is also the firm’s manager, the incen- that more than 60 percent of the dollar
tive to consume perquisites declines as his amount of small business credit outstanding
ownership share rises, because his share of takes the form of bank loans. Petersen and
the firm’s profits rises with ownership while Rajan (1994), Berger and Udell (1995),
his benefits from perquisite consumption and Cole (1998) argue and present
are constant. For a primary owner who evidence that firm-creditor relationships
employs an outside manager, the gains generate valuable information about
from monitoring in the form of reduced borrower quality.
agency costs increase with his ownership Because banks generally require a firm’s
stake. Here, the primary owner fulfills the managers to report results honestly and to
monitoring role that large blockholders run the business efficiently with profit,
perform at publicly traded corporations. bank monitoring complements shareholder
Agency costs should be lower at firms monitoring of managers, indirectly reduc-
where a single family controls more than ing owner-manager agency costs.That is, by
50 percent of the firm’s equity. At a small, incurring monitoring costs to safeguard
closely held corporation where a single their loans, banks lead firms to operate
116 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Professional Services
Business Services
Retail
Wholesale
Transportation
Other Manufacturing
Primary Manufacturing
Construction
Figure 4.1 Operating expense-to-sales ratio by one-digit SIC for a sample of 1,708 small
corporations.
more efficiently by better utilizing assets greater perceived likelihood of the firm
and moderating perquisite consumption in switching its banking business between
order to improve the firm’s reported finan- banks.
cial performance to the bank. Thus, lower The bank’s ability to monitor is proxied
priority claimants, such as outside share- by the length of a firm’s relationship with
holders, should realize a positive externality its primary bank. A longer relationship
from bank monitoring, in the form of lower enables the bank to generate information
agency costs. Additionally, local bankers’ about the firm that is useful in deciding its
ability to acquire knowledge concerning the creditworthiness (Diamond (1984)). Both
firms from various local sources, such as Petersen and Rajan (1994) and Cole
churches, social gatherings, and interactions (1998) find that longer relationships
with the firm’s customers and suppliers, improve the availability of credit to small
makes them especially good monitors. We firms while Berger and Udell (1995) find
use two variables to represent bankers’ that longer relationships improve the terms
incentive, cost, and ability to monitor: the of credit available to small firms.
number of banks used by the firm and The bank’s incentive to monitor is prox-
the length of the firm’s longest banking ied by the firm’s debt-to-asset ratio.
relationship.8 Because our sample consists entirely of
The bank’s cost of monitoring is proxied small businesses, virtually all of the firm’s
by the number of banks from which the debt is private rather than public, and the
firm obtains financial services. The incen- majority of this debt is in the form of bank
tive for each bank to monitor may decrease loans. As leverage increases, so does the
as the number of banks with which the firm risk of default by the firm, hence the incen-
deals increases (Diamond (1984)). Part of tive for the lender to monitor the firm.While
the reduced incentive to monitor is due to a the primary purpose of this monitoring is to
form of lenders’ free-rider problems, and prevent risk-shifting by shareholders to
part is due to the shorter expected length debtholders, increased monitoring should
of banking relationships when there is a also inhibit excessive perquisite consumption
AGENCY COSTS AND OWNERSHIP STRUCTURE 117
Professional Services
Business Services
Retail
Wholesale
Transportation
Other Manufacturing
Primary Manufacturing
Construction
0 1 2 3 4 5 6 7
Sales-to-Assets Ratio
Figure 4.2 Sales-to-asset ratio by one-digit SIC for a sample of 1,708 small corporations.
by managers. (Most of the sample firms’ of 0.65 for finance and real estate and
nonbank debt is in the form of loans from professional services. Figure 4.2 shows the
finance companies and other nonbank ratio of annual sales to total assets by one-
private lenders, who also have greater digit SIC. This efficiency ratio ranges from
incentive to monitor the firm as leverage 3.6 for manufacturing to 6.2 for professional
increases.) services. Hence, these figures underscore the
importance of controlling for differences
D. Control variables across industries in our analysis of agency
costs. We do this by including a set of 35
We realize that the length of banking dummy variables, one for each two-digit
relationship variable may be correlated SIC that accounts for more than one
with firm age, which in turn could be percent of our sample of firms.
related to a firm’s efficiency. Due to the Small firms such as those surveyed by
effects of learning curve and survival bias, the NSSBF seem likely to realize scale
older firms are likely to be more efficient economies in operating expenses (e.g.,
than younger ones and, especially, than overhead items). Thus, there is a need to
start-up firms. Hence, we include firm control for firm size. This adjustment is
age as a control variable in all our tests especially important for comparisons of
involving the variable measuring the operating expenses across firms where the
length of the firm’s relationship with its difference in average size is of several
primary bank. orders of magnitude, as it is with the small
Both of our efficiency ratios vary widely businesses in our sample. Figure 4.3
across industries because of the varying confirms this, showing that the operating-
importance of inventory and fixed assets. expense-to-sales ratio declines monotonically
Figure 4.1 shows the ratio of operating by sales quartile, decreasing from 0.56 for
expenses to sales by one-digit SIC. These the smallest quartile to 0.38 for the largest
ratios vary from a low of 0.39 for quartile. If we regress the expense-to-sales
construction and manufacturing to a high ratio against annual sales, we find a
118 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
0.6
0.5
0.4
0.3
0.2
0.1
0
$0.00–$0.35 $0.35–$1.50 $1.50–$5.20 > $5.20
Sales Quartile ($Millions)
Figure 4.3 Operating expense-to-sales ratio by sales quartile for a sample of 1,708 small corporations.
0
$0.00–$0.35 $0.35–$1.50 $1.50–$5.20 > $5.20
Sales Quartile ($Millions)
Figure 4.4 Sales-to-asset ratio by sales quartile for a sample of 1,708 small corporations.
firms in these two groups for making the firm’s equity. At 368 of these 515
meaningful statistical comparisons of their firms, the owner also serves as manager; at
operating expense ratios. 147 firms, the owner employs an outsider
We find that both the median and as manager. The former group fits the
average ratios of operating expenses to definition of our no-agency-cost base case,
annual sales are considerably higher for where the manager owns 100 percent of
firms managed by outsiders (column 5) the firm and the interests of manager and
than for firms owned by shareholders owner are completely aligned. For the
(column 3). For the full sample (line 1 of latter group, the interests of owner and
Panel A), the average ratios of operating manager are completely unaligned. Thus,
expenses to assets at insider-managed these groups are of interest because they
firms and outsider-managed firms are represent the two ends of the Jensen and
46.9 percent and 51.9 percent, respectively; Meckling’s spectrum of ownership and
the 5.0 percentage-point difference in these managerial structures. Line 2 of Panel A in
means is statistically significant at the Table 4.1 shows that the ratio of operating
1 percent level. expenses to sales for the no-agency-cost
Our data enable us to provide a rough base case firm, where the manager owns
estimate of the agency costs per year 100 percent of the firm’s equity, is 46.4
attributable to the nonalignment of outside percent, as compared with 49.8 percent for
managers and shareholders. A back-of-the- firms whose owners hold all of the firms’
envelope calculation shows that, in equity but hire an outside manager. For
absolute dollars, a five-percentage-point these two groups of firms, the difference in
difference implies that the operating operating expense ratios is 3.4 percentage
expenses at a firm with median annual points. Although this univariate difference
sales of $1.3 million are $65,000 per year in means is not statistically significant, a
higher when an outsider rather than a multiple regression model that corrects for
shareholder manages the firm. The present size and industry effect, shown in Table 4.3
values of these residual equity agency costs later, indicates that firms hiring outside
are of course several times higher.10 managers have operating expenses that are
Included in the full sample are 515 firms 5.4 percent greater than those at firms
in which the primary owner controls all of managed by a shareholder.
120 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Also included in the full sample are interest in the firm (line 3), where a single
1,001 firms in which the primary owner holds family holds a controlling interest in the firm
a controlling interest of more than half of the (line 4), and where no individual or family
firm’s equity. As shown in Table 4.1, Panel A, owns more than half of the firm (line 5).
line 3, the ratio of operating expenses to For the full sample, displayed in line 1,
sales for these firms is 2.8 percentage the average sales-to-asset ratio at insider-
points lower when the owner manages the managed firms is almost 10 percent higher
firm than when the owner hires an outside than at outsider-managed firms at 4.76
manager. However, this difference is not and 4.35, respectively. The 0.41 difference
statistically significant. in these means is statistically significant at
There are also 1,249 firms in which a the 10 percent level.This difference implies
single family holds a controlling interest of that the revenues of a median-size firm,
more than half of the firm’s equity. As which has $438,000 in total assets, are
shown in line 4 of Panel A, the average $180,000 per year higher when a share-
ratio of operating expenses to sales for holder rather than an outsider manages the
these firms is 3.9 percentage points higher firm. In each of the remaining four
when the firm is managed by an outsider comparisons (lines 2–5 of Panel B), the
than when the firm is managed by a average ratio of annual sales to total assets
shareholder. This difference is statistically also is greater when the firm is managed by
significant at the 5 percent level. a shareholder than when the firm is
One final group of interest is composed managed by an outsider. However, this
of 336 firms in which no person or family difference is statistically significant at
holds a controlling interest of more than least at the 10 percent level only when the
50 percent of the firm’s equity. As predicted, primary owner holds a controlling interest
because of the more diffuse ownership of in the firm (line 3).
these firms, the average ratio of operating Overall, the results displayed in Table 4.1
expenses to sales is indeed much higher: suggest that both the ratio of operating
7.2 percentage points more at firms man- expenses to annual sales and the ratio of
aged by outsiders than at firms managed by annual sales to total assets are adequate
shareholders. This difference is statistically proxies for small corporations’ agency costs.
significant at the 5 percent level.To confirm Each provides results consistent with the
that our finding is robust with respect to predictions of agency theory for a wide range
sample distributions, we also perform non- of potentially high to low agency cost
parametric tests on the difference between organizational and management structures.
the medians, and find similar results.
A. 3. Determinants of high- and
A.2. Agency costs as measured by the low-agency cost firms
ratio of annual sales to total assets
Table 4.2 presents descriptive statistics for
In Panel B of Table 4.1 we present results the variables hypothesized to explain
from a similar analysis of agency costs, but agency costs. Statistics are presented both
here we measure agency costs by the ratio for the entire sample and for two groups of
of annual sales to total assets rather than firms constructed by dividing the entire
the ratio of operating expenses to annual sample in half, based on the sample’s
sales.11 As predicted, the results show that median ratios of agency costs. For the
the sales-to-asset ratios are higher in all entire sample (Panel A), ownership and
categories of shareholder-managed firms control is highly concentrated. On average,
versus outsider-managed firms. This is true a shareholder manages the firm 73 percent
for the full sample of 1,249 firms (line 1) of the time, the primary owner controls
and for the subsamples where the primary 65 percent of the firm’s equity, and a single
owner holds all of the firm’s equity (line 2), family owns a controlling interest in the
where the primary owner holds a controlling firm 73 percent of the time. The average
AGENCY COSTS AND OWNERSHIP STRUCTURE 121
Table 4.1 Agency Costs, Ownership Structure, and Managerial Alignment with Shareholders
Agency costs are presented for a sample of 1,708 small corporations divided into two groups of
firms: those managed by owners (aligned with shareholders) and those managed by an outsider (not
aligned with shareholders). Agency costs are proxied alternatively by the ratio of operating expenses
to annual sales and the ratio of annual sales to total assets. Separate analyses are presented for
each agency cost proxy and for subgroups where the primary owner owns 100 percent of the firm,
where the primary owner owns more than half of the firm, where a single family owns more than
half of the firm, and where no owner or family owns more than half of the firm. The last column
shows the difference between the mean (median) ratios of the outsider-managed firms and the
insider-managed firms. Statistical significance of the differences in the mean ratios is based on
the t-statistic from a parametric test (based on the assumption of unequal variances) of whether
the difference in the mean ratios of the two groups of firms is significantly different from zero.
Statistical significance of the differences in the median ratios is based on a chi-square statistic from
a nonparametric test of whether the two groups are from populations with the same median (Mood
(1950)). Data are taken from the Federal Reserve Board’s National Survey of Small Business
Finances.
Type of Manager
*, **,*** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
Table 4.2 Descriptive Statistics for Variables Used to Analyze Agency Costs
Selected variables used to study agency costs at a sample of 1,708 small corporations are identified in column 1, the sample means and medians appear
in columns 2 and 3. In columns 4 and 5 (columns 7 and 8) are the means for two groups of firms constructed by splitting the sample into two equal-size
groups of 854 firms based on the entire sample’s median operating expense-to-annual sales ratio (annual sales-to-total asset ratio). Column 6 (column 9)
shows the difference in the two groups’ means, and the results from a t-test for significant differences in the means of the low- and high-ratio groups of
firms. Data are from the Federal Reserve Board’s National Survey of Small Business Finances.
Ownership variables
Firm manager is a shareholder 0.73 1 0.78 0.69 0.09*** 0.71 0.75 0.04*
One family owns 50 percent of the firm 0.73 1 0.75 0.71 0.04* 0.72 0.74 0.02
Ownership share of primary owner 0.65 0.54 0.66 0.64 0.02 0.62 0.68 0.06***
Number of nonmanager shareholders 3.51 1 3.89 3.14 0.75** 2.35 4.68 2.33***
External monitoring variables
Length of the longest banking relationship 10.6 8 11.3 10.0 1.3*** 11.5 9.7 1.8***
(years)
Number of banking relationships 1.65 1 1.75 1.56 0.19*** 1.73 1.58 0.15***
Debt-to-asset ratio 0.60 0.52 0.61 0.58 0.03 0.55 0.64 0.09***
122 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
Control variables
Annual sales ($millions) 5.9 1.3 8.2 3.6 4.6*** 6.1 5.7 0.4
Firm age (years) 17.6 14 18.4 16.8 1.6** 19.0 16.2 2.8***
*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
AGENCY COSTS AND OWNERSHIP STRUCTURE 123
Ownership variables
Manager is a shareholder 5.4*** 5.7***
(4.1) (4.3)
One family owns 50 percent of the firm 3.0** 0.4
(2.2) (0.3)
Ownership share of primary owner 8.2*** 8.6***
(3.9) (2.8)
Log of the number of nonmanager stockholders 1.9*** 0.01
(2.8) (0.1)
External monitoring variables
Length of the longest banking relationship 0.22*** 0.3***
(3.2) (3.4)
Number of banking relationships 1.1* 1.0*
(1.9) (1.7)
Debt-to-asset ratio 0.6 0.7
(0.6) (0.7)
Control variables
Two-digit SIC dummies Yes Yes Yes Yes Yes Yes Yes Yes
Log of annual sales 2.9*** 3.1*** 3.1*** 3.1*** 2.6*** 2.6*** 2.8*** 2.9***
124 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
AGENCY COSTS AND OWNERSHIP STRUCTURE 125
8.1 percentage points, lower than those at finding conflicts with our hypothesis in
a firm where the primary owner has only a which multiple banking relationships
one percent share. Each of the variables reduce each bank’s incentive to monitor,
analyzed in columns 2 through 4 is statis- and, therefore, increase agency costs. One
tically significant at least at the 5 percent possible explanation reconciling the two
level. seemingly contradictory results is that the
In column 5 of Table 4.3 we analyze (the number of banking relationships may proxy
natural logarithm of one plus) the number for factors other than the banks’ incentive
of nonmanager shareholders. We expect a to monitor the firm. The most prominent
positive relationship between agency costs explanations are the increasing financial
and this variable, as the returns to monitor- sophistication and maturity of the firms
ing decrease and free-rider problems and their managers, and regulatory limita-
increase with the number of nonmanager tions on loans to a single borrower, which
shareholders. We use the natural logarithm may constrain a small bank’s ability to
rather than the level of this variable supply funds to a larger firm.
because we expect that the relationship is In column 8 of Table 4.3 we analyze the
stronger at smaller values of the variable. complex relation between capital structure
The estimated coefficient is positive and and ownership on agency costs. As discussed
significant at better than the one percent in Section II, we expect an inverse rela-
level, confirming our expectations. For a tionship between agency costs and the
firm with 30 non-manager shareholders, debt-to-asset ratio. We do, indeed, find a
the maximum value imposed by our cap at negative relationship, but the coefficient is
the 95th percentile, the estimated coeffi- not significantly different from zero.
cient of 1.9 implies that agency costs are In each of the seven specifications
6.5 percentage points higher, or $85,000 displayed in columns 2 through 8 of Table
greater, than at a firm with zero nonman- 4.3, observe that our size variable, the nat-
ager shareholders. ural logarithm of annual sales, is negative
In columns 6 and 7 of Table 4.3 we and statistically significant at better than
analyze the two bank monitoring variables: the 1 percent level, which is strong evidence
the length of the firm’s longest banking of economies of scale. Not shown in Table
relationship and the firm’s number of bank- 4.3 are statistics indicating that at least 20
ing relationships. As discussed in Section II, of the 35 two-digit SIC indicator variables
we expect agency costs to vary inversely included in each specification are statisti-
with the length of the longest banking cally significant at least at the 5 percent
relationship and directly with the number level. These findings underscore the critical
of banking relationships. To distinguish importance of controlling for differences
between the private information generated across industries when examining the oper-
by bank monitoring and the public informa- ating expense-to-sales ratio. The adjusted
tion generated by a firm’s durability, we R2 for each of the seven specifications in
also include firm age in the specification columns 2 through 8 indicates that the
analyzing the length of the firm’s longest models explain approximately one-quarter
banking relationship. of the variability in the ratio of operating
As shown in column 6 of Table 4.3, expenses to annual sales.
agency costs are reduced by a statistically Our final specification appears in column
significant 0.22 percent for each additional 9 of Table 4.3, where we include each of
year in the length of the firm’s longest four of the ownership variables, the two
banking relationship. The coefficient on bank-monitoring variables, and the capital
firm age is not significantly different from structure variable, along with the control
zero. In column 7, however, a related variables for firm size, age, and industrial
variable, the number of banking relation- classification. We find that each of the four
ships, is negative and statistically significant ownership variables has the predicted sign.
at better than the 10 percent level. This However, only two of the four—the indicator
126 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
2.01* 2.51** 1.27 1.27 2.49** 1.95* 3.26*** 1.34 1.17
Intercept (1.9) (2.4) (1.2) (1.2) (2.5) (1.9) (3.2) (1.2) (1.0)
Ownership variables
Manager is a shareholder 0.51** 0.30 0.46*
(2.0) (1.2) (1.8)
One family owns 50% of the firm 0.087 0.34 0.22
(0.3) (1.2) (0.8)
Ownership share of primary owner 0.012*** 0.011* 0.012***
(3.0) (1.8) (3.1)
Log of the number of nonmanager 0.82*** 1.05*** 1.05***
stockholders (6.2) (5.4) (5.4)
External monitoring variables
Length of the longest banking 0.025* 0.01 0.01
relationship (1.7) (0.4) (0.4)
Number of banking relationships 1.09*** 0.50*** 0.50***
(4.7) (4.3) (4.3)
Debt-to-asset ratio 1.05*** 1.11*** 1.11***
(5.3) (5.5) (5.5)
Control variables
Two-digit SIC dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes
Log of annual sales 0.05 0.03 0.07 0.18*** 0.07 0.11* 0.02 0.28*** 0.28***
(0.7) (0.5) (1.1) (2.7) (1.1) (1.7) (0.3) (3.9) (3.9)
Firm age 0.012 0.01 0.01
(1.2) (0.8) (0.8)
Regression summary statistics
Adjusted R2 0.032 0.030 0.035 0.051 0.035 0.042 0.045 0.080 0.080
F-statistic 2.51*** 2.40*** 2.65*** 3.49*** 2.63*** 3.02*** 3.20*** 4.37*** 4.37***
AGENCY COSTS AND OWNERSHIP STRUCTURE 127
*, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively.
128 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
inversely related to the sales-to-asset ratio, from positive and significant at better than
and is statistically significant at better than the 1 percent level in column 4 to negative
the 10 percent level. This runs counter to and significant at better than the 10 per-
our hypothesis that agency costs are lower cent level in column 9.This counterintuitive
when a firm’s bank has had more time to finding may be attributable to the high
develop valuable private information about correlation of the ownership share variable
the firm. However, this variable is not sig- with the log of number of nonmanager
nificantly different from zero in the full shareholders ( 0.75).
specification shown in column 9. In column Both of the external monitoring variables
7 we see that a related variable, number of are inversely related to the sales-to-asset
banking relationships, is negative and sta- ratio, as they are in columns 6 and 7, when
tistically significant at better than the 1 these variables are analyzed independ-
percent level. This latter finding supports ently. However, the length of the longest
the hypothesis that the values of a bank’s banking relationship variable no longer
monitoring effort and private information even approaches statistical significance
about a firm are dissipated when the firm (t 0.3), but the number of banking
obtains financial services from multiple relationships is significant at better than
sources, but, on the whole, the results the 1 percent level. The debt-to-asset ratio
regarding the bank monitoring variables remains positive and significant at better
are ambiguous. than the 1 percent level, as it is in column
In column 8 we analyze the effect of 8. Overall, the results displayed in column 9
capital structure on the sales-to-asset tend to confirm the findings previously
ratio. The results indicate that firms with discussed when the analysis variables
higher debt ratios have higher sales-to- are examined one-by-one in columns 2
asset ratios, and that this relationship is through 8.
statistically significant at better than the 1 To test whether the correlations among
percent level.This finding is supportive of a the ownership variables are responsible for
version of the theory put forth by Williams this finding, we orthogonalize the four own-
(1987) that additional debt decreases ership variables and then reestimate the
agency costs. model specification appearing in column 9.
Not shown in Table 4.4 are the results The results of this reestimation, which
concerning the industry indicator variables. appear in column 10, confirm our suspi-
In each specification, at least 20 of the 35 cions.Three of the four ownership variables
two-digit SIC indicator variables are signif- are statistically significant at least at the
icant at the 5 percent level. Once again, this 10 percent level.13 The log of the number of
finding underscores the critical importance shareholders remains negative and signifi-
of controlling for differences across indus- cant at better than the 1 percent level, but
tries when comparing agency costs. the primary owner’s ownership share
Our final specification appears in column switches back from negative to positive and
9 of Table 4.4, where we include each of the also is significant at better than the 1 per-
four ownership variables, the two monitoring cent level. The dummy indicating that the
variables, and the capital structure vari- firm is managed by a shareholder also is
able, along with the control variables for positive, but is significant at only the 10
firm size, age, and industrial classification. percent level. The dummy indicating that a
We find that only two of the four correlated family controls the firm is not significantly
ownership variables are statistically signif- different from zero. In sum, the results for
icant at better than the 10 percent level. the four ownership variables are not quali-
The natural logarithm of the number of tatively different from those reported in
nonmanager shareholders varies inversely columns 2 through 5, when each of these
with the sales-to-asset ratio and is signifi- variables is examined independently, and
cant at better than the 1 percent level. The provide strong support for the agency-cost
primary owner’s ownership share switches theory of Jensen and Meckling (1976).14
AGENCY COSTS AND OWNERSHIP STRUCTURE 129
IV. SUMMARY AND CONCLUSIONS pay to the managers as well as perks and firm
level monitoring cost, is the appropriate measure
In this article, we use data on small to test the hypothesis.
2 Data from the NSSBF yield significant and
businesses to examine how agency costs
interesting results that appear in several recent
vary with a firm’s ownership structure. published papers. See the studies on banking
Because the managerial ownership of small relationships and credit markets by Petersen
firms is highly variable, with a range from and Rajan (1994, 1995, 1997), Berger and
zero to 100 percent, we are able to Udell (1995), and Cole (1998).
estimate a firm’s agency costs across a 3 The staff at the Federal Reserve Board
wide variety of management and ownership partially edited the financial statement items
structures.15 By comparing the efficiency for violations of accounting rules, such as when
of firms that are managed by shareholders gross profit is not equal to sales less cost of
with the efficiency of firms managed by goods sold, and some improbable events such as
when accounts receivable are greater than sales,
outsiders, we can calculate the agency
or cost of goods sold equals inventory.
costs attributable to the separation of 4 Operating expenses are defined as total
ownership and control. expenses less cost of goods sold, interest
We also examine the determinants of expense, and managerial compensation.
agency costs in a multivariate regression Excessive expense on perks and other nonessen-
framework and find that our results tials should be reflected in the operating
support predictions put forth by the theo- expenses. Strictly speaking, agency costs that
ries of Jensen and Meckling (1976) and are measured by this ratio are those incurred at
Fama and Jensen (1983a) about owner- the firm level (i.e., shirking and perquisite
ship structure, organizational form, and the consumption by the managers). This may under-
estimate total agency costs since this ratio does
alignment of managers’ and shareholders’
not fully measure firm-level indirect agency costs,
interests. First, we find that agency costs such as the distortion of operating decisions due
are higher when an outsider manages the to agency problems. (See Mello and Parsons
firm. Second, we find that agency costs (1992) for an attempt to measure such costs in
vary inversely with the manager’s owner- the presence of debt.) Nor does it measure
ship share. Third, we find that agency costs off-income-statement agency costs, such as the
increase with the number of nonmanager private monitoring costs by the nonmanagement
shareholders. Fourth, we also find that, to a shareholders or the private costs of bonding
lesser extent, external monitoring by banks incurred by the manager.
produces a positive externality in the form 5 Manne (1967) and Alchian and Demsetz
(1972) agree that limited liability is an attractive
of lower agency costs.
feature of the corporate form of organization.
Jensen and Meckling (1976) point out that
although unlimited liability gives more incentive
NOTES for each shareholder to monitor, in the aggre-
gate it leads to excessive monitoring. Thus, it
* Ang is from Florida State University; Cole is may be more economical to offer a single high
from The University of Auckland, New Zealand; premium to creditors to bear risk of nonpay-
and Lin is from Montana State University. We ment and, thus, monitoring in exchange for
appreciate the comments of David Mauer, limited liability.
Michael Long, René Stulz (the editor), and an 6 Technically, the survey does not provide a
anonymous referee. variable for the number of nonmanager share-
1 Theoretical support for the null hypothesis holders. Rather, we define this variable as the
is due to Demsetz (1983), who suggests that the number of shareholders for firms that have an
sum of amenities for on-the-job consumption outside manager and as the number of share-
and take-home pay for similar quality managers holders less one for firms that have an insider
is the same for both high-cost and low-cost manager.
monitoring organizations. The proportion paid 7 This formulation recognizes the unequal and
to the managers, however, differs according to diminishing role of additional shareholders, and
the cost of monitoring. Here, it would seem that the problem of undefined zero when there is no
total operating expense, which include direct other shareholder.
130 ALTERNATIVE PERSPECTIVES ON CORPORATE GOVERNANCE SYSTEMS
8 These are also the same governance 14 For the sake of completeness, we also
variables used by Berger and Udell (1995) and perform the same procedure on the regression
Cole (1998) in their studies of banking rela- equation in column 9 of Table 4.3. We find that
tionship. However, none of their variables, except orthogonalizing the ownership variables does
for the corporate form dummy, are found to not qualitatively affect the results. Only the
significantly affect either the loan term or the same two ownership variables are statistically
use of collateral. significant. Overall, although both measures of
9 In Fama and Jensen (1983a), the delegation agency costs provide qualitatively similar
of decision control management and residual results, the expense ratio regression yields
owner (i.e., hiring of outsiders as managers) is greater explained variations.
related to the decision skill and the accompanying 15 There are few empirical studies in related
specialized knowledge that are needed to run areas of corporate finance that analyze ownership,
the firm. Shareholders, however, still have to organizational, and management structures in
bear the costs of monitoring. detail. For example, see a study of executive
10 As a way of comparison, Dong and Dow compensation in Israel by Ang, Hauser, and
(1993) estimate that 10 to 20 percent of total Lauterbach (1997).
labor hours are attributed to supervision or
monitoring in the Chinese collective farms.
Dobson (1992) finds that X-inefficiency REFERENCES
measures 0.2 percent of sales among large U.S.
manufacturing firms. Alchian, Armen A., and Harold Demsetz, 1992,
11 The distribution of the sales-to-asset ratio is Production, information costs, and economic
highly skewed by the presence of large outliers. organization, American Economic Review
Consequently, the ratio is capped at the value 62, 777–795.
found at the 95th percentile, a ratio of 19.0. Ang, James S., Samuel Hauser, and Beni
12 Fama and Jensen (1983) realize that for Lauterbach, 1997, Top executive compensa-
shareholders to monitor the firm’s management tion under alternate ownership and
they must hold sufficient ownership; however, governance structure, in Mark Hirschey and
the cost of large ownership shares is suboptimal
M. Wayne Marr, eds.: Advances in Financial
risk-taking, and, possibly, underinvestment. Also,
Demsetz (1983) suggests that firms with concen- Economics (JAI Press Inc., Greenwich,
trated ownership have lower monitoring costs. Conn.).
13 The sales-to-asset ratio is likely subject to Arnold, R. J., 1985, Agency costs in banking
additional biases that render it much noisier firms: An analysis of expense preference
than the operating expense-to-sales ratio. Note behavior, Journal of Economics and Business,
that the adjusted-R2 statistics appearing at the 103–112.
bottoms of Tables 4.3 and 4.4 indicate that we Barclay, Michael J., and Clifford W. Smith Jr.,
are able to explain about 26 percent of the 1995, The maturity structure of corporate
variability in the latter ratio but only about eight debt, Journal of Finance 50, 609–632.
percent of the variability in the former ratio.
Berger, Allen N., and Gregory F. Udell, 1995,
This led us to investigate additional control vari-
ables in the sales-to-asset regression, including Relationship lending and lines of credit in
the ratio of operating expenses to sales. Results small firm finance, Journal of Business 68,
from specifications including the operating 351–382.
expense-to-sales ratio as an additional regressor Cole, Rebel A., 1998, The importance of rela-
are not qualitatively different from those in tionships to the availability of credit, Journal
Table 4.4. In no case was the operating expense- of Banking and Finance, forthcoming.
to-sales ratio statistically significant at even the Cole, Rebel A., and Hamid Mehran, 1998, CEO
20 percent level. Because we view the operating compensation, ownership structure, and taxa-
expense-to-sales ratio as an endogenous vari- tion: Evidence from small businesses, mimeo,
able, we also tested a specification that included
Northwestern University.
a predicted value of this ratio rather than the
actual value. The predicted value was obtained Cole, Rebel A., and John D. Wolken, 1995,
using the model appearing in column 9 of Sources and uses of financial services by
Table 4.3. Again, the results from this robustness small businesses: Evidence from the 1993
check are not qualitatively different from those National Survey of Small Business Finances,
appearing in Table 4.4 Federal Reserve Bulletin 81, 629–667.
AGENCY COSTS AND OWNERSHIP STRUCTURE 131
Cole, Rebel A., John D. Wolken, and Louise agency costs and capital structure, Journal of
Woodburn, 1996, Bank and nonbank compe- Financial Economics 3, 305–360.
tition for small business credit: Evidence Kim, Wi S., and Eric H. Sorensen, 1986,
from the 1987 and 1993 National Surveys of Evidence on the impact of the agency costs of
Small Business Finances, Federal Reserve debt in corporate debt policy, Journal of
Bulletin 82, 983–995. Financial and Quantitative Analysis 21,
Demsetz, Harold, 1983, The structure of owner- 131–144.
ship and theory of the firm, Journal of Law Knight, Frank H., 1921, Risk, Uncertainty and
and Economics 26, 375–393. Profit (University of Chicago Press,
Demsetz, Harold, and Kenneth Lehn, 1985, The Chicago).
structure of corporate ownership: Causes and Manne, Henry G., 1967, Our two corporate sys-
consequences, Journal of Political Economy tems: Law and economics, Virginia Law
93, 1155–1177. Review 53, 259–284.
Denning, Karen C., and Kuldeep Shastri, 1993, Mello, Antonio S., and John E. Parsons, 1992,
Changes in organizational structure and Measuring the agency cost of debt, Journal of
shareholder wealth, Journal of Financial and Finance 47, 1887–1904.
Quantitative Analysis 28, 553–564. Milgrom, Paul, and John Roberts, 1995,
Diamond, Douglas W., 1984, Financial interme- Economics, Organization, and Management
diation and delegated monitoring, Review of (Prentice Hall, Englewood Cliffs, N.J.).
Economic Studies 51, 393–414. Petersen, Mitchell A., and Raghuram G. Rajan,
Dobson, John, 1992, Agency costs in U.S. man- 1994, The benefits of lending relationships:
ufacturing: An empirical measure using Evidence from small business data, Journal
X-efficiency, Journal of Economics and of Finance 49, 3–38.
Finance 16, 1–10. Petersen, Mitchell A., and Raghuram G. Rajan,
Dong, Xiao Y., and Gregory K. Dow, 1993, 1995,The effect of credit market competition
Monitoring costs in Chinese agricultural on lending relationships, Quarterly Journal of
teams, Journal of Political Economy 101, Economics 110, 407–442.
539–553. Petersen, Mitchell A., and Raghuram G. Rajan,
Fama, Eugene F., and Michael C. Jensen, 1997, Trade credit: Theories and evidence,
1983a, Separation of ownership and control, Review of Financial Studies 10, 661–691.
Journal of Law and Economics 26, 301–325. Shleifer, Andrei, and Robert Vishny, 1986,
Fama, Eugene F., and Michael C. Jensen, Large shareholders and corporate control,
1983b, Agency problems and residual claims, Journal of Political Economy 94, 461–488.
Journal of Law and Economics 26, 327–349. Williams, Joseph, 1987, Perquisites, risk, and
Fama, Eugene F., and Michael C. Jensen, 1985, capital structure, Journal of Finance 42,
Organizational forms and investment deci- 29–48.
sions, Journal of Financial Economics 14, Zeckhauser, Richard, and John Pound, 1990,
101–119. Are large shareholders effective monitors?
Jensen, Michael C., 1986, Agency costs of free An investigation of share ownership and
cash flow, corporate finance and takeovers, corporate performance, in R. G. Hubbard, ed.:
American Economic Review 76, 323–329. Asymmetric Information, Corporate Finance,
Jensen, Michael C., and William H. Meckling, and Investment (University of Chicago Press,
1976,Theory of the firm: Managerial behavior, Chicago).
Part 2
Equity ownership structure
and control
INTRODUCTION
outcomes after initial public offerings in terms of takeover incidence, the ownership
structure of bidders, and the differential characteristics of acquired firms that attract
concentrated and dispersed acquirers. Goergen and Renneboog relate these differences to
the differences in shareholder rights and laws protecting minority shareholders in
Germany and the UK.
Although in the UK concentrated ownership of the kind observed in Germany is uncommon,
as noted in the previous paper, large block shareholdings in the form of institutional share-
holding are quite widespread in the UK. Institutional shareholders – insurance companies,
pension funds, mutual funds, investment trusts – collectively hold a substantial majority
of shares in UK listed companies. Given their dominance, do institutional large block
shareholders monitor their investee companies effectively? This question is addressed by
Faccio and Lasfer (Ch.9) with particular reference to occupational pension funds. By
comparing companies in which these funds hold large stakes with a control group of com-
panies listed on the London Stock Exchange, they show that occupational pension funds
hold large stakes over a long time period mainly in small companies. However, the value
added by these funds is negligible and their holdings do not lead companies to comply with
the Code of Best Practice or outperform their industry counterparts. Overall, their results
suggest that occupational pension funds are not effective monitors.
Chapter 5
Hiroshi Osano*
SECURITY DESIGN, INSIDER
MONITORING, AND FINANCIAL
MARKET EQUILIBRIUM
Source: European Finance Review, 2(3) (1999): 273–302.
ABSTRACT
This paper considers a problem of security design in the presence of monitoring done by a large
investor to discipline the management of a firm. Since the large investor enjoys only part of
the benefits generated by her monitoring activities but incurs all the associated costs, the
design and amount of security need to be structured so as to motivate her to maintain an
efficient level of monitoring, if no other mechanism exists to make her commit to specific levels
of monitoring in advance. By assuming that the large investor takes account of the effect of
the issued amount of security on the revenues received, we show that the optimal security is a
debt-like security such as standard debt with a positive probability of default, or debt with call
options. We also verify that the financial market equilibrium is constrained Pareto optimal.
The main thesis of this paper depends on expectation implies that in this situation a
a mechanism in which the design and lower security price will prevail. Such an
issued amount of security influences the indirect effect of the design and issued
intensity of the large investor’s monitoring, amount of security causes a liquidity cost
not only through a direct effect on her that is borne by the large investor, who
ex ante payoff but also through an indirect issues the security. Thus, the large investor
effect via the market security price on her needs to consider not only the direct effect
ex ante payoff. To clarify the basic idea, of the design and issued amount of security,
consider first an entrepreneur (a large including a retention cost associated with a
investor) who holds a project that generates high unsecuritized or unsold portion of the
risky returns in future periods. She may cash flows, but also the indirect effect of
keep some of the future returns for herself the design and issued amount of security
and securitize the other portion of the through a change in the security price,
future returns to raise capital and maintain including a liquidity cost associated with a
liquidity. Once the large investor issues a low rate of monitoring.
security, she turns effective control of the The above arguments show that there is
firm over to a manager. However, the large a potential conflict between the need for
investor is assumed to have access to costly liquidity preference and the efficiency of
technology for monitoring the management monitoring. Furthermore, these arguments
of the firm and thus can affect the return indicate that the design of the security and
structure of the project.3 However, if the the amount issued are incentive devices
large investor carries out such monitoring, independent of one another for reconciling
small and passive investors can free-ride on the friction between these two aims.
part of the benefits of the control function Nevertheless, the interaction between the
of the large investor, although these design of the security, the amount issued,
investors incur none of the costs of and the large investor’s monitoring
monitoring. This externality reduces the remains imperfectly understood in the
likelihood of the large investor’s carrying literature of corporate governance or
out high intensity monitoring unless she finance. Indeed, the literature reveals
can be made to commit to prior monitoring several difficulties.The first problem is that
levels before she designs and issues the only a limited menu of debt and equity has
security. been studied. The second problem is that
In this model, the design and issued most of the arguments in this field depend
amount of security affects the large on the bilateral contract framework but not
investor’s monitoring because of both the on the market equilibrium framework.4
direct effect on her ex ante payoff and the To avoid these difficulties and formalize
indirect effect on her ex ante payoff the trade-off between the retention cost
through a change in the security price. The and the liquidity cost, we incorporate four
logic of the indirect effect is explained as key issues into our model. First, to motivate
follows. If small and passive investors are the entrepreneur (large investor) to issue a
aware that a change in the design and security backed by future risky returns, we
issued amount of security affects the large assume that she must sell her issued
investor’s monitoring, the security price at security for liquidity reasons, as studied in
which the trade takes place will reflect this Allen and Gale (1988, 1991), Gale
awareness. (1992), and DeMarzo and Duffie (1999).
More specifically, small and passive Thus, the retention cost arises from retaining
investors rationally anticipate a low inten- a large amount of unsecuritized returns or
sity of monitoring if the large investor unsold securities. Second, we assume that
designs and sells to small and passive the large investor can choose a level of
investors a security that is more sensitive monitoring to discipline the management of
to the variations of future returns or if she the firm and affect the structure of risky
sells a security more aggressively. This returns in future periods, while she cannot
SECURITY DESIGN AND INSIDER MONITORING 137
after the security claims are paid to cm, where c is the unit monitoring cost. Let
security holders including the large m be the minimum monitoring level
investor herself. required to sustain the monitoring
To formalize the model, we require process.12
specific assumptions about the liquidity All the project returns at each state, (Xg,
motive of the large investor, the preference Xb), and the fraction of the security sold by
of outside investors, and the monitoring the large investor to the market, qL, can be
technology of the large investor. verified with no cost. On the other hand, the
To provide the motivation for liquidation, monitoring level chosen by the large
we assume that the large investor evaluates investor, m, cannot be observed by any
the period 2 cash flows at a rate ∈ (0, 1). outside agents. Thus, the large investor
This assumption implies that the large cannot be committed to any specific levels
investor is indifferent as to whether she of monitoring until period 1. The functions
keeps future project returns or sells them ( g(m), b(m)) and the parameters
for cents in the dollar in cash by securi- (Xg, Xb, c, ) are assumed to be common
tizing them. This also means that, if the knowledge.
large investor retains a portion of the secu- The security is represented by a claim,
rity, the private value of one dollar’s worth that is, any promise to make a future pay-
of security returns to the large investor is . ment, contingent on the state of nature
Similar assumptions are made in Allen and (that is, the project returns). Let F̃ be a
Gale (1988, 1991), Gale (1992), and real-valued random variable that generates
DeMarzo and Duffie (1999).8 Since we a security claim Fg at the good state, and
assume that high transaction costs deter a security claim Fb at the bad state,
the large investor and outside investors respectively. We focus our attention on the
from building a bilateral contract relation, set of admissible securities. These are
the large investor has an incentive to defined by a set of functions satisfying the
securitize some portion of the future risky following limited liability and monotonicity
project returns and sell the security to out- conditions:
side investors.
In contrast, outside investors are 0 Fs Xs, s g, b, (1)
assumed to have no liquidity motives. Since Fb Fg. (2)
outside investors do not care when asset
returns are generated, they have no concern In the subsequent discussion, we denote the
about the timing of payments. set of admissible securities as . In the
The large investor has access to costly limited liability condition (1), the restric-
technology of monitoring the manage- tion of 0 Fs for s g, b expresses limited
ment. The technology affects the structure liability for security holders.The restriction
of the returns of the risky project as of Fs Xs for s g, b implies limited lia-
follows:9,10 bility for the large investor as a residual
claimant. The monotonicity condition (2)
X , with probability (m),
X̃ Xg , with probability g (m),
b b
shows that the security claim is nonde-
creasing in the available firm cash flows.13
These assumptions are quite common in
where Xg Xb 0, ( g(m), b(m)) ∈ the finance literature, and are justified in
{( g(m), b(m)) | g(m) b(m) 1, Innes (1990), Nachman and Noe (1994),
g(m) 0, b(m) 0}, and m is the inten- Hart and Moore (1994), and DeMarzo and
sity of monitoring.11 One part of the Duffie (1999).
efficiency of monitoring is captured by the Now, the security represented by the
function g(m), where g’(m) 0 and claim (Fg, Fb) pays an amount min (Fs, Xs)
”g(m) 0.The other part of the efficiency for s g, b when the returns of the project
of monitoring is characterized by the are realized in period 2.14 The large
cost of monitoring at level m, expressed by investor then receives the cash revenues from
140 EQUITY OWNERSHIP STRUCTURE AND CONTROL
the sale of the fraction qL of the security in Since outside investors do not behave
period 1, and the residual cash flows of strategically, they take the security price p
max (Xs Fs, 0) for s g, b and the as given.
claims for the unsold fraction (1 qL) of A financial market equilibrium is now
the security in period 2. characterized by (m , F̃ , qL, qO, p ) that
The trading process is described as a satisfies the following conditions (see
Walrasian process in which the large Figure 5.1):
investor is strategic. The large investor
FINANCIAL MARKET EQUILIBRIUM:
chooses the fraction of the security sold to
(i) Sequential Optimality of the Large
the market, qL, by taking into account the
Investor. In period 1, the large investor
effect of the issued amount on the security
chooses a monitoring level m such that
price, p. We assume that the large investor
perceives an inverse demand correspon-
dence p ∈ P(qL; F̃) for the issued security
F̃, where P : [0, 1] → ℜ. In the next
L1(F̃, qL) max RL(m, F̃ , qL).
m m
(5)
section, we specify how P(qL; F̃ ) is
determined. On the other hand, outside In period 0, given the knowledge of the
investors behave as representative, risk - price correspondence P(qL; F̃), the large
neutral, price-taking investors. Thus, out- investor chooses a design F̃ and a fraction
side investors do not receive expected net q L of the security sold to the market
gains from buying any security at market such that
equilibrium.
The expected payoff of the large investor L0 max
F̃僆 , qL僆[0,1]
L1(F̃ , qL), (6)
in period 0 is then given by
where is the set of admissible securities
RL(m, F̃, qL) E max (X̃ F̃, 0) defined by (1) and (2).
(ii) Optimality of Outsider Investors. In
(1 qL)E min (F̃, X̃)
period 1, outside investors purchase a fraction
qLP(qL; F̃) cm, (3) qO of the security supplied in the market,
relative to an observed price p , such that
where E is the expectation operator.
The right-hand side of (3) consists of the
expected residual claims retained by the
0 (p ) max Ro( qo, p ).
qo僆[0,1]
(7)
large shareholder, E max (X̃ F̃, 0)
the expected claims for the unsold fraction (iii) Market Equilibrium.The security price
(1 qL) of the security, (1 qL)E min p and the sold and purchased fractions
(F̃, X̃ ), the cash revenues raised from the (q L, q O) must satisfy
sale of the security, qLP(qL; F̃ ), and the
monitoring cost, cm. The first two terms p 僆 P(qL; F̃ ), (8)
are discounted by the rate because the and
returns of these two claims accrue only in qL qO. (9)
period 2. The monitoring cost is not dis-
counted, since it must be paid in period 1. Note that the large investor selects a
Note that the first two terms clearly security design F̃ from the set of admissible
depend on the monitoring level m because securities , a sold fraction qL from [0, 1],
the realized probability of the state is and a monitoring level m from (m, ∞).
affected by the choice of m. Outside investors select a purchased fraction
The expected payoff of outsiders in qO from [0, 1]. Since the large investor can-
period 0 is similarly represented by not be committed in advance to any specific
levels of monitoring, she must solve the
Ro(qo, p) qo [E min (F̃, X̃) p]. (4) sequential problem made up of (5) and (6).
SECURITY DESIGN AND INSIDER MONITORING 141
period 0
large investor risky project
(insider)
security trade
period 1
large investors
period 2
stochastic project returns
Rg
R= g
Rb b
3. FINANCIAL MARKET
0 if E min (F̃, X̃ ) < p,
EQUILIBRIUM q0 僆 [0,1] if E min (F̃, X̃ ) p,
0 if E min (F̃, X̃ ) < p.
In this section, we first derive the inverse
demand correspondence P(qL; F̃) from the (10)
condition of the optimality of outside
investors and the condition of market
Combining (9) and (10), we have
equilibrium by taking as given a security
design chosen by the large investor, F̃ . We
next solve the sequential optimization 0 if E min (F̃, X̃ ) < p,
problem of the large investor in two steps, qL 僆 [0,1] if E min (F̃, X̃ ) p,
and determine the remaining variables 0 if E min (F̃, X̃ ) > p.
endogenously.
(11)
3.1. Inverse demand correspondence
The relation between qL and p determined
Solving the optimization problem of out- by (11) gives the inverse demand corre-
side investors, (7), with (4), we see spondence p ∈ P(qL; F̃).
142 EQUITY OWNERSHIP STRUCTURE AND CONTROL
3.2. The monitoring problem faced We next examine the case of E min
by the large investor (F̃, X̃) p, where qL 僆 [0, 1] from (11).
Rearranging (3) with the definitions of X̃
We now proceed to the sequential and F̃, we explicitly describe the interim
optimization problem of the large investor. maximization problem (5) by
We discuss the optimality conditions for
this problem by moving backward from
period 1 to period 0. max RL(m, F˜, qL)
m
m
We begin with solving the monitoring
problem (5) faced in period 1 by the large
investor, who designs a security F̃ and
max
m
m max(X F 0) (m)
ig,b
i i, i
Fb Fg Xb Xg
∆ (A)
Fb Xb Fg Xg
Fb Xb Fg = Xg
∆ (B)
Fb = Xb Fg Xg
∆ (C)
Fb = Xb Fg = Xg
∆ (D)
solution to the monitoring problem is
specified in the Appendix. In Subsection 3.3 P(1; F˜ ) cm
144 EQUITY OWNERSHIP STRUCTURE AND CONTROL
optimal value for the problem (6) under
each of the four possible restrictions, and F*g min Xb (1 )g(m*)
choose the maximal value among the opti-
(m*)Fm**
1
mal values of the four possible cases. The
g , Xg , (18b)
corresponding solution gives an optimal g
security design F̃ (Fg, F b) and an optimal
level of the sold fraction qL to the problem F*g Xb, (18c)
(6). In contrast to the analysis of the
monitoring problem, the large investor must
consider the effect of a change in her (1 )[Fg*g(m*) Xbb(m*)]
strategy on the security price. A detailed m*
q*L(Fg* Xb)g’ (m*) 0. (18d)
analysis of the optimizing conditions to the q*L
SECURITY DESIGN AND INSIDER MONITORING 145
The large investor’s payoff is then begin with (18a). Given the security price
determined in period 0, an increase in the
A
L0 c [X (m ) X (m )]
L0 g g
*
b b
* monitoring effort in period 1 affects the
ex ante payoff of the large investor
through two routes. One effect increases
q*L(1 )[F*g g(m*) the expected project returns, thereby
further increasing not only the discounted
Xbb(m*)] cm*. (18e) residual returns to be received by the large
investor,that is,(Xg F*g Xb F*b )’
g(m*),
but also the discounted security returns for
(ii) If F̃ is restricted to (B) or (D),
the unsold portion to be received by the
the optimal solution to the constrained
large investor, that is, (1 q*L) (F*g F*b)
maximization problem, (m**, F** ** **
g , Fb , q L ),
is represented by ’g(m*). The combined effects increase the
ex ante payoff of the large investor,
(1 q**
L )(Xg Xb)g(m ) c, (19a)
’ **
expressed by the left-hand side of (18a).
The other effect increases the total moni-
Fg** Xg, (19b) toring cost, c, thus decreasing the ex ante
payoff of the large investor.This cost effect
b Xb,
F** (19c) is represented by the right-hand side of
(1 )[Xgg(m**) Xbb(m**)] (18a). The monitoring level is then chosen
to balance the former benefit against the
m** latter cost.
q**
L (Xg Xb)g(m )
’ **
0.
q**
L Before proceeding to (18b) and (18c),
we discuss the implications of (18d)
(19d)
because it is then more straightforward to
The large investor’s payoff is then understand (18b) and (18c). An increase in
the sold fraction has two potentially
B
Lo D
Lo [ q**
L (1 )]
conflicting effects on the ex ante payoff of
the large investor. One effect causes the
large investor to receive higher revenues
[Xgg(m**) Xbb(m**)] cm**. from the sale of the security. This effect
enables the large investor to satisfy any
(19e)
desire to obtain as much revenue as possi-
ble at date 0 rather than at date 1 thus
Proposition 1 shows that, if F̃ is restricted directly increasing her ex ante payoff,
to (A) or (C), the optimal security can expressed by (1 )p* (1 )
be a debt-like security with F* g Xg and [F*g g(m*) Xb b(m*)]. The other effect
F*b Xb. One interpretation of this kind of of an increase in qL reduces the correlation
security is standard debt with default such between the ex post project returns and the
as that analyzed in DeMarzo and Duffle ex post payoff of the large investor because
(1999) and von Thadden (1995): F* g can be she obtains the more deterministic sales
viewed as the face value, and F* b as the revenues at period 0. Since this effect leads
default payment of debt. Another interpre- to a lower monitoring level, outside
tation is debt with call options (convertible investors rationally anticipate a lower
bond or debt with warrant): Xg F* g can intensity of monitoring. This expectation
then be interpreted as the option premium. reduces the security price, thus indirectly
Proposition 1 also indicates that if F̃ is decreasing the large investor’s revenues
restricted to (B) or (D), the optimal from the sale of the security, represented by
security is equity whose claim fully reflects q*L(F*g Xb) g’(m*)[∂m*/∂q*L].The former
the project returns. direct benefit cancels out the latter indi-
It is instructive to explore the implica- rect cost at the optimal level of the sold
tions of equations (18) and (19). Let us fraction.
146 EQUITY OWNERSHIP STRUCTURE AND CONTROL
marized by q*L δ g(m*), which is negative (m*, F* > ,
g ,F*b , q*
L ) if
LO LO
the revenues as much as possible at date 0 possible outcomes and so on. Second,
rather than at date 1. This is because the Admati, Pfleiderer, and Zechner (1994)
large investor would receive the full discuss the monitoring activity of the large
margin benefit from monitoring as well as investor when only equity can be issued.
bearing the full marginal cost of monitoring. However, Proposition 2 implies that the
However, in the present case, the large optimal security becomes debt-like in the
investor can sell some portion of the secu- presence of the monitoring problem,
rity and receive the revenues from the sale although our framework is restricted. This
of the security. Since the discount factor suggests that the security design problem
is less than one, the sales revenues received needs to be analyzed endogenously in the
in period 0 are higher than the residual study of the monitoring of the large investor.
claims received in period 2. Hence, unless
the monitoring level decreases through a
change in security payments, the large 4. NORMATIVE ANALYSIS OF THE
investor is motivated to increase the sales FINANCIAL MARKET EQUILIBRIUM
revenues received in period 0. This implies
that, in the absence of monitoring consid- In this section, we compare the financial
erations, the large investor is motivated to market equilibrium in the preceding section
set security claims at each state to with several benchmark cases: the compet-
increase by the maximum amount that the itive equilibrium allocation, the passive
firm can pay. If the security claims need equilibrium allocation, and the constrained
not be constrained by monitoring consider- social surplus maximizing allocation.
ations, then the optimal security becomes The competitive equilibrium allocation
equity because F*g Xg and F*b Xb. On deals with the situation in which the large
the other hand, if the security claims are investor does not take account of the effect
constrained by monitoring considerations of the issued amount of security on her
to some extent, the large investor needs to sales revenues. The passive equilibrium
enhance her monitoring activity by allocation arises from the situation in
increasing the residual revenues received which the large investor chooses the mini-
at the good state in period 2. The large mum monitoring level. The constrained
investor would therefore face the possibil- social surplus maximizing allocation corre-
ity that the optimal security becomes stan- sponds to the constrained Pareto optimal
dard debt with default because she must allocation such that the social planner
set F*g less than Xg while keeping F*b Xb. maximizes the social surplus subject to sev-
More specifically, in the presence of eral information constraints. In the analysis
monitoring considerations, the large that follows, we show that the financial
investor needs to consider the trade-off market equilibrium in the preceding section
between a decline in Fg from Xg and a rise is constrained Pareto optimal, whereas
in qL. In other words, she needs to balance the competitive and passive equilibrium
the costs of securitizing a smaller portion allocations are not. With regard to the
of the project returns against the costs of competitive and passive equilibrium
selling a smaller portion of the issued secu- allocations, we suggest that they are identical.
rity.16 Since the total effects of a decrease in
Fg from Xg and an increase in qL on the ex PROPOSITION 3. If the large investor does
ante payoff of the large investor per issued not take account of the effect of the issued
amount are positive, the optimal security amount of security on her sales revenues, the
involves debt-like types such as standard optimal allocation, (m̂, F̂g, F̂b, q̂L), would be
debt with default or debt with call options. given such that the monitoring level of the
Several remarks about this proposition large investor, m̂, is minimized: more specif-
are in order. First, we should notice that the ically, m̂ = m, F̂g = Xg, F̂b = Xb, and q̂L 1.
result of this proposition depends on Proof. If the large investor does not take
the simplifying assumptions such as two into account the effect of the issued
148 EQUITY OWNERSHIP STRUCTURE AND CONTROL
amount of security on her sales revenues, Let S(m, F̃, q) be the social surplus
the optimality conditions for the problem defined by
(6) provided by (A5)–(A8) in the Appendix S (m, F̃, q) EX̃ q(1 )
show that her optimal strategy is given by E min (F̃, X̃) cm, (23)
m̂ m, F̂g Xg, F̂b Xb, and q̂L 1 in all
possible cases, provided the security market where the first term represents the project
does not collapse. returns, the second term expresses the liquidity
benefits of the large investor, and the third
The intuitive explanation for Proposition 3
term indicates the monitoring cost.
is clarified as follows. If the large investor
Then, the constrained social surplus
does not consider the effect of the issued
maximizing allocation is formally charac-
amount of security on her sales revenues, it
terized by the following sequential
is optimal for the large investor to securi-
maximization problem:
tize all the project returns and sell all of the
security, because she evaluates the sales Constrained social surplus maximizing
revenues received in period 0 more than the allocation. In period 1, the social planner
residual claims received in period 2. Since chooses a monitoring level m such that
the large investor need not take account
of the monitoring problem, she can minimize S1(F̃, q) m
max
m S(m, F̃, q).
(24)
the monitoring level. However, this might
cause the collapse of the security market In period 0, the social planner chooses a
because outside investors would rationally design F̃ and a fraction q of security sold to
anticipate the large investor’s strategy and outside investors such that
participate in the security market only if the
security price is low enough.
S0 max S1(F̃, q). (25)
Proposition 2 indicates that in the F̃僆 ,q僆[0,1]
presence of the monitoring problem, the
optimal security is a debt-like security if
Given (A1) and (A3) with E min (F̃, X̃) p
the large investor takes account of the
in the Appendix, it can be seen immediately
effect of the amount sold on the revenues
from (23) that the constrained social
received. On the other hand, Proposition 3
surplus maximization problem is equivalent
suggests that even in the presence of the
to the sequential optimization problem of
monitoring problem, the optimal security is
the large investor at financial market equi-
equity if the large investor does not take
librium. Thus, we obtain the following
account of the effect of the amount of
proposition:
security issued on her sales revenues, that
is, if she chooses the minimum monitoring
PROPOSITION 4. If the large investor
level. The result of Proposition 3 depends
takes account of the effect of the amount
on the assumption that the large investor
sold on the revenues received, the financial
cannot be committed to any prior levels of
market equilibrium characterized in the pre-
monitoring until she designs and issues the
ceding section is constrained Pareto optimal
security.
in the sense that it attains the constrained
We next examine the relation between
social surplus maximizing allocation.
the financial market equilibrium and
constrained social surplus maximizing In view of Propositions 2–4, we see that
allocations, defined such that the social the competitive equilibrium allocation is
planner maximizes the social surplus subject not constrained Pareto optimal. Thus, the
to the constraints: (i) that the large investor large investor must take into account the
initially owns the risky project,17 (ii) that her effect of the issued amount of security on
monitoring level is unobservable to outside her sales revenues to attain the constrained
investors, and (iii) that she cannot be com- social surplus maximizing allocation.
mitted to any specific levels of monitoring This finding is reminiscent of the results of
before she designs and issues the security. Allen and Gale (1991) in a different context.
SECURITY DESIGN AND INSIDER MONITORING 149
The intuition behind this proposition is constrained Pareto optimal or the market
that the social surplus can be transformed collapses. Furthermore, unless the market
into the surplus of the large investor as a breaks down, the optimal security in this
result of the price adjustment at financial case is always equity; and the monitoring
market equilibrium. This finding mainly level of the large investor is minimized.
depends on both the price-making behavior of As suggested in DeMarzo and Duffie
the large investor and the risk neutrality (1999), the model rather closely fits into
of the large investor and outside investors. the context of the design of corporated
securities that are sold by an informed
underwriter who retains some fraction of
5. CONCLUSION the security of issuing firms. In Germany
and Japan, the recent trend in the financing
patterns of nonfinancial firms away from
This paper has developed a framework for
(indirect) bank borrowing and toward
analyzing a problem of security design in
(direct) bond financing has been affecting
the presence of monitoring done by a
their corporate governance systems.
large investor to discipline the manage-
However, since German house banks and
ment of a firm. Since the large investor
Japanese main banks usually retain the
enjoys only a part of the benefits brought
security of their borrowing firms, our result
about by her monitoring activities but
indicates that these intermediaries may still
incurs all the associated costs, she needs
discipline the management of firms by
to structure the problem of security
designing a debt-like security for their
design to motivate herself to make an
borrowing firms and retaining some portion
efficient level of monitoring if she cannot
of the security.
be committed to any prior levels of moni-
Within our framework, many issues
toring before she designs and issues the
remain to be investigated. Among other
security. In fact, it is costly to design and
things, the problem of multi-security design
issue the security to attain an efficient
is an important question that we hope to
level of monitoring because of both the
pursue in future research. Furthermore, the
direct effect on the ex ante payoff of the
spanning issue caused by the risk averting
large investor and the indirect effect on
actions of investors should also be exam-
her ex ante payoff through a change in the
ined. Finally, as has been recently studied
security price.
under the incomplete contract model, the
By assuming that the large investor
bankruptcy and liquidation problem is an
takes account of the effect on her sales rev-
important topic of corporate governance
enues of the amount of security issued, we
(see Hart and Moore (1989), Bolton and
have shown that the optimal security is a
Scharfstein (1990), Aghion and Bolton
debt-like security such as standard debt
(1992), Berglöf and von Thadden (1994),
with a positive probability of default, or
and Dewatripont and Tirole (1994)). It will
debt with call options in the presence of the
be very interesting to explore how the design
monitoring problem. This result suggests
of securities can resolve the bankruptcy and
that it is not necessarily optimal to issue a
liquidation problem at financial market
security with zero default probability such
equilibrium.
as risk-free debt, and that the monitoring
model of Admati, Pfleiderer, and Zechner
(1994) is restricted in the sense that only APPENDIX
equity is permitted to be issued. We have
also proved that the financial market equi- 5.1. First-order condition (16) for
librium is constrained Pareto optimal. In each divition of the space F̃:
contrast, if the large investor does not take
into account the effect of the amount of For each possible set of F̃, we obtain the
security issued on her sales revenues, the corresponding value of the objective
competitive equilibrium allocation is not function of (5) and the corresponding
150 EQUITY OWNERSHIP STRUCTURE AND CONTROL
expression for the first-order condition 5.2. Optimizing conditions for the
(16): Objective function of (5) in the case problem (6):
of E min (F̃, X̃) p:
For this analysis, we need to respecify the
(A) F̃ 僆 (A): objective function L1 (F̃, qL) of the problem
RL(m, F̃, qL) [(Xg Fg)g(m) (6) for each possible case generated from
(Xb Fb)b(m)] the restriction of F̃. Substituting E min
(1 qL)[Fgg(m) (F̃, X̃) for p in (A1A)–(A1D) yields:
Fbb(m) pqL cm,
(A1A) (A) F̃ 僆 (A):
(B) F̃ 僆 (B): ˜
L1 (F, qL) [Xgg(mA) Xbb (mA)]
A
(1 qL)[Fgg(m)
qL(1 )[Xg g(mB)
Xbb(m)] pqL cm,
(A1C) Fbb(mB)] cmB,
(D) F̃ 僆 (D): (A3B)
RL(m, F̃, qL) (1 qL)[Xgg(m)
Xbb(m)] pqL cm. (C) F̃ 僆 (C):
(A1D)
˜
L1 (F, qL) [Xgg(mC) Xbb (mC)]
C
[Xgg(mD) Xbb(mD)]
(C) F̃ 僆 (C): cmD, (A3D)
[Xg Xb qL)(Fg Xb)]’g(m) c,
(A2C) where mA, mB, mC, and mD denote the
monitoring levels that satisfy the optimality
(D) F̃ 僆 (D): conditions (A2A), (A2B), (A2C), and
(A2D), respectively.
(1 qL)(Xg Xb)’g(m) c. (A2D) Now, for each possible set of F̃, we solve
the following constrained maximization
We should again notice that the large investor problem:
takes the security price as given when solving
the monitoring problem in period 1.
In Subsection 3.3, we will use (A1) and
i
L0 max
F̃僆 (i),qL僆[0,1]
i
L1(F̃, qL),
(A2) to solve the problem (6) faced by the
large investor in period 0. i A, B, C, D. (A4)
SECURITY DESIGN AND INSIDER MONITORING 151
(A) F̃ 僆 (A):
(mC)
mC
Fg g0 g1 0, (A7a)
qL (1 )g(mA) (Fg Fb)
(1 )[Fgg(mC) Xbb(mC)]
mC
g’ (mA)
mA
Fg g0 g1 0, (A5a)
qL(Fg Xb) ’g (mC)
qL
0,
(A7b)
qL (1 )b(mA) (Fg Fb) where g0, g1, and ς are the nonnegative
multipliers associated with Fg Xb, Xg
Fg and 1 qL, respectively. The partial
’g (mA)
mA
Fb g0 b0 b1 0, derivatives ∂mC /∂Fg and ∂mC /∂qL are
constructed by totally differentiating
(A5b)
(A2C) with respect to mC, Fg, and qL. Note
(1 )[Fgg(mA) Fbb(mA)] that Fb is equal to Xb in this case.
m
qL(Fg Fb) ’g (mA) A 0, (D) F̃僆 (D):
qL
(A5c) (1 )[Xgg(mD) Xbb(mD)]
mD
where g0, g1, b0, b1, and ς are the non- qL(Xg Xb) ’g (mD) 0,
qL
negative multipliers associated with Fg
(A8)
Fb, Xg Fg, Fb 0, Xb Fb, and 1 qL,
respectively. The partial derivatives where ς is the nonnegative multiplier asso-
∂mA/∂Fg, ∂mA/∂Fb, and ∂mA/∂qL are derived ciated with 1 qL. The partial derivative
by totally differentiating (A2A) with ∂mD /∂qL is derived by totally differentiating
respect to mA, Fg, Fb, and qL. (A2D) with respect to mD and qL. Note that
Fg Xg and Fb Xb in this case.
(B) F̃ 僆 (B):
Proof of Proposition 1. We begin by
qL (1 )b(mB) (Xg Fb)’g evaluating the first-order conditions for
F̃ and qL, assuming F̃ 僆 (A), that is,
(A5a)–(A5c). To do so, we first need to
(mB)
mB
Fb
b0 b1 0, (A6a)
specify ∂mA/∂Fg, ∂mA/∂Fb, and ∂mA/∂qL,
which are obtained from (A2A). Totally
differentiating (A2A) with respect to mA,
(1 )[Xgg(mB) Fbb(mB)] Fg, Fb, and qL yields
mB
qL(Xg Fb) ’g (mB) 0, mA qLg(mA)
’
qL
< 0, (A9)
(A6b) Fg A
where b0, b1, and ς are the nonnegative mA qL’g(mA)
> 0, (A10)
multipliers associated with Fb 0, Xb Fb A
Fb, and 1 qL, respectively. The partial
mA (Fg Fb)g(mA)
’
derivatives ∂mB/∂Fb and ∂mB/∂qL are 0, (A11)
obtained by totally differentiating (A2B) qL A
with respect to mB, Fb, and qL. Note
that Fg is equal to Xg in this parameter where ΛA [Xg Xb qL(Fg Fb)]”g
constellation. (mA) 0. Note that ΛA is negative
152 EQUITY OWNERSHIP STRUCTURE AND CONTROL
because the second-order condition must be (A7b), we totally differentiate (A2C) with
satisfied.The signs of (A9)–(A11) then fol- respect to mC, Fg, and qL:
low from (2), 0 qL 1, and ’g(m) 0.
Since g0, b0, and b1 are the nonnegative mC qL’g(mC)
< 0, (A14)
multipliers associated with Fg Fb, Fb Fg ΛC
0, and Xb Fb, (A5b) and (A10) show that
g0 0 or b1 0 or both, that is, Fg Fb mC (Fg Xb)’g(mC)
0, (A15)
or Xb Fb or Fg Fb Xb. If g0 0 so qL ΛC
that Fg Fb, it is seen from (A5a) that
g1 0, which means Xg Fg Fb. This where ΛC [Xg Xb qL (Fg Xb)]
contradicts Fb Xb Xg for F̃ 僆 (A). ”g (mC) 0, which is negative from the
Thus, we must have b1 0 and Fb Xb. second-order condition. Suppose that
Furthermore, if Fg Fb Xb, it again Xb Fb Fg.Then, it is found from (A7a)
follows from (A5a) that g1 0, which that g1 0, that is, Xg Fg. Since this con-
implies Xg Fg. However, this contra- tradicts Fg Xb, we must have Xb Fb Fg
dicts Fg Fb Xb Xg. Therefore, we for F̃ 僆 (C). Given ∂mC/∂Fg 0 from (A14)
must have Xb Fb Fg and g1 0. and ∂mC/∂qL 0 from (A15) with Fg Xb,
Given ∂mA/∂ Fg 0 from (A9) and we can assume that Fg and qL are determined
∂mA/∂qL 0 from (A11) with Xb Fb as an interior solution by (A7a) and (A7b).
Fg, we can assume that Fg and qL are Thus, rearranging (A2C), (A7a), and (A7b)
determined as an interior solution by with Xb Fb and g0 g1 ς 0, we
(A5a) and (A5c). Thus, rearranging (A2A), have (18a)–(18e) if F̃ 僆 (C).
(A5a), and (A5c) with Xb Fb and Finally, we investigate the first-order
g0 g1 ς 0, we obtain (18a)–(18e) condition for qL when F̃ 僆 (D), that is,
if F̃ 僆 (A). (A8). We totally differentiate (A2D) with
We next discuss the first-order condi- respect to mD and qL, and see
tions for F̃ and qL, assuming F̃ 僆 (B),
that is, (A6a) and (A6b). To specify
mD (Xg Xb)g(mD)
’
∂mB /∂Fb and ∂mB /∂qL, we totally differen-
< 0. (A16)
tiate (A2B) with respect to mB, Fb, qL ΛD
and qL:
(A7a) that g1 0 in this case. Thus, we Combining (A18) and (A19) yields
see Fg* Xg even if F̃ is restricted by the set
(A) 傼 (C). This finding implies that the ,
{ [Xg Xb q**
L (1 )
optimal security is equity.
(Xg Xb)] ’g(m䉫) c}(m** m䉫)
We now return to the situation in which
’g(m) 0. Let us take the solution (Fg, Fb, L (1 )(Xg Xb)
q**
qL (Fg**, F** b , q L ) (Xg, Xb , q L ) as a
** **
[g (m䉫) g(m**)]
starting point.Then, consider a permutation q** (1 )[(Xg )
L
(Fg,䉫 Fb䉫, qL䉫) from the solution (Fg**, F** **
b , q L );
that is, (Fg,䉫 Fb䉫, qL䉫) (Xg – ,Xb, q**( 1 + )), g(m**) Xbb(m**)]
L
where Xg – Xb 0 and 0. Choose q**
L (1 ) g(m ), for m m .
** ** 䉫
m䉫 that satisfies
Since m䉫 satisfies (A17), this inequality
[Xg Xb qL(Fg 䉫 䉫
Xb)]’g(m䉫) c, reduces to
which means Γ ,
q**
L {(1 )(Xg
[Xg Xb qL** (1 )(Xg Xb)[g(m䉫) g(m**)]
Xb)]’g(m䉫) c. (A17) (1 )[(Xg )g(m**)
Xb b(m**)] (1 ) g(m**)},
Since (F䉫g, F䉫b) 僆 (A) 傽 (C), the large
investor’s payoff is represented by for m** m䉫. (A20)
A ˜䉫 䉫
L1(F , qL) [Xgg(m䉫) Xbg(m䉫)] Now, suppose that we can take a pair
(, ) 僆 {( , ) | Xg – Xb 0 and
q**
L (1 )(1 )[(Xg ) 0} which satisfies
g(m䉫) Xbb(m䉫)] cm䉫.
Xg Xb q**
L (1 )
Now, for B
L0 defined in Proposition 1, (Xg Xb) Xg Xb
q**
L (Xg Xb), (A21)
construct
[(Xg )g(m**) Xbb
Γ ,
A ˜䉫 䉫
L1(F , q L)
B
L0 (m**)] g(m**). (A22)
ABSTRACT
We preview empirical work by the European Corporate Governance Network on the size of
block shareholdings in Europe. The most salient finding is the extraordinarily high degree of
concentration of shareholder voting power in Continental Europe relative to the U.S.A. and the
U.K. Thus the relationship between large controlling shareholders and weak minority share-
holders is at least as important to understand as the more commonly studied interface between
management and dispersed shareholders.
80%
20%
Holding
Company (3) 2%
20% 50%
Listed
Company (4)
Table 6.1 Median size (%) of largest ultimate outside voting block for listed industrial companies
Number of
Country companies Median largest voting block
Austria 50 52.0
Belgium 121 50.6
– BEL20 20 45.1
Germany 374 52.1
– DAX30 30 11.0
Spain 193 34.2
France – CAC40 40 20.0
Italy 216 54.53
The Netherlands 137 43.5
United Kingdoma 250 9.9
United States – NYSE 1309 0b
– NASDAQ 2831 0b
Sources: Austria: Gugler et al. (1999); Belgium: Becht et al. (1999); France: Bloch and Kremp (1999);
Germany: Becht and Böhmer (1999); Italy: Bianchi et al. (1999); Spain: Crespí-Cladera and Garcia-Cestona
(1999); The Netherlands: De Jong et al. (1999); UK: Goergen and Renneboog (1999); USA: Becht (1999a).
a Random sample of 250 companies.
b More precisely, below the 5% disclosure threshold.
at the other extreme, Germany, Austria and voting power concentration is inversely
Italy all exceed 50%. The figures for the related to company size: blocks held in
German DAX30 are somewhat atypical smaller companies tend to be much larger
because there is a minimum turnover on average, see some of the country reports
requirement for membership of the forthcoming in Barca and Becht (1999).
DAX30, leading these companies to widen The numbers mask strong differences in
their shareholder base in order to ensure the way the large voting stakes are built up,
their continued inclusion in the index. which are discussed more fully in the various
Similarly, the French CAC40 is an unrepre- country reports. For example, in Italy,
sentative group of the largest and most liq- France and Belgium, pyramidal holding
uid companies on the exchange. In general, company groups are a preferred method of
160 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 6.2 Voting power concentration, percentage of companies for which largest voting power
stake lies in (in and below) various ranges
[x, y] (%) Cum (%) Cum (%) Cum (%) Cum (%) Cum
(%) (%) (%) (%) (%)
amassing voting power without concentrating and of course the 50% majority threshold.
(cash flow) ownership. In the Netherlands, Such thresholds differ from country to
many companies use a device called an country and are reflected in differences in
Administratiekantoor, a foundation controlled clustering of voting blocks.
by company insiders that owns all or most of
the voting stock and distributes non-voting
share certificates to ordinary shareholders; 4. CONCLUSION
as well as a defensive arsenal of preference
or priority shares, with large blocks of An immediate very clear conclusion from
potential voting power triggered by impor- the EGCN’s work on blockholdings in
tant control-relevant events. In Germany, Europe is that voting power is highly
voting pacts are a commonly used device for concentrated in Continental Europe. Our
concentrating voting power, together with survey, which was purely descriptive, raises
voting caps and the default voting power of more questions than it answers. Why is
banks where shares are placed in custody (a there this strong preference for voting
source of voting power excluded from the power, and to what extent is it matched by
figures in Tables 6.1 and 6.2). concentration in cash flow ownership
Figures 6.2 and 6.3, taken from the stakes (which, unfortunately, need not be
ECGN U.K. and Germany Country Reports, disclosed under the EU transparency
respectively, illustrate how countries’ legal directive)? Becht (1999b) provides an
environment affects the pattern of voting analysis of the trade-offs involved. Is the
power concentration. Thus is the U.K. the current level of disclosure of voting and
Takeover Panel’s mandatory bid rule that ownership stakes sufficient or excessive?
prescribes a general cash offer to all share- Can EU policy on these matters do better
holders whenever a blockholder crosses the in ensuring the smooth functioning of capi-
30% threshold, has effectively ensured that tal markets: in the words of U.S. Supreme
the growth of large blocks stops short of Court justice Louis Brandeis, is trans-
30%. For the 250 companies examined, all parency the best disinfectant? And Bianco
blocks over 30% were held by founders and and Casavola (1999) make progress
their families. In Germany, important towards tackling the most fundamental
thresholds induce clusters of holdings: the question: what are the effects of corporate
10% anti-director threshold at which share- governance variables such as (pyramidal)
holders can take to court board members, ownership concentration on company per-
the 25% blocking minority and 75% super formance?3 There is enormous scope for
majority for changes to company statutes, further empirical work on these issues.
BLOCKHOLDINGS IN EUROPE 161
100
Maximum Outside Voting Block %
75
50
25
0
0 .25 .5 .75 1
Fraction of the data
100
Maximum Outside Voting Block %
75
50
25
0
0 .25 .5 .75 1
Fraction of the data
ABSTRACT
We analyze the ultimate ownership and control of 5,232 corporations in 13 Western European
countries.Typically firms are widely held (36.93%) or family controlled (44.29%). Widely held
firms are more important in the UK and Ireland, family controlled firms in continental Europe.
Financial and large firms are more likely widely held, while non-financial and small firms are
more likely family controlled. State control is important for larger firms in certain countries.
Dual class shares and pyramids enhance the control of the largest shareholders, but overall
there are significant discrepancies between ownership and control in only a few countries.
Western European firms are most likely (1999), in that we find fewer State-controlled
to be widely held (36.93%) or family firms and more widely held firms, fewer
controlled (44.29%). Widely held firms are pyramids, and more dual class shares.
especially important in the UK and Ireland, Compared to the findings of Claessens et al.
while family control is more important in (2000) for East Asia, we find that families
continental Europe. Widely held firms are control a higher proportion of firms; each
more important for financial and large family controls fewer firms on average; top
firms, while families are more important families control a lower proportion of total
for non financial and small firms. In certain stock market capitalization;1 a higher
countries, widely held financial institutions proportion of family controlled companies
also control a significant proportion of have family members in top management;
firms, especially financial firms. In some and the largest shareholder is less often
countries of continental Europe, the State alone, but averages much higher cash-flow
also controls a significant proportion of rights, control rights, and ratio of cash-flow to
firms, especially the largest. Widely held voting rights. These differences may be due
corporations control few firms. to weaker law enforcement in Asia that
We report the use of multiple classes of allows controlling owners to achieve
shares, pyramidal structures, holdings effective control of a large number of firms
through multiple control chains, and cross- by controlling and owning a smaller part of
holdings, which are devices that give the each firm.
controlling shareholders control rights in Section 2 describes the data. Section 3
excess of their cash-flow rights. Pyramiding discusses ultimate ownership patterns.
occurs when the controlling shareholder Section 4 discusses the means whereby
owns one corporation through another owners gain control rights in excess of own-
which he does not totally own. Firm Y is ership rights and Section 5 measures the
held through “multiple control chains” if it extent to which this has been achieved.
has an ultimate owner who controls it via a Section 6 presents conclusions.
multitude of control chains, each of which
includes at least 5% of the voting rights at
each link. “Cross-holdings” means com- 2. DATA
pany Y directly or indirectly controls its
own stocks. Dual class shares are used by Some of the previous studies of corporate
few firms in Belgium, Portugal, and Spain, ownership and control, such as Lins and
but by 66.07%, 51.17%, and 41.35% of Servaes (1999a, b), rely primarily on
firms in Sweden, Switzerland, and Italy. Worldscope. However, we find its coverage
Pyramids and holdings through multiple inadequate. For example, Worldscope
control chains are used to control only includes only 176 of 632 Spanish listed
19.13% and 5.52% of listed firms firms at the end of 1997. In this case, we
respectively, being less important for family instead rely upon the Spanish Stock
controlled firms and more important for Exchange regulatory authority’s files
firms controlled by the State and by widely (Comision Nacional del Mercado de
held financial institutions. 53.99% of Valores, 1998) which provides quarterly
European firms have only one controlling information on all shareholders with at
owner. More than two-thirds of the family least 5% of control rights, as well as direc-
controlled firms have top managers from tors’ ownership for all listed firms.
the controlling family. Overall, we find a Moreover, ownership data is sometimes
substantial discrepancy between ownership missing: in this case, Worldscope reports
and control in Belgium, Italy, Norway, zero ownership stakes. To ensure accuracy,
Sweden, and Switzerland, but much less we include only countries for which we can
elsewhere. obtain alternative sources (especially pri-
Our results for the 20 largest firms differ mary or official) to permit cross-checking.
slightly from those of La Porta et al. We do not rely on Worldscope if we have
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 165
an official data source (i.e., the Stock stake above 2%: Compart with a stake of
Exchange ownership files). When official 33.45% and Mediobanca with a stake of
data sources are not available, we collect 3.77%. Compart is indicated in the Italian
data from alternative sources. We use supervisory authority’s files as the “ultimate”
Worldscope for ownership data only when owner of Montedison. However, we found
information for a specific firm can not that Compart has three shareholders with
otherwise be identified. Cross-checking stakes above 10%: Credit (11.01%),
excludes Luxembourg, Greece,2 Denmark3 Cassa di Risparmio di Roma (10.14%),
and Holland4 leaving comprehensive, and Mediobanca (15.26%). According to
reliable ownership data for 13 Western our definition, Compart is the ultimate con-
European countries: Austria, Belgium, trolling shareholder of Montedison at the
Finland, France, Germany, Ireland, Italy, 20% threshold. However, at the 10%
Norway, Portugal, Spain, Sweden, threshold, Mediobanca would be the largest
Switzerland, and the UK. For these ultimate owner of Montedison, with a
countries, Table 7.1 lists the data sources 15.26% 3.77% 19.03% control
for ownership and multiple classes of stake. Furthermore, mechanisms used to
shares. Data limitations confined our sam- secure control rights in excess of ownership
ple to the period from 1996 to the end of rights are not systematically analyzed.
1999.5 This is not a significant restriction Over the sample period, a total of 5,547
since ownership structures tend to be rather firms were listed in these 13 countries. Of
stable, as noted in La Porta et al. (1999). these, we excluded 167 firms whose owner-
The European Corporate Governance ship was not recorded, 61 that use nominee
Network (ECGN), too, has sponsored sev- accounts (mostly in the UK and Ireland,
eral studies on ownership structures within where firms are not required to disclose the
the European Union.6 However, compliance identity of their “true” owners) and 87 affil-
with the European Union directive on large iates of foreign firms (i.e., a foreign firm
shareholdings (88/627/EEC) restricts controls at least 50% of their voting rights)
meaningful cross-country analysis with where we could not follow their ownership
non-European Union countries. In particu- chain. We retain the affiliates of foreign
lar, the ownership measures documented in firms in our database whenever the holding
those studies represent neither ultimate firm is included in our sample. This screen-
ownership nor ultimate control stakes. ing left 5,232 firms comprising 94.32% of
Specifically, they do not consider the use of the listed firms in the 13 countries.7 For
multiple voting shares, and simply add up these firms, our database records all owners
direct and indirect control stakes without who control at least 5% of voting rights. In
tracing them to the ultimate owners. The France, Germany, and Spain, such owners
controlling owner is defined as the one who must disclose their identity. The disclosure
controls an absolute majority (i.e., over threshold is 2% in Italy and 3% in the UK.
50%) of voting rights, or holds enough vot- The difficulty of organizing dispersed
ing rights to have de facto control. For shareholders means that if the largest
example, French regulation defines de facto shareholder holds a substantial block
control as occurring when a person or a of shares, then that shareholder has effec-
legal entity owns, directly or indirectly, tive control. In line with earlier studies, we
more than 40% of voting rights and no shall assume that 20% of the voting shares
other partners or shareholders own a suffices to ensure control. We shall also
higher percentage directly or indirectly discuss some cases that assume a control
(Bloch and Kremp, 1998). This corre- threshold of 10%. If no shareholder
sponds to a control threshold of 40%. To exceeds a given control threshold, then the
illustrate the bias that this definition firm is said to be widely held at that thresh-
introduces, consider the ultimate control old.Table 7.2 sets out the screening process
structure of Montedison (Italy). and lists the samples of firms analyzed in
Montedison has two shareholders with a later tables.
Table 7.1 List of country specific data sources used to collect data on direct ownership (i.e., names of large shareholders, and stakes held in the company),
and dual class share structures
Countries Direct ownership data Dual class shares
Austria Wiener Börse (2000), “Yearbook 1999” Wiener Börse (2000), “Yearbook 1999”
166 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Worldscope (1998)
Belgium Brussels Stock Exchange (www.bxs.be) Datastream
Worldscope (1998) Brussels Stock Exchange
Finland Helsinki Media Blue Book, “Major Finnish Firms Internet Database” Datastream
(http://www.bluebook.fi/en/tuotteet/haku/majorfinnishfirms.html) Helsinki Media Blue Book, “Major Finnish Firms Internet
Hugin, Annual Report CD (1998) (http://www.huginonline.com) Database”
France The Herald Tribune (1997), “French Firm Handbook 1997,” Datastream
SFB-Paris Bourse Financial Times (1997), “Extel Financial”
Financial Times (1997), “Extel Financial” Les Echos (1996)
Worldscope (1998) Muus (1998)
http://www.bourse-de-paris.fr/fr/market8/fsg830.htm
Germany Commerzbank (1997), “Wer gehört zu wem” Datastream Die Welt (1996),
(http://www.commerzbank.com/navigate/date_frm.htm) available at: http://www.welt.de/
Financial Times (1997), “Extel Financial” Financial Times (1997), “Extel Financial”
Worldscope (1998) Becht and Boehmer (1998)
Ireland http://www.hemscott.com/equities/firm/ Datastream Worldscope (1998)
Worldscope (1998)
Italy CONSOB (1997), “Bollettino – edizione speciale n. 4/97 – Compagine Datastream
azionaria delle società quotate in borsa o ammesse alle negoziazioni nel Il Sole 24 ore (1997), “Il taccuino dell’azionista”
mercato ristretto al 31 dicembre 1996”
(http://www.consob.it/trasparenza_soc_quot/trasp_soc_quot.htm)
Il Sole 24 ore (1997), “Il taccuino dell’azionista”
Norway Hugin, Annual Report CD (1998) Datastream
Worldscope (1998) Hugin, Annual Report CD (1998)
Portugal Bolsa de Valores de Lisboa, “Sociedades Cotadas 1997” (1998) Bolsa de Valores de Lisboa,
“Sociedades Cotadas 1997”
Spain Comision Nacional del Mercado de Valores (1998), ABC (1996)
“Participaciones significativas en sociedades cotizadas” Financial Times (1997), “Extel Financial”
(http://www.cnmv.es/english/cnmve.htm) Crespi-Cladera and Garcia-Cestona (1998)
Sweden Hugin, Annual Report CD (1998) Datastream
Worldscope (1998) Hugin, Annual Report CD (1998)
Switzerland Union Bank of Switzerland (1998), “Swiss Stock Guide 96/97,” Union Bank of Switzerland (1998),
Zurich “Swiss Stock Guide 96/97,” Zurich
United Financial Times (1997), “Extel Financial” Datastream
Kingdom London Stock Exchange (1997), “The London Stock Exchange Financial Times (1997), “Extel Financial”
Yearbook” Financial Times (1996)
Financial Times
Worldscope (1998)
http://www.hemscott.com/equities/firm/
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 167
168 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 7.2 Screenings used to identify and remove corporations without reliable ownership data,
and sub-samples of firms used in the different sections of the paper. Panel A describes
the procedure used to eliminate companies with insufficient or unreliable data on
ownership, due to unavailable data, or related to the impossibility to trace ownership
back to the ultimate owner (i.e., when the direct controlling shareholder is from a coun-
try for which we do not have ownership data, or when the direct owner is a nominee).
Panel B reports the total number of firms remaining after this screening, as well as the
number of firms with an ultimate owner at the 5% and 20% cutoffs. It also lists the
different subsets of firms used in the various sections of the paper
The exclusion of firms that use nominee by each owner at the 10% threshold, but
accounts may overstate the proportion of family controlled at the 20% threshold.
widely held firms in our sample. However,
nominee accounts are the largest share- 2.1. Calculation of cash-flow rights
holders in only a small proportion of firms
and control rights
(less than 5%), so any bias is likely to be
marginal. Ireland and the UK have the Corporate ownership is measured by
highest proportion of nominee accounts so cash-flow rights, and control is measured
this bias would be strongest there. These by voting rights. Ownership and control
countries also have the highest proportions rights can differ because corporations can
of widely held firms, so the exclusion of issue different classes of shares that provide
firms using nominee accounts is unlikely to different voting rights for given cash-flow
distort our cross-country comparisons. rights. Ownership and control rights can
A shareholder of a corporation is said to also differ because of pyramiding and hold-
be an ultimate owner at a given threshold if ings through multiple control chains.
he controls it via a control chain whose links Firm Y is said to be controlled through
all exceed that threshold. If a firm has two pyramiding if it has an ultimate owner, who
owners with 12% of control rights each, then controls Y indirectly through another corpo-
we say that the firm is half controlled by ration that it does not wholly control. For
each owner at the 10% threshold, but that example, if a family owns 15% of Firm
the firm is widely held at the 20% threshold. X, that owns 20% of Firm Y, then Y is
In the case of a firm with two owners (a fam- controlled through a pyramid at the 10%
ily with 20% of control rights and a widely threshold. However, at the 20% threshold,
held corporation with 19% of control rights) we would say that Firm Y is directly con-
we would say that this firm is half controlled trolled by Firm X (which is widely held at the
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 169
family, which has a 100% O&C stake. At the 20% level, financial firms control
Another case is Heidelberger Zement AG, 4.9% of unlisted firms, the State 2.67%,
whose largest owner (with a 19.07% O and cross-holdings 1.83%. Thus, both
stake and a 21.8% C stake) is Schwenk State and widely held financial firms are
Gmbh, again an unlisted firm. We find that insignificant as ultimate owners of unlisted
Schwenk is 90% owned and controlled by firms. Furthermore, since the State and the
the Schwenk family, and 10% owned by the largest financial institutions are included in
Babette family. In some cases, the owner- the Wer gehört zu wem database, we were
ship structure of unlisted firms is more able to identify them and trace their
complex. For example, the direct owner of ownership chain. This further reduces the
Thyssen AG is Thyssen Beteiligunsverw possibility of bias in our sample.
Gmbh, which is unlisted. We find that this For Italy, Bianchi et al. (1998) consid-
firm is 49.999% owned and controlled by ered a sample of 1,000 manufacturing
Commerzbank (a listed, widely held finan- firms surveyed by the Bank of Italy. They
cial firm), and 49.981% owned and con- reported that the largest immediate share-
trolled by Allianz. A number of complex holder of these firms held, on average, a
cases of ownership of unlisted firms were direct stake of 67.69%. Among all types of
also reported by La Porta et al. (1999). immediate shareholders, individuals held
Our database records the unlisted sub- 48% of equity, non financial firms 36.9%,
sidiaries of widely held corporation or financial firms 0.17%, the State 4.6%,
financial institution, so any listed firm con- and foreign investors 8%. We secured a
trolled by an unlisted firm that is controlled wider sample of 3,800 unlisted Italian
by a widely held corporation or financial firms with ultimate owners from the AIDA
institution is recorded as controlled by the database.8 We found that the State is the
latter in our database. Thus, an unlisted largest owner for 0.4% of firms, financial
firm that we identify as the ultimate con- institutions for 0.2%, and families (domes-
troller of a listed firm is unlikely to be, tic and foreign) for 99.4%. The average
in fact, controlled by a widely held corpo- ultimate control stake of the largest
ration or financial institution. The unlisted controlling shareholder was 70.71%.
firm isn’t likely controlled by the State For France, Bloch and Kremp (1998)
because State-controlled firms tend to be summarize the ownership structure of a
listed. In any case, State ownership has sample of 282,322 (mostly unlisted)
decreased dramatically in Europe after the firms.9 For firms with more than 500
privatization wave of the 1990s. The low employees, the largest owner held, on aver-
likelihood that an unlisted firm is, in fact, age, 88% of the capital. For 56% of the
controlled by a widely held corporation, unlisted firms, the largest owner was a
widely held financial institution, or the family; for the remaining 44% it was
State leaves families as the most likely another corporation, usually an unlisted
controller of an unlisted firm. firm. For the UK, Goergen and Renneboog
For Germany, we collected a sample of (1998) reported that 78% of unlisted firms
500 unlisted firms with ultimate owners are fully controlled by one shareholder, while
from Wer gehört zu wem. We considered the remaining 22% have a shareholder who
firms in alphabetical order, including a firm (directly) holds a majority stake.10
in our sample if we could trace its ultimate
owners. We stopped when our sample 2.3. Examples
numbered 500. We found that the average
control stake is 89.44%. In 68% of Figure 7.1 illustrates dual class shares and
the cases, the largest owner is the sole pyramiding in the Nordström family group
owner; in the remainder, the largest owner of Sweden. All holdings of more than 5% of
holds an average stake of 67%. Families, a firm’s shares are listed. All firms shown in
both domestic and foreign, are the largest Figure 7.1 have dual class shares: A-shares
ultimate owners of 90.6% of these firms. carry one voting right while B-shares
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 171
Humlegården Columna
Sweco AB Fastigheter AB
Fastigheter AB
(16.2%O; (26.1%O; (35%O;
23.7%C) 17.7%C) 48.1%C)
(11.3%O;
Latour
5.3%C) (2%O; Realia
(2%O; 9.5%C) (2.66%O; 39.3%C)
Robur 9.5%C)
Minne family (13.7%O; (1.5%O;
Kapitalf. 9%C) 8.7%C)
Figure 7.1 The Nordström family group (Sweden). This figure describes the major listed
firms controlled by the Nordström family. All control stakes of at least 5%
are reported. All the firms have dual class shares with A-shares each carrying
one voting right and B-shares each carrying a tenth of a vote. The procedure
used to compute direct ownership and control rights is illustrated in Panel B.
Ownership stakes are denoted by “O” and voting stakes by “C.” Direct
ownership and control stakes are shown alongside each of the arrows, while
the resulting ultimate ownership and control stakes for Realia (the only case
of pyramiding) by the Nordström family are reported in the firm’s box.
carrying one-tenth of a voting right. Realia carry 4.292 million votes ( one-tenth of
has two classes of shares: 2.641 million 42.922 million). Thus, A-shares carry
A-shares and 42.922 million B-shares. 38.09% of the votes ( 2.641 million/
A- and B-shares have the same face value, (2.641 million 4.292 million)), while
so the total stock capital is 45.563 million B-shares carry 61.91% of votes.
shares. A-shares constitute 5.8% of stock Realia has three direct shareholders:
capital ( 2.641 million/45.563 million) Columna Fastigheter, Lingfield Investment,
while B-shares constitute 94.2%. A-shares and the Eriksson family. Columna owns
carry 2.641 million votes, while B-shares 1.933 million A-shares and 14.007 million
172 EQUITY OWNERSHIP STRUCTURE AND CONTROL
B-shares. Thus, Columna owns 35% chains (Ifi-Ifil and Ifi-Unicem, denoted by
( (1.933 million 14.007 million)/ dotted lines in Figure 7.2).
45.563 million) of cash-flow rights, and To compute the Agnelli family’s ultimate
controls 48.1% ( (1.933 million ownership of Unicem, we form the
1.401 million)/ (2.641 million 4.292 product of its ownership stakes along
million)) of votes in Realia. Realia’s second the pyramidal chain of Agnelli-Ifi-
largest direct shareholder, Lingfield Carfin-Ifil-Unicem, namely 100% *
Investments, owns no A-shares and only 41.23% * 100% * 20.55% * 8.76%
6.259 million B-shares, which comprise 0.7422%. We then add the stake from the
13.7% (6.259 million/ 45.563 million) of two further (i.e., multiple) control chains
the cash-flow rights, and 9% (0.6259/(2.641 (the dotted lines).The first chain is from Ifi
million 4.292 million)) of the control into Ifil (the left side of Figure 7.2) which
rights. Finally, the Eriksson family owns gives the Agnelli family a stake of
0.591 million A-shares and 0.101 million 100% * 41.23% * 7.97% * 8.76%
B-shares, comprising 1.5% ( (0.591 0.2878%. The second chain is from Ifi to
million 0.101 million)/ 45.563 million) of Unicem (the right side of Figure 7.2) which
the ownership rights and 8.7% ( (0.591 gives the Agnelli family a stake of
million 0.0101 million)/ (2.641 million 100% * 41.23% * 19.42% 8.0069%.
4.292 million)) of the control rights.Through The family’s overall ownership stake in
Columna Fastigheter, the Nordström family Unicem is the sum of these three
has sole control of Realia at the 20% stakes: 0.7422% 0.2878% 8.0069%
threshold. However, at the 10% threshold, 9.0369%.
Realia has a second large shareholder, The Agnelli family’s ultimate control
Blockfield Properties. In Relia’s control stake is the sum of the weakest links along
structure there are two pyramiding chains: each control chain. The weakest link in the
Nordström family/Columna/Realia and pyramidal chain is min(100%; 82.45%;
Blockfield/Columna/Realia. 100%; 37.64%; 14.81%) 14.81%. In
Figure 7.2 illustrates the control of the control chain from Ifi to Ifil, the weak-
Unicem by pyramiding, holdings through est link is min(100%; 82.45%; 14.16%;
multiple control chains, and dual class 0%) 0%. Note that Ifil controls 14.81%
shares within the Agnelli family group, the of Unicem; this 14.81% stake has already
largest Italian group. The methodology been considered in the pyramidal chain, so
presented in Figure 7.1 is used to compute we use 0%. In the control chain from Ifi to
cash-flow rights (O) and control rights (C), Unicem, the weakest link is min(100%;
taking account of dual class shares. 82.45%; 32.83%) 32.83%. The Agnelli
Unicem is directly controlled by two major family’s control stake in Unicem is the sum
shareholders: Ifi and Ifil. Ifil is controlled of these three weakest links, namely
by Ifi with a direct stake (O 7.97%; 14.81% 0% 32.83% 47.64%.
C 14.6%) and an indirect stake The ratio of cash-flow to voting rights is
(O 20.55%; C 37.64%) through 0.1897 (9.04%/47.64%).
Carfin, a wholly owned, non financial We conclude that Unicem is controlled
unlisted firm. Since Carfin is wholly con- by the Agnelli family at both the 10% and
trolled by Ifi, we consider Ifi’s stake in Ifil the 20% thresholds. The company has only
as a direct holding rather than a pyramid, one controlling shareholder, and its control
although we say there is holdings through structure includes pyramiding, multiple
multiple control chain (see section 2.1). Ifi control chains, and multiple class shares.
is controlled by a single major shareholder,
Giovanni Agnelli & C. S.p.A. (the Agnelli
family). The Agnelli family’s control of 3. ULTIMATE OWNERSHIP PATTERNS
Unicem is thus exercised through pyramid-
ing (Ifi-Carfin-Ifil-Unicem), non voting Table 7.3 analyzes the ultimate controlling
shares (within Ifi, Ifil, and Unicem), and owners of Western European corporations
two holdings through multiple control at the 20% threshold. In all countries,
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 173
Agnelli Family
(100% O&C)
(41.23%O; 82.45%C)
IFI S.p.A.
(41.23%O; 82.45%C)
(100% O&C)
Carfin S.p.A.
(41.23%O; 82.45%C )
(20.55%O; 37.64%C)
Figure 7.2 Unicem (Italy). This figure describes the principal shareholders of Unicem, a complex
case of pyramiding, holdings through multiple control chains, and the use of dual class
shares within the Agnelli family group, the largest Italian group. All control stakes of
at least 5% are reported.The company has only one ultimate owner, the Agnelli family,
which controls 47.64% of the voting rights and owns 9.04% of the cash-flow rights in
Unicem. Hard lines indicate pyramiding while dotted lines indicate holdings through fur-
ther (i.e., multiple) control chains. Ownership stakes are denoted by “O” and voting
stakes by “C.” Direct ownership and control stakes are shown alongside each arrow,
while the Agnelli family’s ultimate ownership and control stakes (when different from
the direct stakes) are shown in each firm’s box.
widely held and family controlled firms are percentages are in the UK (23.68%) and
the most important category. 36.93% of Ireland (24.63%); in continental Europe,
the firms in our sample are widely held and the lowest percentages are in Norway
44.29% are family controlled. However, (38.55%), Sweden (46.94%),
there is a sharp cleavage between Switzerland (48.13) and Finland
ownership patterns in continental Europe (48.84%). In every other Western
and in the UK and Ireland. Widely held European country, family controlled firms
firms comprise 63.08% of UK firms and are in the majority.
62.32% of Irish firms; in continental The UK and Ireland also stand apart in
Europe the highest percentages of widely the low percentages of corporations that
held firms are all in Scandinavia but are are State controlled (0.08% and 1.45%).
substantially lower (Sweden 39.18%, In continental Europe, the State controls
Norway 36.77%, Finland 28.68%). The more than 10% of the listed firms in
lowest percentages of widely held firms are Austria, Finland, Italy, and Norway.
in Germany (10.37%), Austria (11.11%), 9.03% of Western European firms are
and Italy (12.98%). The picture for family controlled by widely held financial institu-
control is reversed. Here, the lowest tions, this being especially significant in
Table 7.3 Ultimate control of publicly traded firms. Data relating to 5,232 publicly traded corporations are used to construct this table.The table presents
the percentage of firms controlled by different controlling owners at the 20% threshold. Controlling shareholders are classified into six types.
Family: A family (including an individual) or a firm that is unlisted on any stock exchange. Widely held financial institution: A financial firm
(SIC 6000-6999) that is widely held at the control threshold. State: A national government (domestic or foreign), local authority (county,
municipality, etc.), or government agency. Widely held corporation: A non financial firm, widely held at the control threshold. Cross-holdings:The
firm Y is controlled by another firm, that is controlled by Y, or directly controls at least 20% of its own stocks. Miscellaneous: Charities, voting
trusts, employees, cooperatives, or minority foreign investors. Companies that do not have a shareholder controlling at least 20% of votes are
classified as widely held
Family of which:
Number Widely Identified Unlisted Widely held Widely held
Country of Firm held Family families firms State Corporation Financial Miscellaneous Cross-holdings
Austria 99 11.11 52.86 12.12 40.74 15.32 0.00 8.59 11.11 1.01
Belgium 130 20.00 51.54 7.31 44.23 2.31 0.77 12.69 12.69 0.00
Finland 129 28.68 48.84 16.28 32.56 15.76 1.55 0.65 4.52 0.00
174 EQUITY OWNERSHIP STRUCTURE AND CONTROL
France 607 14.00 64.82 26.11 38.71 5.11 3.79 11.37 0.91 0.00
Germany 704 10.37 64.62 27.03 37.59 6.30 3.65 9.07 3.37 2.62
Ireland 69 62.32 24.63 13.04 11.59 1.45 2.17 4.35 5.07 0.00
Italy 208 12.98 59.61 39.50 20.11 10.34 2.88 12.26 1.20 0.72
Norway 155 36.77 38.55 10.59 27.96 13.09 0.32 4.46 4.54 2.27
Portugal 87 21.84 60.34 5.17 55.17 5.75 0.57 4.60 6.90 0.00
Spain 632 26.42 55.79 6.25 49.54 4.11 1.64 11.51 0.47 0.05
Sweden 245 39.18 46.94 22.65 24.29 4.90 0.00 2.86 5.71 0.41
Switzerland 214 27.57 48.13 22.66 25.47 7.32 1.09 9.35 6.31 0.23
UK 1,953 63.08 23.68 12.22 11.46 0.08 0.76 8.94 3.46 0.00
Total 5,232 36.93 44.29 16.93 27.36 4.14 1.68 9.03 3.43 0.51
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 175
Belgium (12.69%) and Italy (12.26%) firms will affect its results. Financial firms
but insignificant in Finland (0.65%). constitute only 10.1% of sample firms in
Control by widely held corporations is Finland and 14.5% in Germany, but
trivial (1.68%) in all sample countries. 30.3% in Austria and 39.2% in Belgium.
Cross-holdings are only marginally significant
in Norway, Germany, and Austria (accounting 3.2. Size effects
for 2.27%, 2.62%, and 1.01% of cases
respectively) but are insignificant elsewhere. To explore the relationship between owner-
If we lower the threshold to 10%, then ship patterns and firm size, we identify in
the proportion of widely held firms falls each country the 20 largest, the middle 50,
to 13.72%; family control increases from and the 50 smallest firms by market capi-
44.29% to 55.87%; control by financial talization. For Austria, Ireland, and
institutions also increases from 8.73% to Portugal which had less than 120 listed
18.34%.This increase is especially signif- firms, we defined the 20 largest firms as
icant in the UK and Ireland. By contrast, “Large”, then divided the rest equally into
lowering the control threshold to 10% those that were “Medium” and “Small.”
has little effect on the percentages of Table 7.5 shows that family ownership is
State-controlled firms, firms controlled less likely for larger firms, being particularly
by widely held firms, and firms with cross- weak among the largest firms in the UK
holdings. and Sweden. In Austria, Finland, Italy,
Norway, and Portugal, State control is
quite important for the largest firms. Large
3.1. Financial versus non financial firms are also more likely to be widely held
firms than small firms in all our sample countries
Table 7.4 compares the ownership struc- except Austria, Norway, and Portugal. In
ture of financial firms and non-financial the UK, 90% of large firms are widely held
firms. Financial firms are slightly more at the 20% threshold, but only 14% of
likely to be widely held than non-financial small firms are widely held at the 10%
firms (39.92% versus 36.39%), are much threshold. Cross-country differences
more likely to be controlled by a widely become less significant among small firms.
held financial institution (21.48% versus
5.96%), and are much less likely to be
family controlled (26.54% versus 4. MEANS OF ENHANCING CONTROL
48.15%). For other types of controlling
owners, the differences between financial
This section reports the major mechanisms
and non financial firms are insignificant.
used to enhance control. We neglect less
In both the UK and Ireland, the proportion
significant mechanisms such as firm-
of widely held financial firms is about the
specific voting caps,11 golden shares,12
same as that of non financial firms. informal alliances (i.e., voting blocks), or
Financial firms are more likely than transfer restrictions on shares.
non-financial firms to be widely held for
continental countries except Austria,
Finland, and Portugal. Non-financial firms 4.1. Dual class share structures
are more likely to be family controlled in
all countries except Finland and Portugal. Consistent with the hypothesis that control
Differences in patterns of ownership and provides large private benefits (see,
control across countries can be explained for example, Barclay and Holderness,
by differences in the regulations on finan- 1989; Claessens et al., 2002; Jensen and
cial firms (Barth et al., 2000). Since the Meckling, 1976; Johnson et al., 2000;
ownership structure of financial firms dif- La Porta et al., 1997, 2002) in Western
fers from that of non financial firms, a Europe, several studies report that voting
country’s ratio of financial to non financial shares trade at a premium over non voting
Table 7.4 Concentration of control: financial versus non financial firms. Data relating to 5,232 publicly traded corporations are used to construct this
table, which groups corporations into financial firms (SIC: 6000-6999) and non financial corporations. The table presents the percentage of
firms controlled by different controlling owners at the 20% threshold. Controlling shareholders are classified into six types. Family: A family
(including an individual) or a firm that is unlisted on any stock exchange. Widely held financial institution: A financial firm (SIC 6000-6999)
that is widely held at the control threshold. State: A national government (domestic or foreign), local authority (county, municipality, etc.), or
government agency. Widely held corporation: A non financial firm, widely held at the control threshold. Cross-holdings:The firm Y is controlled by
another firm, that is controlled by Y, or directly controls at least 20% of its own stocks. Miscellaneous: Charities, voting trusts, employees, coop-
eratives, or minority foreign investors. Companies that do not have a shareholder controlling at least 20% of votes are classified as widely held
Austria Financial 30.3 10.00 38.33 13.33 0.00 18.33 16.67 3.33
Non-financial 69.7 11.59 59.18 16.18 0.00 4.35 8.70 0.00
Belgium Financial 39.2 37.25 32.35 1.96 1.96 13.73 12.75 0.00
Non-financial 60.8 8.86 63.92 2.53 0.00 12.03 12.66 0.00
Finland Financial 10.1 23.08 64.10 10.26 0.00 2.56 0.00 0.00
Non-financial 89.9 31.71 44.11 17.07 1.22 0.61 5.28 0.00
France Financial 16.1 22.45 33.16 6.63 0.00 36.22 1.53 0.00
Non-financial 83.9 12.38 70.92 4.81 4.52 6.58 0.79 0.00
Germany Financial 14.5 12.75 37.58 5.88 0.00 28.27 1.96 13.56
Non-financial 85.5 9.97 69.20 6.37 4.26 5.82 3.61 0.76
176 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Ireland Financial 15.9 63.64 13.64 9.09 4.55 0.00 9.09 0.00
Non-financial 84.1 62.07 26.72 0.00 1.72 5.17 4.31 0.00
Italy Financial 23.6 24.49 18.37 8.16 7.14 38.78 0.00 3.06
Non-financial 76.4 9.43 72.33 11.01 1.57 4.09 1.57 0.00
Norway Financial 16.1 56.00 12.00 24.00 0.00 4.00 4.00 0.00
Non-financial 83.9 33.08 43.65 10.99 0.38 4.55 4.64 2.71
Portugal Financial 24.1 19.05 71.43 0.00 0.00 4.76 4.76 0.00
Non-financial 75.9 23.44 55.47 7.81 0.78 4.69 7.81 0.00
Spain Financial 16.1 29.41 23.37 7.35 0.00 38.40 1.47 0.00
Non-financial 83.9 25.85 62.03 3.49 1.95 6.34 0.28 0.06
Sweden Financial 17.1 50.00 40.48 4.76 0.00 2.38 2.38 0.00
Non-financial 82.9 37.43 47.49 5.03 0.00 3.35 6.70 0.00
Switzerland Financial 24.3 32.69 24.04 17.31 0.00 16.35 9.62 0.00
Non-financial 75.7 25.93 55.86 4.12 1.44 7.10 5.25 0.31
UK Financial 0.185 60.11 18.01 0.00 0.42 17.94 3.53 0.00
Non-financial 0.815 63.76 24.97 0.09 0.84 6.90 3.44 0.00
All Financial 0.183 39.92 26.54 5.05 1.31 21.48 4.00 1.71
Non-financial 0.817 36.39 48.15 3.85 2.13 5.96 3.30 0.21
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 177
shares. For example, Zingales (1994) as of May 1998. Prior to this, German
reports that in Italy this premium is firms could be authorized to issue shares
81.5%, Megginson (1990) reports a with multiple voting rights by State authorities.
13.3% premium in the UK, Muus and For example, at the end of 1996, Rwe AG
Tyrell (1999) find a 29% premium in had multiple voting stocks with a 20
Germany, and Muus (1998) documents a voting right, and Siemens AG had multiple
51.35% premium in France. In addition, voting stocks with a 6 voting right.
Horner (1988) documents a 27% pre- Multiple voting shares were issued in
mium in Switzerland and Rydqvist (1992) Germany by Bewag, Frankisches
finds a 6.5% premium in Sweden. In the Uberlandwerk, Hamburger Hochbahn,
US, Lease et al. (1983) find a voting pre- Hamburgische Electricitats Werke, Lech
mium of 5.4%. Nenova (2000) analyzes a Elektrizitatswerke, and Uberlandwerk
sample of 661 dual class share firms in Unterfranken. Multiple voting shares are
18 countries and reports that control benefits legal in France and most firms grant two
constitute 50% of firm value in Mexico, votes for each ordinary share, as long as
around 0% of firm value in Denmark, and they have been held for at least two con-
between one-quarter and one-half of firm secutive years; for publicly traded firms this
value for Brazil, Chile, France, Italy, and minimum holding period can be extended
South Korea. Nenova adds that a large up to four years. These multiple voting
fraction (i.e., more than 70%) of the stocks do not represent a special category
differences in benefits can be explained by of stocks in France, so we treat them as
the legal systems, in particular the quality ordinary shares.
of general investor protection, minority Table 7.6 ranks countries by this last
rights in the case of control transfer, and statistic. Some ironies emerge when we
standards of law enforcement. compare this ranking with the ranking of
Panel A of Table 7.6 displays the legal countries by the severity of legal restric-
restrictions in each country on dual class tions on dual class shares, which are of
shares, the proportion of firms issuing dual four types: (1) the one-share-one-vote
class shares for each country, and the rule; (2) a cap on the proportion of non
average minimum percentage of the stock voting stocks; (3) a rule placing a mini-
capital required to control 20% of the mum on the votes accruing to any type of
voting rights (denoted by Own 20% share; (4) no restrictions on the type of
Con). For example, Realia (Figure 1) has stock issued.
2.641 million A-shares with one vote per The European countries with the lowest
share and 42.922 million B-shares with average minimum percentages of shares
one-tenth of a vote per share. A-shares required to ensure 20% control are
represent 5.8% of stock capital (2.641 Sweden (9.83%), Switzerland (15.26%),
million/45.563 million) and 38.09% (i.e., and Finland (15.42%). Thus, Sweden and
2.641 million/(2.641 million 4.292 Finland, the only countries that impose a
million)) of votes, while B-shares represent lower limit on the voting rights of shares,
61.91% of votes. Hence, it takes 3.05% exhibit a greater discrepancy between own-
( 0.2 * (5.8%/38.09%)) of A-share ership and control than countries that
capital to acquire 20% of the votes place no limits on voting rights, such as
(Own 20% Con is 0.0305). Austria and Ireland. Therefore, the explicit
Minimum percentages are calculated for floor on the voting rights of a share pro-
each firm from its capital structure, then vides a clear indication of the unfairness in
averaged over all sample firms to construct voting rights that regulators will tolerate,
the table. These figures need not account as well as a defense against public criticism.
for the use of multiple voting shares Sweden also has the highest percentage of
(except in Scandinavian countries). The firms issuing dual class shares (66.07%),
issue of multiple voting shares was out- followed by Switzerland (51.17%),
lawed in Italy, Spain, the UK, and Germany, Italy (41.35%), and Finland (37.60%).
Table 7.5 Concentration of control and firm size. Data relating to 5,232 publicly traded corporations are used to construct this table, which groups
corporations according to their size (market capitalization), and, for each variable, presents means for the largest 20 firms, 50 medium-size
firms, and the smallest 50 firms. For countries with less than 120 firms, we group the largest 20 firms in the “Largest 20” group, and then
divide the rest into two equal groups. The table presents the percentage of firms controlled by different controlling owners at the 20% threshold.
Controlling shareholders are classified into six types. Family: A family (including an individual) or a firm that is unlisted on any stock exchange.
Widely held financial institution: A financial firm (SIC 6000-6999) that is widely held at the control threshold. State: A national government
(domestic or foreign), local authority (county, municipality, etc.), or government agency. Widely held corporation: A non financial firm, widely
held at the control threshold. Cross-holdings: The firm Y is controlled by another firm, that is controlled by Y, or directly controls at least 20%
of its own stocks. Miscellaneous: Charities, voting trusts, employees, cooperatives, or minority foreign investors. Companies that do not have a
shareholder controlling at least 20% of votes are classified as widely held.
Average
market cap Widely Widely held Widely held
178 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Country Category (m US$) held Family State corporation financial Misc. Cross-holding
Austria Largest 20 1,593.0 10.00 32.50 42.50 0.00 5.00 10.00 0.00
Middle 40 227.8 14.29 47.14 11.43 0.00 15.71 8.57 2.86
Smallest 39 33.5 11.76 71.57 4.90 0.00 2.94 8.82 0.00
Belgium Largest 20 5,386.4 20.00 37.50 0.00 0.00 27.50 15.00 0.00
Middle 50 722.2 18.00 47.00 4.00 0.00 20.00 11.00 0.00
Smallest 50 58.0 16.00 65.00 0.00 2.00 2.00 15.00 0.00
Finland Largest 20 2,310.2 38.10 33.33 28.57 0.00 0.00 0.00 0.00
Middle 50 274.3 41.18 38.23 11.76 2.94 1.47 4.41 0.00
Smallest 50 30.1 17.65 65.20 9.80 0.00 0.00 7.35 0.00
France Largest 20 16,467.1 60.00 30.00 0.00 0.00 10.00 0.00 0.00
Middle 50 149.9 14.00 68.00 6.00 2.00 10.00 0.00 0.00
Smallest 50 6.8 8.00 77.00 2.00 4.00 9.00 0.00 0.00
Germany Largest 20 27,510.2 45.00 15.00 10.00 0.00 12.50 5.00 12.50
Middle 50 158.3 10.00 75.00 2.00 7.00 4.00 1.00 1.00
Smallest 50 12.0 14.00 81.00 3.00 0.00 2.00 0.00 0.00
Ireland Largest 20 2,260.2 70.00 20.00 0.00 2.50 0.00 7.50 0.00
Middle 25 198.9 50.00 29.16 0.00 4.17 8.33 8.33 0.00
Smallest 24 37.0 68.00 24.00 4.00 0.00 4.00 0.00 0.00
Italy Largest 20 10,398.8 35.00 20.00 25.00 0.00 15.00 0.00 5.00
Middle 50 320.4 8.00 63.00 6.00 6.00 14.00 2.00 1.00
Smallest 50 70.4 14.00 67.00 6.00 2.00 10.00 1.00 0.00
Norway Largest 20 2,302.7 30.00 45.00 20.00 0.00 0.00 5.00 0.00
Middle 50 310.8 38.00 34.83 16.50 0.00 3.83 5.67 1.17
Smallest 50 62.4 40.00 35.67 7.40 0.00 7.33 4.40 5.20
Portugal Largest 20 2,338.9 15.00 60.00 17.50 2.50 0.00 5.00 0.00
Middle 34 170.4 28.57 60.00 2.86 0.00 5.71 2.86 0.00
Smallest 33 42.8 18.75 60.94 1.56 0.00 6.25 12.50 0.00
Spain Largest 20 4,919.2 45.00 20.83 10.00 9.17 15.00 0.00 0.00
Middle 50 302.4 34.00 46.00 8.00 2.00 10.00 0.00 0.00
Smallest 50 65.3 36.00 56.00 2.00 0.00 6.00 0.00 0.00
Sweden Largest 20 9,575.2 80.00 5.00 5.00 0.00 0.00 10.00 0.00
Middle 50 397.9 49.02 42.16 2.94 0.00 0.98 4.90 0.00
Smallest 50 44.0 32.00 63.00 0.00 0.00 3.00 2.00 0.00
Switzerland Largest 20 24,589.9 50.00 35.00 0.00 0.00 15.00 0.00 0.00
Middle 50 352.2 18.00 58.00 10.00 2.00 5.00 7.00 0.00
Smallest 50 58.0 28.00 55.00 7.33 2.67 6.00 1.00 0.00
UK Largest 20 40,563.4 90.00 0.00 0.00 0.00 10.00 0.00 0.00
Middle 50 151.7 64.71 19.60 0.00 0.00 13.73 1.96 0.00
Smallest 50 4.2 42.00 38.67 0.00 0.00 8.00 11.33 0.00
All Largest 12,027.2 45.24 27.24 12.20 1.09 8.46 4.42 1.35
Middle 295.0 29.28 48.82 6.53 2.01 8.65 4.25 0.46
Smallest 38.9 25.32 59.66 3.72 0.90 5.17 4.80 0.44
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 179
Table 7.6 Legal restrictions on issuing dual class shares. Data relating to 5,232 publicly traded corporations are used to construct this table. Own 20%
Con is the average minimum percent of the book value of equity required to control 20% of votes (as computed for each firm in our sample);
Dual class shares (%) is the percentage of firms with outstanding dual class shares.
By contrast, Rydqvist’s survey (1992) give these shares “adequate voting rights”
reported that 75% of Swedish firms issue on resolutions on (1) dividends in arrears,
non voting stocks. Bergström and Rydqvist (2) reducing share capital and winding-up
(1990) reported that stocks issued by the firm, and (3) actions affecting their
Swedish firms can be restricted or unre- class rights. The voting rights attached to
stricted to foreigners, but unrestricted preference shares are defined by the arti-
shares cannot exceed 40% of equity and cles of the firm and vary across firms. We
20% of the votes. Swedish firms seem to combine limited voting (i.e., preference)
have taken this as a license to issue shares shares with non-voting shares to calculate
with inferior voting rights to foreigners. the ratio of votes to share capital.The con-
Shares with superior voting rights are often sequence is that 19.91% of firms issue
not traded in Sweden. In Switzerland, firms dual class shares and the Own 20% Con
can issue different classes of shares: ratio is 18.74%.
Bearer (B-shares), registered (R-shares),
and non-voting shares. R- and B-shares
have identical cash-flow rights; however, 4.2. Pyramiding, holdings through
R-shares have more voting power than multiple control chains, and
B-shares since they are issued with a lower cross-holdings
face value (Gardiol et al., 1997; Horner,
1988). Italian firms can issue limited vot- Table 7.7 displays the control enhance-
ing (i.e., preference) shares and non voting ments employed by firms that have a
shares. Only eight firms have limited voting controlling shareholder at the 20%
shares outstanding. Both limited voting and threshold. Pyramids are used by 19.13%
non voting shares have legally prior claims of such firms in our sample, being most
on dividends, and on reimbursement in case prevalent in Norway (33.90%) and least
of liquidation, see Zingales (1994). Finnish prevalent in Finland (7.46%). Holdings
regulations require a control contestant to through multiple control chains are used
treat all classes of shares in a “fair and by 5.52% of controlling shareholders,
equitable” manner, so bidders need to offer being most prevalent in Norway
all classes of shares the same tender price. (20.34%), Italy (8.78%), and Germany
The protection of minority shareholders (7.22%) but insignificant elsewhere.
during takeover contests may explain why Cross-holdings are used by 0.73% of the
small investors are willing to hold limited controlling shareholders, being most
voting and non voting shares. prevalent in Germany (2.69%) and
At the opposite extreme, dual class Norway (2.04%)14 but marginal in other
shares are rare in Portugal (0%), Belgium countries, whose regulations typically set a
(0%), Spain (0.16%), and France 10% cap on these stakes.
(2.64%).13 These countries have either a
one-share-one-vote rule or a cap on the 4.3. Other control-enhancing
proportion of non voting stocks. In mechanisms
Norway, departures from the one-share-
one-vote principle require governmental A controlling shareholder is said to be
approval. This seems readily given since “alone” if no other owner controls at least
13% of firms have multiple classes of 10% of the voting rights. Table 7.7 shows
shares. In the UK, non-voting shares have that this is true of 53.99% of the
been outlawed since 1968, but firms can firms that are not widely held. Austria
issue preference shares that have a prior exhibits the highest percentage of such
claim on dividends and cumulative divi- firms (81.82%) and Norway the lowest
dends, but limited voting rights. Preference (38.78%). Bennedsen and Wolfenzon
shares may have no voting rights at gen- (2000) and Gomes and Novaes (1999)
eral meetings. However, listing rules on the discuss the role of the second largest
London Stock Exchange require firms to shareholders.
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 183
Table 7.7 Percentage of firms adopting control-enhancing devices. We only include the 3,300 firms
with controlling shareholders at the 20% level. Pyramids reports the percentage of firms
whose largest controlling shareholder adopts pyramids as control devices. Holdings through
multiple control chains reports the percentage of firms whose largest controlling share-
holder adopts at least 5% of holdings through multiple control chains as control devices.
Cross-holdings reports the percentage of firms whose largest controlling shareholder adopts
cross-holdings as control devices. Controlling owner alone reports the percentage of firms
that have single controlling owners. Management reports the percentage of firms whose top
managers come from the largest shareholder’s family. The table presents the percentage of
firms controlled by different controlling owners at the 20% threshold.
Holdings
Number through multiple Cross- Controlling
Country of firms Pyramids control chains holdings owner alone Management
These averages are computed over firms largest average number of firms controlled
where at least one owner owns at least 5% by a single family (1.46), while Finland
of the control rights. The largest ultimate (1.05) and Switzerland (1.10) have the
owners average the highest cash-flow rights smallest.
in Germany (48.54%), Austria (47.16%) Another perspective on the concentra-
and France (46.68%), but the lowest tion of family control is provided by the
cash-flow rights in Ireland (18.82%) and percentage of total market capitalization
the UK (22.94%), followed by the controlled by the top families in each coun-
Scandinavian countries of Norway try. For each family, we sum the market
(24.39%), Sweden (25.15%), and Finland capitalization of all firms in which the
(32.98%). family is the largest controlling shareholder,
The largest ultimate owners average the then divide by the market capitalization of
highest voting rights in Germany (54.50%) all firms in our sample from that country.
and Austria (53.52%), but the lowest in We then rank families by the share of
Ireland (21.55%) and the UK (25.13%), market capitalization that they control and
again followed by the Scandinavian coun- calculate the share of market capitalization
tries of Sweden (30.96%), Norway controlled by the top 1, 5, 10, and 15
(31.47%), and Finland (37.43%). families.
Table 7.9, Panel C shows that the largest The top family controls 17.89% of total
ultimate shareholder’s average ratio of market capitalization in Switzerland and
cash-flow to voting rights is 0.868, being 10.40% in Italy,15 but only 1.40% in the
lowest in Switzerland (0.740), Italy UK, 1.66% in Spain and 3.40% in
(0.743), and Norway (0.776), and highest Ireland. The top 15 families control
in Spain (0.941), Portugal (0.924), and 36.77% of total market capitalization in
France (0.930). Portugal and 36.63% in Belgium, but only
To measure the concentration of corpo- 6.55% in the UK, 13.48% in Spain, and
rate control in families,Table 7.10 displays 15.38% in Ireland.
the average number of firms controlled by
each family via control chains that include
at least 10% of the control rights at each 6. CONCLUSION
link. For example, if a family controls 20%
of Firm A, Firm A controls 15% of Firm In this paper, we document the ultimate
B, and Firm B controls 9% of Firm C, then ownership of 5,232 listed firms in 13
we include only A and B amongst the firms Western European countries. Widely held
controlled by that family. Italy has the and family controlled firms predominate.
186 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 7.9 Cash-flow and control rights. This table includes data relating to 4,806 publicly traded
corporations (including both financial institutions and non financial institutions) where
the largest controlling owner has at least 5% of voting rights. Cash-flow rights repre-
sents the ultimate ownership stake held by the largest controlling shareholder. Control
rights the percentage of voting rights controlled by the largest controlling shareholder.
Number of Standard
Country firms Mean deviation Median 1st quartile 3rd quartile
Table 7.10 How concentrated is control by families/unlisted firms? This table provides summary
statistics on the concentration of control by families and unlisted firms. Average
number of firms per family refers only to firms in the sample. Percent of total market
value of listed corporate assets that families control is the aggregate market value of
common equity of firms controlled by the largest, the top 5, the top 10, and the top 15
families divided by the total market value of common equity for all firms included in
the sample for a given country.
Average Percent of total market value of listed corporate assets that families control
number of
firms per Top Top 5 Top 10 Top 15
Country family family families families families
Austria 1.16 5.64 15.59 19.47 22.15
Belgium 1.22 5.63 20.38 30.45 36.63
Finland 1.05 3.94 13.98 21.82 25.90
France 1.18 5.94 22.04 29.18 33.80
Germany 1.24 5.43 15.66 21.29 25.01
Ireland 1.16 3.40 11.88 14.46 15.38
Italy 1.46 10.40 16.83 20.18 21.92
Norway 1.29 3.73 16.01 23.13 27.18
Portugal 1.23 6.11 24.59 34.23 36.77
Spain 1.19 1.66 6.97 10.92 13.48
Sweden 1.27 3.37 9.29 13.26 15.66
Switzerland 1.10 17.89 24.35 28.94 30.93
UK 1.17 1.40 4.11 5.85 6.55
Widely held firms are more important University of Singapore, Washington University
amongst financial and large firms, while at St Louis, the Capacity Building Seminar in
families are more important for non finan- Manila (sponsored by the Asian Development
cial and small firms. In some continental Bank), the European Financial Management
Association meeting (Athens), the European
European countries, the State controls a
Finance Association meeting (London), the
significant number of larger firms. We also Financial Management Association meeting
document the means whereby owners gain (Seattle), and the Scottish Institute for
control rights in excess of their cash-flow Research in Investment and Finance
rights. The use of multiple class voting (Edinburgh). We also thank Bolsa de Valores de
shares contribute only marginally to this, Lisboa, Commerzbank, Helsinki Media Blue
except in just a few countries. The use of Book, Hugin, the Union Bank of Switzerland,
pyramids and holdings through multiple and the Vienna Stock Exchange for generously
control chains is also marginal. providing us with their data sets. Mara Faccio
acknowledges research support from Università
Cattolica, Milan, and MURST (research grant
#MM13572931). Larry Lang acknowledges
NOTES research support from a Hong Kong Earmarked
Grant.
* Corresponding author. 1 Japan stands apart from the rest of
† We are grateful to Marco Bigelli, Lorenzo East Asia in this regard: the top Japanese
Caprio, Edith Ginglinger, Arun Khanna, John family controls only 0.5% of total market
McConnell, Stefano Mengoli, Robert Pye, capitalization.
Gordon Roberts, Bill Schwert (the editor), 2 In addition, Greek shares are often held in
George Tian, and especially Andrei Shleifer and bearer form, masking the identities of their
an anonymous referee for providing helpful owners.
comments. We benefited from comments from 3 La Porta et al. (1999) use Hugin for
workshop participants at the National Denmark. However, Hugin covers less than 20%
188 EQUITY OWNERSHIP STRUCTURE AND CONTROL
of all listed firms and we could not locate 11 Voting caps are used, for example, by
information from other official sources. BASF (2.62%), Bayer (5%), Deutsche Bank
4 The Stichting Toezicht Effectenverkeer (5%), Linde (10%), Mannesmann (5%),
(STE, the Securities Board of the Netherlands), Phoenix (10%), Schering (3.51%), and
responded to our request for data as follows: Volkswagen (20%) in Germany; and Telefonica
“The implementation of the 1992 Act has (10%) in Spain. In Italy, the law requires vot-
resulted in the list of disclosed notifications ing caps for cooperative banks; they are rather
being no more than an historical overview that common in recently privatized firms such as
has been overtaken by countless events and no Comit (3%) and Credit (3%). Voting caps also
longer provides the desired transparency of the obtain in France, e.g., in Alcatel, Danone, and
(Dutch) stock market. Furthermore, the file Pernod Ricard. In Switzerland, 12 out of the
corruption will become greater with the passage 50 largest firms have voting caps. Examples
of time. Given the above and the fact that from the ten largest corporations include
the 1996 Act does not empower the STE to take Novartis (2%), Nestlé (3%), UBS (5%),
any measures in this regard which would lead to Swiss Bank Corporation (5%), and Zurich
an up-to-date overview, the STE does not issue Insurance (3%).
this list to third parties.” See also De Jong et al. 12 As reported in Crespi-Cladera and Garcia-
(1998). Cestona (1998), the State holds these golden
5 Data are from 1996 for France, Germany, shares in some recently privatized Spanish
Italy, Switzerland, and the UK; from 1997 for firms, such as Repsol, Telefonica, and Endesa.
Spain and Portugal; from 1998 for Sweden and The use of golden shares is also popular among
Norway; and from 1999 for Austria, Belgium, privatized firms in Italy, as emerged from the
Finland, and Ireland. vicissitudes of Telecom Italia and Enel in the
6 Becht and Boehmer (1998) for Germany; late 1990s.
Bianchi et al. (1998) for Italy; Bloch and 13 Similar evidence is reported for Spain in
Kremp (1998) for France; Crespi-Cladera and Crespi-Cladera and Garcia-Cestona (1998) and
Garcia-Cestona (1998) for Spain; De Jong et al. for France in Muus (1998).
(1998) for the Netherlands; Renneboog (1998) 14 These figures are, respectively, 2.62% and
for Belgium; Goergen and Renneboog (1998) 2.27% if we also include second ultimate
for the UK. owners (see Table 7.3).
7 Our sample coverage of countries is: Austria: 15 Brioschi et al. (1989) show that, in the
100.00 (%), Belgium: 89.04, Finland: 87.76, mid-1980s, more than a quarter of total market
France: 89.26, Germany: 99.15, Ireland: 82.14, capitalization could be traced to the control of
Italy: 100.00, Norway: 72.77, Portugal: three single families.
100.00, Spain: 100.00, Sweden: 94.96,
Switzerland: 100.00, and the UK: 95.69. The
low figure for Norway is due more to limited
data coverage than to our screening. REFERENCES
8 AIDA is a private database provided by
Bureau Van Dijk. The AIDA database Barclay, M., Holderness, C., 1989. Private
provides accounting information for about
benefits from control of public corporations.
130,000 Italian firms and ownership informa-
tion for 25,314 Italian firms. Firms with Journal of Financial Economics 25,
ownership data in the AIDA database are not 371–395.
ranked in any order. We traced ownership to Barth, J.R., Caprio Jr., G., Levine, R., 2000.
ultimate owners starting from the first firm Banking systems around the globe: do
listed in the AIDA database using ownership regulation and ownership affect performance
information contained in that database. A firm and stability? Unpublished working paper,
was included if we could trace its ultimate own- World Bank, Washington DC.
ers. We stopped when our sample reached 15% Becht, M., Boehmer, E., 1998. Ownership and
of firms with ownership data. voting power in Germany. Unpublished work-
9 Their results were based on data compiled
ing paper, European Corporate Governance
by the French central bank (Fiben), that is
unavailable to the public. Network.
10 The figures are based on a sample of Bennedsen, M., Wolfenzon, D., 2000. The
12,600 unlisted firms from the Jordan’s balance of power in closely held corporations.
database, which is compiled by a private data Journal of Financial Economics 58,
vendor, Bureau Van Djik. 113–139.
ULTIMATE OWNERSHIP OF WESTERN EUROPEAN CORPORATIONS 189
Bergström, C., Rydqvist, K., 1990. Ownership of Gardiol, L., Gibson-Asner, R., Tuchschmid, N.S.,
equity in dual-class firms. Journal of Banking 1997. Are liquidity and corporate control
and Finance 14, 255–269. priced by shareholders? Empirical evidence
Berle, A., Means, G., 1932. The Modern from Swiss dual class shares. Journal of
Corporation and Private Property. Corporate Finance 3, 299–323.
MacMillan, New York. Goergen, M., Renneboog, L., 1998. Strong man-
Bianchi, M., Bianco, M., Enriques, L., 1998. agers and passive institutional investors in
Pyramidal groups and the separation between the UK. Unpublished working paper,
ownership and control in Italy. Unpublished University of Reading.
working paper, European Corporate Gomes, A., Novaes, W., 1999. Multiple large
Governance Network. shareholders in corporate governance.
Bloch, L., Kremp, E., 1998. Ownership and con- Unpublished working paper, University of
trol in France. Unpublished working paper, Pennsylvania, Wharton.
INSEAD Business School. Holderness, C., Kroszner, R., Sheehan, D., 1999.
Bolsa de Valores de Lisboa, 1998. Sociedades Were the good old days that good? Changes
Cotadas 1997. Lisbon Stock Exchange, Lisbon. in managerial stock ownership since the great
Brioschi, F., Buzzacchi, L., Colombo, M.G., depression. Journal of Finance 54, 435–470.
1989. Risk capital financing and the separa- Horner, M.R., 1988. The value of the corporate
tion of ownership and control in business voting right: evidence from Switzerland.
groups. Journal of Banking and Finance 13, Journal of Banking and Finance 12, 69–84.
747–772. Il Sole 24 ore, 1997. Il taccuino dell’azionista.
Claessens, S., Djankov, S., Lang, L.H.P., 2000. Il Sole 24 Ore Radiocor, Milan.
The separation of ownership and control in Jensen, M., Meckling, W., 1976. Theory of the
East Asian corporations. Journal of Financial firm: managerial behavior, agency costs, and
Economics 58, 81–112. ownership structure. Journal of Financial
Claessens, S., Djankov, S., Fan, J., Lang, L.H.P., Economics 11, 5–50.
2002. Disentangling the incentive and Johnson, S., La Porta, R., Lopez-de-Silanes, F.,
entrenchment effects of large shareholdings. Shleifer, A., 2000. Tunnelling. American
Journal of Finance, forthcoming. Economic Review Papers and Proceedings
Comision Nacional del Mercado de Valores, 90, 22–27.
1998. Participaciones significativas en La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
sociedades cotizadas. (Data file). CNMV, Vishny, R.W., 1997. Legal determinants of
Madrid. external finance. Journal of Finance 52,
Commerzbank, 1997. Wer gehört zu wem, 19th 1131–1150.
Enlarged Edition. Commerzbank AG, La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Frankfurt-am-Main. 1999. Corporate ownership around the world.
Commissione Nazionale per le Società e la Journal of Finance 54, 471–518.
Borsa, CONSOB, 1997. Bollettino - edizione La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
speciale n. 4/97 - Compagine azionaria delle Vishny, R.W., 2002. Investor protection and
società quotate in borsa o ammesse alle corporate valuation. Journal of Finance 57,
negoziazioni nel mercato ristretto al 1147–1170.
31 dicembre 1996. CONSOB, Milan. Lease, R.C., McConnell, J.J., Mikkelson, W.H.,
Crespi-Cladera, R., Garcia-Cestona, M.A., 1983. The market value of control in publicly
1998. Ownership and control: a Spanish sur- traded corporations. Journal of Financial
vey. Unpublished working paper, European Economics 11, 439–471.
Corporate Governance Network. Lins, K., Servaes, H., 1999a. International
De Jong, A., Kabir, R., Marra, T., Roell, A., evidence on the value of corporate
1998. Ownership and control in the diversification. Journal of Finance 54,
Netherlands. Unpublished working paper, 2215–2239.
European Corporate Governance Network. Lins, K., Servaes, H., 1999b. Is corporate
Financial Times, 1997. Extel Financial, diversification beneficial in emerging
Company Research CD-rom #36/1997. FT, markets? Unpublished working paper, London
London. Business School.
190 EQUITY OWNERSHIP STRUCTURE AND CONTROL
London Stock Exchange, 1997. The London Renneboog, L., 1998. Shareholding concentration
Stock Exchange Yearbook. LSE, London. and pyramidal ownership structures in
Megginson, W., 1990. Restricted voting stock, Belgium: stylized facts. Unpublished working
acquisition premiums, and the market value paper, European Corporate Governance
of corporate control. The Financial Review Network.
25, 175–198. Rydqvist, K., 1992. Dual-class shares: a review.
Muus, C.K., 1998. Non-voting shares in Oxford Review of Economic Policy 8, 45–57.
France: an empirical analysis of the voting Shleifer, A., Vishny, R., 1997. A survey of cor-
premium. Unpublished working paper, porate governance. Journal of Finance 52,
Johann Wolfgang Goethe-Universitat, 737–783.
Frankfurt-am-Main. The Herald Tribune, 1997. French Firm
Muus, C.K., Tyrell, M., 1999. An empirical Handbook 1997. SFB-Paris Bourse.
analysis of the voting premium in Germany. Union Bank of Switzerland, 1998. Swiss Stock
Unpublished working paper, Johann Wolfgang Guide 96/97. UBS, Zurich.
Goethe-Universitat, Frankfurt-am-Main. Wiener Börse, 2000. Yearbook 1999. Vienna
Nenova, T., 2000. The value of corporate votes Stock Exchange, Vienna.
and control benefits: a cross-country analysis. Zingales, L., 1994. The value of the voting right:
Unpublished working paper, Harvard a study of the Milan stock exchange experience.
University. Review of Financial Studies 7, 125–148.
Chapter 8
ABSTRACT
We analyze why the control of listed German and U.K. companies is so different. As
shareholders in Germany are less protected and control is less expensive, German investors
prefer controlling stakes. We also focus on economic factors such as profitability, risk, and
growth to predict the probability of occurrence of different states of control six years after the
flotation. Large U.K. companies become widely held, whereas in large German firms new
shareholders control significantly larger stakes. Wealth constraints become binding for U.K.
shareholders, whereas German shareholders can avoid this by using pyramids. We find
substantial differences between a takeover by a concentrated shareholder and one by a widely
held company. For the United Kingdom, the probability of the former increases when the
company is risky, small, and poorly performing. Conversely, the latter is more likely when the
target is large, fast growing, and profitable. Poor performance and high risk require control
and monitoring by a concentrated shareholder. Conversely, high growth and profitability attract
widely held companies. Founders are less inclined to dilute their stake to retain private benefits
of control. When German firms are profitable and risky, control is likely to go to a concentrated
shareholder, but growth and low profitability increase the probability of a control acquisition
by a widely held firm.
Ownership disclosure
Minimum disclosure level 5% of voting rights 3% (5% prior to 1989)
(25% prior to 1995)
Minimum threshold of None 15%
control to reveal
strategic intent
(takeover intentions)
Dual-class shares Nonvoting shares: No nonvoting shares
common
Voting caps Common at 5–10%, No
but caps were
abolished in 1998
with a grandfather
clause for existing
voting caps
in by-laws
Control structure Complex, multilayered Simple, one ownership
cascades of tier
ownership levels
Takeover code (since 1995)
Minimum level 75% 30%
of control for
obligatory tender
offer on all shares
Minimum price of tender Can be lower than Not lower than
market value the highest share
price during last
12 months
Using voting rights
Casting votes Presence at AGM Allowed by mail, fax, or
required in person Internet. But, in practice,
or representative presence (in person or
by representative) is
required for show of
hands. Proxy votes only
matter in poll
Registration of votes Deposit of shares Shares need to be
with notary, registered
depository bank,
or company itself
Degree of minority Weak Strong
protection
Boards Two-tier board One-tier board
with a large- with independent
shareholder nonexecutive directors
dominated super-
visory board
Corporate aim of Pursuing stake- Maximizing share-
management holder interest holder value
(continued)
196 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 8.1 Continued
owning large equity stakes in terms of lost issued nonvoting shares converted them
liquidity, an intricate web of multilayered into voting shares under the pressure of the
tiers of shareholdings is often used by London Stock Exchange and institutional
families or companies. The main reason for investors during the early 1990s. Whereas
building such pyramids is control leverage.4 German corporate law (Aktiengesetz, para.
As companies can issue up to 50% of the 139) prohibits the issue of multiple voting
equity as nonvoting shares, a mere 25% of rights, nonvoting shares up to the amount
the cash flow rights (50% of the voting of ordinary shares outstanding are explic-
rights) at every tier of shareholdings is suf- itly allowed.8 As such, the issue of nonvot-
ficient for a shareholder to retain control ing shares increases the relative power of
over a target company while the capital the initial shareholders and allows them to
investment is minimized. In addition, a retain control with limited investment.9
shareholder (e.g., a family) could also hide The only regulatory element which
his identity prior to the introduction of discourages shareholders in Germany from
the disclosure regulation.5 While pyramids holding large equity stakes is that voting
are not explicitly forbidden in U.K. corpo- limitations can be imposed by the by-laws.
rate law, it is surprising that this control Usually the voting power of a shareholder
leverage technique is not used at all. The is limited to 5% or 10%.10 Only recently
main reason is that ownership disclosure have voting caps been prohibited in
regulation in the United Kingdom not only Germany (Act on Control and Transparency
applies to individuals or companies but also of Enterprises of 1998 (KonTraG)), but a
to individuals and companies with voting grandfather clause applies for existing
agreements. Such voting agreements voting caps. Voting caps in listed U.K.
consist of obligations or restrictions companies are not allowed. Hence, the
between shareholders with respect to the one-share-one-vote principle is better
use, retention, or disposal of their stakes. A upheld in the United Kingdom than in
coalition of shareholders with a voting Germany.
agreement will be considered by the regu-
latory authorities as a single shareholder.
This implies, for instance, that if the com- 2.2.7 Proxy voting and voting practice
bined direct and indirect shareholdings of a Another important element of the protec-
coalition amount to at least 3%, disclosure tion of shareholder rights is the voting
is compulsory.6 Furthermore, a coalition procedure. In both the United Kingdom and
controlling directly and indirectly 30% or Germany, a person or legal entity can
more of the equity will be obliged to make represent a shareholder at the AGM. At
a tender offer for all shares outstanding.7 first sight there seems to be less need to
This section has shown that the costs of resort to proxy voting in the United
holding large share stakes are reduced by Kingdom because exercising one’s voting
pyramids that allow the combination of rights seems substantially easier. However,
strong control with limited investment. the general voting practice requires that a
shareholder of a U.K. firm or his represen-
2.2.6 Nonvoting shares and voting caps tative is present at the general meeting in
order to cast his votes. The U.K. voting
By issuing nonvoting shares, initial owners practice is captured by this statement from
can dilute their cash flow rights in the firm the IPO prospectus of Compel plc: “At a
without relinquishing control. Although general meeting every member present in
nonvoting shares are in principal admitted person shall, on a show of hands, have one
by the London Stock Exchange, issues of vote and every member present in person or
nonvoting shares have been actively dis- by proxy shall, on a poll, have one vote for
couraged (Brennan and Franks, 1997). every ordinary share of which he is the
Goergen and Renneboog (2001) report holder.” On uncontroversial issues, voting
that the few listed U.K. companies that had takes place by a show of hands, which
198 EQUITY OWNERSHIP STRUCTURE AND CONTROL
risk and high risk triggers control transfers Profitable firms generating sufficient
to large shareholders with strong monitoring cash flow allow the initial shareholders to
skills (Proposition 4). As German firms are retain control for a longer period than
older at the float (and have more stable poorly performing firms (Dennis and Sarin,
and less risky cash flows than their U.K. 1999). Poor performance may reflect not
counterparts) and as shareholders can only weak management but also poor mon-
control a German firm with a lower itoring, which may activate the market for
percentage of voting rights, Proposition 4 corporate control or persuade another
is more likely to hold for U.K. firms. shareholder with superior monitoring skills
Growth may also force changes in con- to acquire control (Franks et al., 2001).
trol: rapidly growing firms may have to Thus we expect poor performance to trig-
resort to attracting more external equity ger changes in control (Proposition 7).
than firms operating in mature industries. Such changes in control are likely to occur
Thus the stronger the growth, the higher is differently in the United Kingdom than in
the probability that control by the initial Germany. In the former, the market for cor-
shareholders is diluted and that either a porate control consists in full takeovers
new shareholder emerges (by a partial or (Franks and Mayer, 1996), whereas in
full takeover) or that the firm becomes Germany it operates via partial takeovers
widely held (Crespi, 1998) (Proposition 5). or a transfer of a controlling stake to a new
We expect this relation to hold more major shareholder (Jenkinson and
strongly for U.K. than for German firms Ljungqvist, 2001).
because, in the latter, pyramids enable
shareholders to attract external capital
while limiting control dilution. 4. DATA SOURCES, DESCRIPTION OF
In line with DeAngelo and DeAngelo VARIABLES, AND METHODOLOGY
(1985) for the United States, voting shares
in German IPOs are normally held by the 4.1 Sample description and data
initial owners, whereas nonvoting shares sources
are issued to the outsiders in the IPO
(Goergen, 1999). Immediately after the Even though the German economy is about
IPO, the family shareholder owns on 1.8 times larger in terms of gross domestic
average 57% of the voting rights, but only product (GDP) than the U.K. economy, its
47% of total equity (voting plus nonvoting capitalization is substantially smaller, as
shares). Thus, we expect that issuing non- the number of quoted companies on the
voting shares at the IPO leads to higher German exchanges is less than one-third
control retention by initial shareholders the number of listed firms on the London
and that control retention will be higher in Stock Exchange (which is around 2,000
Germany than in the United Kingdom firms, including 550 financial institutions
(Proposition 6). For those cases where and investment funds). Similarly the num-
control is transferred, we expect new share- ber of IPOs on the London Stock Exchange
holders to be able to acquire control more is much larger. In the United Kingdom, 764
easily in Germany than in the United firms went public during the period
Kingdom. For instance, suppose that there 1981–1988, of which 284 were floated on
are two firms with a different equity struc- the Official Market and 480 on the
ture but which are otherwise identical: one Unlisted Securities Market (USM), that is,
firm has only voting shares outstanding, the secondtier market. Over the same
while the other has issued 50% voting and period, a total of only 96 German firms
50% nonvoting shares. If the initial share- went to the stock exchange, 51 of which
holders decide to sell a controlling stake, were listed on the Official Market and 45
the new controlling shareholder will only chose a listing on the Regulated Market
need half of the capital to acquire control (the second-tier market). This study con-
in the firm with the nonvoting equity.15 centrates on domestic IPOs listed on the
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 201
official and secondary markets, as data for The U.K. firms are riskier (at the 1% level
lower market tiers16 are usually not available. of significance), measured by the standard
To ensure comparability across the deviation of their monthly share returns.
German and U.K. IPOs, in this study we German firms also have a higher cash flow
only retain those IPOs controlled by an both as a proportion of their book value of
individual or a group of persons, such as a assets (CF1) and as a proportion of the
family or unrelated associates. Thus we do sum of the market value of equity and the
not include privatizations or equity carve- book value of debt (CF2).
outs. Consequently this study analyzes Information on the identity of initial
more than 90% of all German IPOs with shareholders and on pre- and post-IPO
available data.17 holdings was obtained from the IPO
The distribution of the population of prospectuses. Substantial share stakes
96 German and 764 U.K. IPOs across were traced over the period after the IPO
industries reveals significant differences.18 through the company reports as well as the
Although the industry with the highest fre- London Stock Exchange Yearbooks for the
quency of IPOs is the same for both coun- United Kingdom and the Saling
tries (electricals, electronics, and office Aktienführer for Germany. Information on
equipment), there are proportionally more both the direct and ultimate voting stakes
German IPOs in mature industries (such as was collected for the German firms. To
mechanical engineering with 15.5% of the identify the ultimate shareholder and his
total number of IPOs and motor compo- level of control, the ownership pyramids
nents with 5.2%). Conversely, there is a were reconstructed on a year-by-year basis.
higher proportion of U.K. IPOs in cyclical Higher tiers of ownership were traced for
service industries with 29% of the sample those large shareholders owning 25% or
(service agencies with 9.0%, property with more of the voting rights as long as the
6.0%, leisure with 5.7%, chain stores with control chain in the ownership cascade was
3.6%, and construction with 4.9%). not interrupted. The ultimate level of con-
Within each industry, U.K. IPOs are also trol is reached when the ultimate share-
usually smaller than German IPOs. holder is either an individual or a family, or
Table 8.2 shows that there are also is a widely held company (i.e., a firm that
marked differences between the population does not have a shareholder owning at least
of German and U.K. IPOs in terms of size, 25% of its voting rights). Whereas ultimate
age, industry, and risk. The German IPOs control is only relevant for German compa-
are twice as large as the U.K. IPOs, with nies, ownership in the United Kingdom is
market capitalizations at the end of the equivalent to control, as multiple or nonvoting
first day of listing of £113 million and shares are not used. Also, U.K. ownership
£56 million, respectively.19 On average, the structures are simple and pyramids are rare
German IPOs were founded 51 years prior (Goergen and Renneboog, 2001).
to the float whereas U.K. firms were set up Share prices were collected from the
only 14 years before the IPO. The age of Karlsruher Kapitalmarktdatenbank
the German and U.K. sample IPOs (KKMDB) and the London Share Price
matched by size is similar to the popula- Database (LSPD). IPO characteristics
tion: German and U.K. firms are, respec- (age and industry) and the closing market
tively, 50 and 12 years old prior to the capitalization for the first day of listing
float. The Herfindahl index demonstrates were obtained from the Deutsche Börse AG
that voting rights in the German firms at and the London Stock Exchange.
the IPO are more concentrated than in their Accounting information was collected from
U.K. counterparts (the difference is the IPO prospectuses, company reports, the
significant at the 1% level; see Table 8.2, Extel Financial Company Research and
panel B). German founder families own Global Vantage CD-ROMs for both coun-
shares at the float in 92% of the firms, tries, and from Datastream and the Extel
while this is the case in 85% of U.K. firms. Microfiches for the United Kingdom.
202 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 8.2 Summary Statistics of Independent Variables
The table is based on a sample of German and U.K. IPOs matched by size (market capitalization).The Z-test in
panel B is a two-tailed test for the equality between two proportions following a binomial distribution. Age is
the firm’s age since registration. Size is in million pounds sterling and is the market capitalization. Concentrated
ownership is measured by a Herfindahl index of all pre-IPO stakes. Founder is a dummy variable set equal to
one when the founder or his heirs still own a large equity stake in the firm at the float. Risk is the standard devi-
ation of the monthly share price return over the five years following the IPO. CF1 is the annual cash flow defined
as the published profit gross of depreciation, interest, taxes, and changes in provisions divided by the sum of the
book values of equity and debt. CF2 is annual cash flow divided by the market value of equity and the book value
of debt. ***, **, * represent the level of statistical significance at the 1%, 5%, and 10% level, respectively.
Sources: IPO prospectuses, company reports, Datastream, Extel Financial, and London Stock Exchange.
4.2 Methodology and model For the IPO samples, the following
description cross-sectional industry-fixed effects are
estimated:
The direct comparability of control in the
two countries was enhanced by matching Percent voting controli
the German firms with U.K. IPOs by industry 1Countryi 2Sizei
and size.This way we select twin companies 3Founderi 4Riski
whose control evolution we analyze while 5Growthi 6Nonvotingi
controlling for factors such as risk, 7Profiti 8Agei 9Countryi *Sizei
profitability, and growth. The size-matched 10Countryi * Founderi
sample consists of 54 German and U.K. 11Countryi * Riski
firms, whereas the industry-matched sample 12Countryi * Growthi
contains 58 IPOs from each country.20 The 13Countryi * Profiti
average difference in size between a German 14Countryi * Agei time dummies
IPO and its matched U.K. IPO is 2.7%, with industry dummies i.
a median of 0.5% and a standard deviation The percent voting control is the percent-
of 4.7%. For seven German firms, a close ages of voting rights held six years after
match, defined as a match within 25% the float by (i) the initial shareholders,
difference in size, could not be found, neither (ii) the new large shareholders, and (iii)
was it possible to find a U.K. industry match small shareholders (the free float),
for three German IPOs. The two German respectively. As the dependent variable is
samples have 52 firms in common.21 censored (the variable is zero in 33%,
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 203
27.4%, and 22.2% of the cases in the correlated with Age, Profit, Nonvoting,
models with initial shareholders, new and Country, and (ii) Age is related to
shareholders, and the free float as Country, Growth, and Nonvoting. These
dependent variables, respectively), we seven correlation coefficients are statisti-
estimate Tobit models. cally significant but are not large in
Firm size (Size) is the natural logarithm absolute terms. As these correlations may
of the market capitalization at the closing cause multicollinearity problems, we
price on the first day of trading, converted subtract from the variables Age and Risk
into 1985 pounds sterling. Founder involve- the country averages of these variables.
ment is captured by the dummy variable Table 8.A1 in the appendix shows that the
Founder, which equals one if the founder or statistical significance of the correlations
his heirs hold a control stake in the firm at between the variables Riskminuscountryavg on
the IPO and zero otherwise. The level of a the one side, and Age, Country, and
firm’s risk (Risk) is measured by the stan- Nonvoting on the other has disappeared, as
dard deviation of monthly stock returns well as the one between Ageminuscountryavg on
over the five-year period after the flotation. one side and Country, Growth, and
The growth rate (Growth) is defined as the Nonvoting on the other side. The high cor-
average annual growth rate of total assets relation between Nonvoting and Country
over the first five years after the float. cannot be eliminated, as nonvoting shares
Nonvoting shares (Nonvoting) is a dummy are only issued by German firms.
variable capturing the issue of nonvoting For the sake of robustness, we also use a
shares at the time of going public (dummy set of alternative independent variables.
equals one). Twenty-three and 25 German Size is now measured by the book value of
firms issued nonvoting shares in the size- total assets at the end of the financial year
and industry-matched samples, respec- covering the IPO date. Founder stands for
tively. The profit rate (Profit) is defined as founder involvement in the management of
the annual cash flow divided by the book the firm: the dummy variable equals one if
value of total assets with annual cash flow the founder or a member of the founder
measured as the profit gross of deprecia- family is an executive director of the firm,
tion, interest, taxes, and changes in tax, and zero otherwise. Risk is the standard
pension, and special provisions. Age meas- deviation of the cash flows at the IPO and
ures the age of the firm in years at the time during the five years subsequent to the IPO.
of the IPO. Nonvoting is now a continuous variable
As both German and U.K. IPOs are measuring the proportion of nonvoting
included in the model, a dummy variable shares in the firm’s total equity immedi-
Country is used, which equals one for a ately after the IPO. For Growth, the same
German company and zero otherwise. The definition is used as for the first set of inde-
coefficient on this dummy registers a pendent variables. The profit rate (Profit)
possible difference in the intercept is now the cash flow over the sum of the
between German and U.K. IPOs. In order market value of equity and the book value
to determine whether there is a country- of debt.
specific effect for each of the variables As the dependent variables of the three
described above, interactive terms Tobit models are interrelated, we also
consisting of the product of each of the estimate multinomial logit models. Here
above variables with the Country dummy we distinguish between four states of
is included in the model. The interactive control six years before the float: firm i is
terms pick up the differential effect (i.e., (i) controlled by the original owners (SC),
the differential slope coefficient) for the (ii) widely held (WH), (iii) taken over by a
German firms. closely held bidder (TC), or (iv) taken
Table 8.A1 in the appendix shows the over by a widely held bidder (TW). Six
Pearson correlation coefficients for the years after the float, two-thirds of the
independent variables and the p-values for German firms in the size-matched sample
the coefficients. We find that (i) Risk is were still controlled by their initial owner.
204 EQUITY OWNERSHIP STRUCTURE AND CONTROL
This proportion is three times higher than sample matched by size. Initial control is
for the U.K. sample. However, the number statistically different in the German and
of (full and partial) takeovers is similar: 19 U.K. firms immediately after the IPO: the
and 22 German and U.K. firms, respec- initial shareholders in German firms retain
tively. As in most U.K. and German firms, 76.4% versus only 62.8% in U.K. compa-
the initial shareholder retains majority nies. Initial shareholder control is diluted
control at the float, we use this state as the much more rapidly in the United Kingdom
benchmark case in the multinomial logit than in Germany. The original shareholders
regressions.22 The specifications include a of British IPOs lose majority control on
differential intercept and interactive slope average only two years after the float,
coefficients, which pick up possible differ- whereas their German counterparts retain
ential effects for the German IPOs. For the majority control up to five years. Although
size- and industry-matched samples, the the reduction in control is slower in
following cross-sectional multinomial German IPOs, as much as 35% of these
model is estimated by voting rights change hands during the six-year
period.The differences in initial shareholder
Uij ’ zij ij and for
concentration in German and U.K. IPOs are
j 1, 2, . . . , J:
statistically significant at the 1% level for
1 each of the six years following the float.
Pr(Y 0) At the float of U.K. firms, most of the
1 k1e
J kxi
shares are purchased by small shareholders,
as the rationing schemes for share distribu-
kxi
e tion in case of oversubscription often favor
Pr(Y j)
1
J kx k small shareholders for control reasons
k1e
(Brennan and Franks, 1997). Consequently
More specifically, Uij is the state of the free float is significantly higher (at the
control j in company i: 1% level of significance) in U.K. firms
(37.2%) than in German ones (22.2%),
State of controlij but remains relatively stable in both
1Countryi 2Sizei countries over the six-year period after the
3Founderi 4Riski IPO. Over this period control is transferred
5Growthi 6Nonvotingi from the initial shareholders to new large
7Profiti 8Agei shareholders who, on average, control
9Countryi * Sizei about 30% of the voting rights in both
10Countryi * Founderi the German and U.K. samples six years
11Countryi * Riski subsequent to the float.
12Countryi * Growthi
13Countryi * Profiti
14Countryi * Agei 5.2 Changes of control and control
time dummies states
industry dummies ij. Whereas Table 8.3 shows how control
evolves over time, Table 8.4 reports the
5. CONTROL EVOLUTION AFTER THE control state six years after the float. Panel
FLOAT: STYLIZED FACTS A of Table 8.4 shows that full takeovers are
uncommon in Germany, with only one case
5.1 Control evolution by the initial in our sample.23, 24 In contrast, 35% of the
and new large shareholders U.K. companies were acquired in a full
takeover within six years after the float. In
The voting power held by the three categories Germany, control is less frequently transferred
of owners (initial, new large, and new small via a full takeover than via the sale of large
shareholders) over the six-year period after voting blocks (which in fact constitutes a
the IPO is recorded in Table 8.3 for the “partial” takeover). Such partial takeovers
Table 8.3 Proportion of Voting Rights Held by Initial Shareholders, by New Large Shareholders, and by Small Shareholders in Recent German and U.K. IPOs
The German IPOs are matched by size with the U.K. IPOs. If a company is taken over and delisted, it is classed as a case of 100% control by the new shareholder as of the
year of the takeover. If a company is taken private by its original shareholders, it will be recorded as a company owned 100% by its original shareholders.
*** ** *
, , indicate that the average stake in U.K. firms is statistically different from the stake in German firms at the 1%, 5%, and 10% significance level, respectively, of a
t-test on the means. Sources: IPO prospectuses, Saling, company reports, Hoppenstedt, London Stock Exchange, and Extel.
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 205
206 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 8.4 State of Control of IPOs Six Years After Flotation
For panels A and B, the disclosed state is the one six years after the IPO or the one the last year of listing if the
firm was delisted before the sixth year after the float. A full takeover is a takeover of the entire voting rights of
the firm, followed by the delisting of the firm. A partial takeover is a change of the major shareholder, the major
shareholder being the largest shareholder holding at least 25% of the voting equity. A firm is widely held if no
single shareholder owns more than 25% of the voting rights. A firm is still controlled by an initial shareholder if
the initial shareholder is the largest shareholder and holds more than 25% of the votes. In panel C, a firm is
classed as widely held if its ultimate largest shareholder is widely held, that is, is not controlled by a person or
family holding more than 25% of the votes. A firm is classed as closely held if the ultimate largest shareholder
is a person or a family. Sources: McCarthy microfiches, Saling Aktienführer, Financial Times, company reports.
are infrequent in the United Kingdom control across the remaining German and
because of the legal requirement to make a U.K. firms is still substantially different, as
tender offer for the entire equity of the firm 62% of the German firms remain in the hands
as soon as an investor acquires 30% or of their initial owners, compared to only 24%
more of a firm’s equity.25 Thus, if share- of the U.K. firms in the size-matched sample
holders in a U.K. company do not want to (and 35% in the industry-matched sam-
end up with 100% control, they will delib- ple). Only one German company becomes
erately remain below the 30% threshold widely held, whereas 32% of U.K. firms are
(Goergen and Renneboog, 2001). already widely held six years after the IPO.
As a matter of fact, partial takeovers in The reason why we follow the control
Germany and full takeovers in the United evolution for a six-year period subsequent
Kingdom bring about similar results in to the float can be found in panel B of Table
terms of control change (Goergen, 1998). 8.4.26 Control changes in U.K. IPOs take
Panel A of Table 8.4 shows that, after amal- place three to four years after the float,
gamating all control changes (both full and whereas in German IPOs they happen within
partial takeovers), the percentage (36% a four-to-six year period.
and 38%) is almost equal in German and Even though the percentage of German
U.K. IPOs. However, the distribution of and U.K. firms with a control change
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 207
Table 8.5 Voting Rights in Excess of 25%, Six Years After Flotation
This table is based on a sample of German and U.K. IPOs matched by market capitalization. The sample is
unbalanced, that is, if a firm is delisted prior to its sixth year after going public, the largest shareholder in the
last year prior to the delisting is reported.
(full and partial takeovers) may be roughly companies which are fully or partially
similar, the question whether the (ultimate) taken over, 14 are taken over by a bidder
bidder is widely held or is concentrated with dispersed control.
remains important. The potential agency Not only are there differences in the
problems in widely held (and hence man- evolution of control between Germany and
agement-controlled) firms are different the United Kingdom, but there are also
from those in closely held firms (Bratton differences in terms of who owns the votes.
and McCahery, 1995; Shleifer and Vishny, Table 8.5 shows the identity of the major
1997). For the former, the target firm will shareholder six years after the float for the
be monitored by the bidder’s management, size-matched sample. In German firms,
which may have objectives in conflict with families control two-thirds of the voting
the maximization of shareholder value. For rights and other companies (mainly foreign
the latter, it is the ultimate large share- ones) control about one-third. For U.K.
holder who is expected to monitor the tar- firms, other firms (mainly domestic ones)
get. Thus the objectives of the bidder may and families control 35% and 28% of the
differ depending upon who ultimately voting rights, respectively.
controls the bidder. For example, the
management of the bidder with dispersed
ultimate control may be more interested in 6. PREDICTION OF CONTROL IN
acquiring large, profitable, low-risk firms, GERMAN AND U.K. FIRMS
whereas the bidder with a strong
shareholder may prefer a more-risky, 6.1 Determinants of the evolution of
high-growth firm. Empirical evidence control
suggests that it is the ultimate shareholder
at the top of the ownership pyramid who Table 8.6 investigates the determinants
disciplines management rather than the of the evolution of control six years after
shareholders at the first ownership tier (see the float. We not only analyze the dilution
Renneboog, 2000). Panel C of Table 8.4 of stakes held by the initial shareholders,
differentiates between takeovers by an but also the evolution of the free float
ultimately dispersed or concentrated and the emergence of new large share-
bidder. The bidder in 52% of the German holders. All shareholdings which are
partial takeover targets in the size- directly or indirectly controlled by the
matched sample is ultimately closely held. same (ultimate) owner are aggregated.
The remaining German companies are Large U.K. companies (measured by
ultimately widely held. Out of the 24 U.K. market capitalization) tend to evolve
208 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 8.6 Determinants of the Evolution of Control Concentration in Recent German and U.K.
IPOs (Tobit Regressions)
Parameter estimates
(1) Initial (2) Free (3) New
shareholders float shareholder
This table shows the determinants of the concentration of voting equity 6 years after flotation for the
size-matched sample of German and U.K. IPOs using Tobit regressions. The dependent variables Free float and
New shareholder are the shares held by small shareholders and the proportion of shares held by new share-
holders (large shareholders who did not hold shares prior to the IPO), respectively. Stakes below 25% are
ignored.The dummy variable Country equals one for a German company and zero otherwise. Size is the natural
logarithm of the market capitalization at the closing price on the first day of trading and converted into 1985
pounds sterling. Founder equals one if the founder or his heirs held an equity stake in the firm immediately prior
to the IPO, and equals zero otherwise. Risk is measured by the standard deviation of monthly stock returns over
a five-year period subsequent to the IPO. Growth is the average annual growth rate of total assets over the first
five years after the float. Nonvoting is a dummy variable which is set to one if nonvoting shares were issued at
the time of going public, and zero otherwise. Profit is the annual cash flow standardized by the book value of
total assets. Age is the number of years since the registration of the firm. From the variables Age and Risk, the
average by country was subtracted in order to avoid multicollinearity. The p-values are shown in parentheses.
***, **, * represent statistical significance at the 1%, 5%, and 10% level, respectively. Industry and time
dummies are included.
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 209
toward a widely held equity structure as is higher than in firms with below average
wealth constraints put a limit on holding risk. Moreover, the free float in risky U.K.
large blocks of equity (column 2). However, firms is lower. High-risk German companies
these wealth constraints do not seem to be also require a strong new shareholder with
binding for German firms, as the market potentially strong monitoring skills
capitalization does not influence control (column 3). Still, column 2 also shows that
concentration (the parameter estimates of a high degree of risk also leads to a higher
0.162 and 0.189 largely cancel out). free float in German firms. Thus control
Larger U.K. firms also have fewer new concentration is related more strongly to
large shareholders, but this is not the case risk in the United Kingdom than in
for larger German firms in which new Germany, and this partially supports
shareholders control significantly larger Proposition 4.
equity stakes. Thus these findings support Although we do not find any impact of
Proposition 1, stating that there is a growth on the control of U.K. firms, strong
stronger negative relation between control growth in German firms leads to a reduc-
and corporate size for U.K. firms. tion in the initial shareholders’ control and
To control for the fact that the average control is transferred to new large share-
German firm is substantially older, we have holders. The findings support Proposition 5
subtracted the country average from Age. for the German firms, but not for the U.K.
Table 8.6 shows that the initial sharehold- ones. None of the U.K. firms had issued
ers of German firms with an above-average nonvoting shares versus 38% of the
age tend to retain smaller voting stakes German firms. (Nonvoting shares enable
than those of relatively older U.K. firms. the initial shareholders to raise additional
This supports Proposition 2. The control equity capital while maintaining control.)
relinquished by the initial shareholders in Table 8.6 confirms that the voting rights of
the relatively older German firms is not dis- the initial shareholders do not tend to be
sipated but is taken over by new large dissipated after the IPO when the company
shareholders (column 3). has nonvoting outstanding shares (column 1).
Table 8.6 also investigates the influence Thus both a transfer of control to a new
of the founder family on control six years large shareholder and a dissipation of con-
after the float. The table reports different trol are less likely (columns 2 and 3). This
results for U.K. and German IPOs. If the supports Proposition 6. Proposition 7
founder of a German firm is still a share- states that when sufficient cash flows are
holder at the IPO, control is likely to remain generated, there is a higher probability that
tight. Given that the initial shareholders the initial shareholder retains control.
retain large stakes, it is less likely for large Unlike Heiss and Köke (2001), we do not
new shareholders to acquire controlling find any such evidence in Table 8.6 and
stakes in German firms (column 3). In con- reject Proposition 7.27
trast, the presence of the founder family To conclude: in fast-growing, older
among the initial shareholders in the German firms, the initial shareholders sub-
United Kingdom has little impact on the stantially reduce their controlling share
degree of control exerted by the initial and stakes over the six years after the float,
new shareholders, although the likelihood provided they are not part of the founder
that the firm evolves toward widely held family. In contrast, the initial shareholders
ownership decreases. These findings sup- of German firms tend to retain control in
port Proposition 3. firms with nonvoting shares. New share-
Demsetz and Lehn (1985) found a nega- holders acquire control in larger, older, and
tive relation between risk and ownership strongly growing German firms in which
concentration for U.S. firms. Our findings the founder family is no longer involved and
support this picture for U.K. firms: in com- which have no nonvoting equity. A higher
panies with risk above the average 12.8%, free float tends to occur in highrisk German
the control held by new large shareholders firms. In U.K. firms, we find a strong
210 EQUITY OWNERSHIP STRUCTURE AND CONTROL
relation between control and size: the to reject Proposition 2. If the founder or
larger the firm, the lower is the control by his heirs are involved in the firm at the
the new shareholders and hence the larger IPO, the likelihood that they retain control
is the free float. When the founder does not over the six-year period subsequent to the
sell out at the float, the probability that IPO is large and hence the likelihood that
control is dissipated is low. High risk seems the firm evolves toward diffuse control is
to necessitate high control by the new large significantly reduced. This supports
shareholders.28 Proposition 3. The fact that founder
involvement at the flotation does not
6.2 Multinomial logit prediction of influence the probability of being taken
the state of control over provides some support for the predic-
tions of the models by Zingales (1995) and
Whereas in the previous section we investi- Mello and Parsons (1998) and is also
gated the determinants of the control levels consistent with the empirical findings for the
of initial and new shareholders, we now United States (Chung and Pruitt, 1996).
estimate the likelihood that six years after In companies with high risk, we expect
the float the following four different states the initial shareholders to reduce control in
of control arise: initial shareholder control favor of new large, stable shareholders with
(SC), diffuse control (WH), takeover by good monitoring abilities. Table 8.7 reveals
a closely held bidder (TC), or takeover by a that high risk significantly increases the
widely held bidder (TW). This analysis is probability of full and partial takeovers
performed for two reasons. First, it sheds relative to the possibility that the firm
some light on how control concentration remains controlled by the initial sharehold-
comes about (via different types of ers (panel A) and relative to the state of
takeover, or via control retention). Second, diffuse ownership (panel B). Panel C also
the Tobit models estimated in the previous reveals an interesting result: high risk
section (Table 8.6) assume that the increases the likelihood of a takeover by a
dependent variables (control held by initial concentrated bidder rather than by a widely
or new large shareholders or diffuse con- held firm. So it seems that tight control
trol) are independent, which is not the with increased monitoring is the best solu-
case. Therefore we estimate multinomial tion to high risk. This contradicts
logits predicting the control state.29 Proposition 4 and the predictions of Bolton
Table 8.7 confirms that there is a higher and von Thadden’s (1998) model. While we
probability for large firms to end up with hypothesized that Proposition 4 would hold
diffuse control (panel B). This supports more strongly for the U.K. firms, we find no
Proposition 1. In addition, large firms are differences in the risk-control relation
more likely to be taken over by a widely between the United Kingdom and Germany.
held firm than by a concentrated bidder. High growth could lead to control dilu-
This is expected for two reasons. First, a tion resulting from the limited availability
concentrated bidder (a family or individ- of internally generated funds and the
ual) is more likely to be wealth constrained implied need to attract external capital to
than a widely held firm or may be less able finance investment opportunities.30 Table 8.7
to raise external funds to finance the confirms Proposition 5: both diffuse
control acquisition. Second, the reason for control and a takeover by a widely held
a takeover by a widely held firm may be company are the more likely “equilibrium”
agency related: the management of a states six years after the float for growth
widely held firm may be tempted to “build companies. This proposition is expected to
an empire” when it is insufficiently moni- hold more strongly for the U.K. firms than
tored (Berkovitch and Narayanan, 1993; for the German ones. Panel C of Table 8.7
Goergen and Renneboog, 2003). shows that this is the case: high growth
Age does not seem to matter for either triggers more partial takeovers, but not
country’s control levels such that we have necessarily by widely held bidders.The reason
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 211
Log (PWH /Psc) Log (PTC /PSC) Log (PTW /PSC) Log (PTC /PWH) Log (PTW /PWH) Log (PTW /PTC)
This table shows the probability of control for a size-matched sample of German and U.K. firms using a multinomial logit model with four possible states of control: SC (control
remains in the hands of the initial [pre-IPO] shareholders), WH (the firm is widely held [no shareholder owns a stake of more than 25%]), TW (the firm is taken over by a
widely held company), and TC (the firm is taken over by a concentrated shareholder).The dummy variable Country equals one for a German company and zero otherwise. Size
is the natural logarithm of the market capitalization at the closing price on the first day of trading, converted into 1985 pounds sterling. Founder equals 1 if the founder or
his heirs held an equity stake in the firm immediately prior to the IPO, and equals zero otherwise. Risk is measured by the standard deviation of monthly stock returns over a
five-year period after the IPO. Growth is the average annual growth rate of total assets over the first five years after the float. Nonvoting is a dummy variable capturing the
issue of nonvoting shares at the time of going public (dummy equals 1). Profit is the annual cash flow standardized by the book value of total assets. Age is the number of
years since the registration of the firm. From the variables Age and Risk, the average by country was subtracted to avoid multicollinearity. The p-values are in parentheses.
***, **, * represent statistical significance at the 1%, 5%, and 10% level, respectively. Industry and time dummies are included in the model.
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 213
APPENDIX
Country Size Founder Risk Growth Nonvoting Profit Age Riskminuscountryavg Growthminuscountryavg Ageminuscountryavg
Country 1
Size 0.026 1
(0.483)
Founder 0.121 0.081 1
(0.133) (0.139)
Risk 0.533 0.117 0.016 1
(0.000) (0.135) (0.296)
Growth 0.154 0.099 0.177 0.070 1
(0.080) (0.286) (0.053) (0.244)
Nonvoting 0.541 0.152 0.094 0.281 0.009 1
(0.000) (0.172) (0.123) (0.006) (0.478)
Profit 0.191 0.069 0.099 0.250 0.056 0.180 1
(0.291) (0.367) (0.277) (0.008) (0.154) (0.194)
Age 0.342 0.003 0.176 0.098 1
214 EQUITY OWNERSHIP STRUCTURE AND CONTROL
This table shows the Pearson correlation coefficients of the determinants of the concentration of ownership six years after the float. The dummy variable Country equals one
for a German company and zero otherwise. Size is the natural logarithm of the market capitalization at the closing price on the first day of trading and converted into 1985
pounds sterling. Founder equals one if the founder or his heirs held an equity stake in the firm immediately prior to the IPO and equals zero otherwise. Risk is measured by
the standard deviation of monthly stock returns over a five-year period subsequent to the IPO. Growth is the average annual growth rate of total assets over the first five years
after the float. Nonvoting is a dummy variable capturing the issue of nonvoting shares at the time of going public (dummy equals 1). Profit is the annual cash flow standardized
by the book value of total assets. Age is the number of years at the IPO since the creation of the company. Riskminuscountryavg Growthminuscountryavg, and Ageminuscountryavg are the
respective Risk, Growth, and Age variables of which the country average is subtracted. The statistical significance is in parentheses.
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 215
most players are familiar with. Internal rent of the 1980s and computer and software
seeking by parties who participate in corporate manufacturers were first assigned to group 19,
control may also explain why such parties would then to group 69, and only later to group 35.
attempt to impede change toward a more 19. Market capitalization is adjusted for U.K.
efficient control structure. For example, the inflation by the annual GDP deflator (base year
management of widely held companies may 1985) of the International Monetary Fund
prefer to retain a diffuse control structure, as (IMF). Several German firms in our sample
this enables them to maintain their private ben- have dual-class shares of which one class is not
efits at the expense of shareholders. Likewise, listed. The market capitalization for these firms
Bebchuk (1998) argues that concentrated was computed by multiplying the total number
ownership—and hence uncontested corporate of shares by the market price of the listed class
control—prevails in continental Europe because of shares.
the lax corporate-governance regulation allows 20. It may be argued that a different type of
large shareholders to reap substantial private matching based on firm age should also have
benefits of control. Hence, as long as continental been performed. However, a reasonable match
European regulation does not change, control (plus or minus two years of difference) could
concentration will resist change (Bratton and only be found for about 19 German firms.
McCahery, 1999). However, we control directly for age in the
13. See the discussion on minority protection regressions. Similarly we tried to match firms
in Section 2. Evidence of the extraction of pri- simultaneously by size and industry. However,
vate benefits of control by large shareholders in again for more than three-quarters of the firms,
Germany is given by Franks and Mayer (2001), a satisfactory match within the 25% range
and by Dyck and Zingales (2001) in an interna- could not be found.
tional comparison. 21. The German size-matched sample includes
14. Brennan and Franks (1997) show that two companies which are not included in the
control retention is an important reason for industry-matched sample, because there were no
underpricing: share-rationing schemes enable U.K. IPOs in these specific industries during the
the original shareholders to disperse shares to period of study. The German industry-matched
atomistic subscribers. sample comprises six firms not included in the
15. This argument assumes that the voting size-matched sample.
share prices in both companies are equal. 22. As log PWH /PSC log PSC/PWH, we
16. During the period of the study, lower-tier report results for (i) log(PWH/PSC),
markets were the Unregulated Unofficial Market log(PTC/PSC), log(PTW/PSC), (ii) log(PTC/PWH),
and the OTC for Germany and the Third Market log(PTW/PWH), and (iii) log(PTW/PTC).
and OTC for the United Kingdom. 23. Franks and Mayer (1998) report that
17. Control and ownership concentration of hostile takeovers have been a rare phenomenon
61 of these 80 German firms could be traced in Germany. Since World War II and prior to
reliably over time. For most of the other IPOs, 2000, there have only been three attempts, two
the identity of the shareholders was available, of which failed. Since then there has been a
but not the exact size of their holdings. fourth successful hostile takeover (the
18. The industry classification is based on the Mannesmann takeover by Vodaphone in 2000).
two-digit U.K. SE groups, the industrial classifi- 24. Whereas panel A considers only direct
cation used by the London Stock Exchange to holdings, panel C takes both direct and indirect
compile its quarterly lists of new issues. The holdings into account.
groups are covered by the F.T. Actuaries 25. The Takeover Panel may grant an excep-
Investment Index Classification with the amal- tion to this rule, which makes a tender offer
gamation of certain related groups. For each mandatory to a party which reaches the 30%
German firm the industry description at the time threshold of equity (or voting rights), in the case
of the IPO in the Satling Aktienführer was where a shareholder takes a large stake in a
recorded. Subsequently German firms were financially distressed company.
reclassified into two-digit U.K. SE groups. We 26. Expanding the analyzed time period of six
have merged specific groups which only had a years subsequent to the IPO to eight years
small number of IPOs: for example, groups 27 changes neither the data presented in Table 8.3
(Misc. Mechanical Engineering) and 28 nor the results of this article.
(Machine and Other Tools) were merged. Groups 27. As a robustness check, similar Tobit
19 (Electricals), 35 (Electronics), and 69 regressions but with alternative independent
(Office Equipment) were also merged since variables (for the definition, see Section 4.2)
groups 35 and 69 did not exist at the beginning were estimated. The country-dummy variables
218 EQUITY OWNERSHIP STRUCTURE AND CONTROL
confirm that the share stakes of initial discussed above are robust to alternative
shareholders are higher in Germany, whereas variable specifications.
the percentage controlled by new large share- 28. All the results in this section—apart from
holders is higher in the United Kingdom (as a the conclusions concerning corporate growth—
result of the more frequent occurrence of full are also valid for the industry-matched sample.
takeovers). Firm size (book value of total These results are not reported in the article, but
assets) is positively related to the control held are available upon request. The only difference
by small shareholders and by new large share- in results is that the Growth coefficient is not
holders, which confirms the findings of this significantly different from zero at the 10%
section. We find that, if the founder is on the level for the industry sample. This is probably
management board (Vorstand) of a German due to the fact that firms matched by industry
firm, control by the initial shareholders is less have more comparable growth rates than those
likely to be reduced over the six years after the matched by size.
IPO. These initial shareholders retain a higher 29. It should be noted that the state of control
percentage of the voting rights than those in is based on ultimate control. The predictive
firms without founder involvement. As power of these models with control based on
expected, a higher retention rate goes hand in direct shareholding was lower than the one
hand with a reduced presence of new share- based on ultimate control. Statistical signifi-
holders. Growth, in contrast, is strongly related cance for the direct ownership models was
to control concentration. High growth in total always worse for the case of the size-matched
assets leads to a reduction in the stakes of the sample, and worse or similar for the industry-
initial shareholders in the United Kingdom and matched sample.
even more so in Germany. Furthermore, a 30. All results in this section—apart from
higher proportion of nonvoting shares in the conclusions concerning corporate growth—
Germany allows the initial shareholders to are also valid for the industry-matched sample.
retain their control. In contrast to the findings These results are not reported in the article,
presented in Table 8.6, we find different results but are available upon request. The only differ-
for our risk and profitability measures. We do ence in results is that the Growth coefficient is
not find a relation between the risk measured as not significantly different from zero at the
the standard deviation of cash flows over the 10% level for the industry sample. This is
five-year period after the IPO and control. This probably due to the fact that firms matched by
may be due to the fact that our original risk industry have more comparable growth
measure (based on the volatility of stock rates than those matched by size (see also
returns) is the better one. Whereas the prof- Goergen, 1999).
itability measure (cash flow/book value of total 31. This finding is confirmed by a study on
assets) was not related to ownership concentra- hostile takeovers in the United Kingdom by
tion in Table 8.6, the alternative profit rate Franks and Mayer (1996).
(cash flow over market value of equity plus
book value of debt) shows a statistically signif-
icant relation with control concentration. These REFERENCES
results support the proposition by Burkart,
Gromb, and Panunzi (1997) that different
states of the world require different control Baums, T. 1997. “Shareholder Representation
structures. In some states of the world (such as and Proxy Voting in the European Union:
the one with low profitability), a major, stable A Comparative Study,” working paper,
shareholder is needed, while in other cases a University of Osnabrück.
dispersed share structure provides the neces- ——. 2000.“Corporate Governance in Germany—
sary managerial discretion. High profitability in System and Current Developments,” working
the United Kingdom ensures that the initial paper, University of Osnabrück.
shareholders retain control and consequently ——, and C. Fraune. 1995. “Institutionelle
that the voting blocks held by new shareholders
Anleger und Publikumsgesellschaft: Eine
in the more profitable firms are smaller. For
Germany, in contrast, we find the opposite Empirische Untersuchung,” 40 Die
result: initial shareholders seem to consider Aktiengesellschaft 97–112.
high profitability as an opportunity to transfer ——, and R. Schmitz. 2000. “Shareholder
control stakes to new large shareholders. We Voting in Germany,” working paper 76,
conclude that the findings of Table 8.6 University of Osnabrück.
WHY ARE LEVELS OF CONTROLS DIFFERENT IN GERMANY AND UK? 219
ABSTRACT
In this paper we analyze the monitoring role of occupational pension funds in the UK. We argue
that because of their objectives, structure and overall share holding, occupational pension
funds are likely to have more incentives to monitor companies in which they hold large stakes
than other financial institutions. By comparing companies in which these funds hold large
stakes with a control group of companies listed on the London Stock Exchange, we show that
occupational pension funds hold large stakes over a long-time period mainly in small
companies. However, the value added by these funds is negligible and their holdings do not lead
companies to comply with the Code of Best Practice or outperform their industry counterparts.
Overall, our results suggest that occupational pension funds are not effective monitors.
For example, while Strickland et al. that could be used to finance growth
(1996) show that monitoring by share- opportunities (e.g., Hutton, 1995).4 In
holders enhances firm value, Karpoff et al. contrast, other studies show that pension
(1996) do not find evidence that share- funds do not get involved in corporate
holder proposals increase firm value or monitoring because they find it easier and
influence firm policies. cheaper to sell their holdings, they do not
The purpose of this paper is to extend want to sit on the board for fear of getting
previous research that documents the price sensitive information or because of
impact of large holdings on corporate the agency problems within the funds
performance by analyzing the monitoring themselves.5 At the same time, policy
role of occupational pension funds in the makers in the UK tend to rely on these
UK. We identify separately these funds institutions to promote corporate governance
from other financial intermediaries (e.g., Cadbury, 1992; Greenbury, 1995).6
because of the large dimension of their Second, previous studies were mainly
overall stakes in the UK market, the undertaken under the US framework.
particular structure of their portfolios and Franks and Mayer (1997) show that,
their investment objectives. As defined by despite the fact that the US and the UK
Brickley et al. (1988; 1994), occupational governance systems are both market-
pension funds are typical pressure-resistant based, the two countries differ in two major
institutions as opposed to other institu- respects: the US has more quoted compa-
tions, such as banks, investment and unit nies than the UK and, while the largest
trusts and insurance companies, which are category of shareholders in the UK is
pressure-sensitive. Unlike previous studies pension funds, most of the equity in the US
(e.g., Strickland et al., 1996; Bethel et al., is held by individuals (if each different type
1998) we do not analyze block purchases of institutional shareholder is treated as a
because of event date uncertainty and we different category). In addition, unlike US
do not concentrate on formal targeting by pension funds where investments and
pension funds because we could not find activism programs are developed and
any event where occupational pension implemented by fund staff then overseen
funds in the UK target companies, i.e., and approved by trustees (Del Guercio and
make particular proposals at the Hawkins, 1999), the UK pension funds,
company’s annual meetings or negotiate despite the size of their holdings, are not
privately some corporate governance issues known for their monitoring and hardly vote
(as in Carleton et al., 1998). Instead, we at the annual general meetings (NAPF,
compare firms in which occupational 1996b; Financial Times, 1999). Thus, the
pension funds hold large stakes against a testing of the empirical hypotheses in a
control group of companies listed on the pension fund dominated market such as the
London Stock Exchange and test the UK where corporate governance issues are
hypotheses that monitoring increases with debated and companies suffer from the
ownership concentration and, as a result, same free cash flow problems as their US
these funds reduce agency conflicts and counterparts, will strengthen the evidence
lead companies to better performance. provided to-date.7 Third, since our analysis
Further research in this area is is centered on pension funds that are tax
warranted for a number of reasons. First, exempt and hold large stakes, the evidence
the issue of involvement of pension funds in we provide should be of relevance to tax and
the running of companies is controversial. market regulators and to policy makers
The popular belief is that pension funds are involved in the growth of the UK economy.
short-termists and impose their views We construct from the financial
on companies in which they invest. In statements of all UK quoted non-financial
particular, given their tax-exempt status, companies a test sample of companies in
UK pension funds are criticized for which pension funds hold more than 3% of
making companies pay high cash dividends the issued share capital and a control sample
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 223
by matching our test firms by industry and pension funds. A large proportion of these
size.8 We find that pension funds hold large holdings is concentrated in the portfolios of
stakes in 289 out of 1640 (18%) compa- large funds. In this section we review the
nies. These holdings are mainly in small literature on the role of large shareholders,
companies and they did not change signifi- discuss the literature on pension funds
cantly over the 1992–1996 sample period. activism, and analyze the structure of the
We show that our test companies are not UK occupational pension funds, the impor-
more likely to restrain management com- tance of equities in their portfolios, the
pensation and/or adopt the Code of Best relatively high concentration of their indus-
Practice, i.e., split the roles of chairman try and the management approaches and
and CEO, have more non-executive direc- objectives.
tors and/or narrow the size of their board
than our control firms. These results are
not consistent with the recommendations 2.1. Review of the literature
of Cadbury (1992) and Greenbury (1995). Corporate governance deals with how
In addition, we find that our test companies companies are managed in the long-term
are not more profitable and do not pay interest of their shareholders.The literature
higher dividends than the control firms has identified two main corporate governance
despite the tax credit pension funds could systems that predominate in the developed
claim during the sample period. We report economies: the market-based systems of
weak and even negative relationship the UK and the US characterized by liquid
between occupational pension funds block- markets and unconcentrated company own-
holdings and firm value, and, over a longer ership, and the relationship-based systems
time period, our test firms do not overper- of Japan and Germany where ownership is
form their peers. Our overall results are concentrated and markets are relatively
consistent with previous US evidence (e.g., illiquid.9 The issue, although not trivial, has
Romano, 1994 and Wahal, 1996) and cast been considered in the literature only
doubt on the effectiveness of UK occupa- recently when agency theorists argue that
tional pension funds’ monitoring role. At public corporations suffer from excessive
the same time, our results provide support costs as managers pursue their own
to the proposition of Coffee (1991) and interests rather than the interests of share-
imply that pension funds do not follow an holders (e.g., Jensen, 1986). As a result,
‘exit’ policy which is increasingly more there is a need for setting up mechanisms
expensive because they must accept to make managers maximize shareholder
substantial discounts in order to liquidate wealth. These mechanisms include share-
their holdings. Thus, once “locked in”, holding of managers, intermediaries and
occupational pension funds avoid costly large blockholders (McConnell and
monitoring and refrain from selling their Servaes, 1990; Morck et al., 1988), out-
large stake for fear of losing from large side directors (Cotter et al., 1997), debt
discounts and/or conveying information to policy (Lasfer, 1995; McConnell and
the market. Servaes, 1995; Lang et al., 1996), the
The rest of the paper is structured as market for corporate control and incentive
follows. Section 2 presents the theoretical contracts (Hart and Holmstrom, 1987;
background. Section 3 describes the data. Hart, 1995), large intermediaries
Section 4 presents the results. Conclusions (Diamond, 1984; Admati et al., 1994), and
are in Section 5. long-term relationships (Ayres and
Cramton, 1993).
In theory, Diamond (1984) suggests that
2. THEORETICAL FRAMEWORK a large intermediary can represent a better
solution to agency conflicts because of
In recent years, more than a third of all economies of scale and diversification.
listed equities in the UK are held by Admati et al. (1994) argue that when
224 EQUITY OWNERSHIP STRUCTURE AND CONTROL
monitoring is costly, the intermediary will of pension funds. They find that, unlike
monitor only if this will result in a modifi- proposals sponsored by externally-managed
cation in firm’s payoff structure and lead to funds, those made by internally managed
net gains. When the intermediary does not funds are not associated with general
hold all the firm’s equity and the transac- increases in governance-related events at
tion costs are not excessive, the level of target firms. Other studies that looked at
commitment will be sub-optimal even when the characteristics of companies in which
optimal risk-sharing is attained. Maug institutions hold large stakes find that the
(1998) extends this analysis and argues relationship between such holdings and
that liquid markets reduce large shareholders firm value or accounting rates of return is
incentive to monitor because they can sell weak (e.g., Demsetz and Lehn, 1985;
their holdings easily, but such markets Agrawal and Knoeber, 1996).
make corporate governance more effective In sum, the primarily US-based studies
as it is cheaper and easier to acquire and provide mixed results on the monitoring
hold large stakes. Kahn and Winton (1998) role of pension funds. The testing of these
distinguish between liquidity, speculation hypotheses under a different institutional
and intervention and argue that interven- framework is, thus, warranted.
tion is a function of the size of the
institution’s stake, firm specific factors
and institution’s trading profit. Shleifer 2.2. Institutional settings
and Vishny (1997) and Agrawal and In this section we describe the pension funds
Knoeber (1996) suggest that large industry and the corporate governance
investors, because of the relevance of the system in the UK and set up the hypotheses.
resources invested, have all the interest and
the power to monitor and promote better
governance of companies. 2.2.1. The UK pension fund system
Previous empirical studies show that
institutions behave differently from individ- The UK pension fund system includes, in
uals in sponsoring initiatives (Jarrell and addition to the public pension scheme,
Poulsen, 1987; Karpoff et al., 1996) and occupational pension schemes, which are
that the institutional behavior is not homo- organized and sponsored by employers, and
geneous as it depends on the sensitivity to individual pension schemes offered by
managerial pressure (Brickley et al., 1988; financial institutions.The occupational pen-
Gordon and Pound, 1993). However, they sion schemes are usually defined-benefits
disagree on the effectiveness of share- (DB) where the amount of benefits relates
holder activism.10 In particular, Wahal to the final salary of the member while
(1996) and Karpoff et al. (1996) find little individual pension schemes are defined-
evidence that operating performance of contribution (DC) where pension benefits
companies that are the target of pension depend on the contributions paid during the
funds proposals improves. These results are working life of the members and the
consistent, among other things, with the returns realized on the investment (Blake,
arguments of Murphy and Van Nuys 1995).This difference has significant impli-
(1994) and Romano (1994) that pension cations on these two schemes’ investment
funds are not effective monitors because of policies. With defined contribution plans,
the agency problems within the funds them- individuals bear the investment risk and
selves. In contrast, Nesbitt (1994) and require a more cautious investment strat-
Smith (1996) find that companies targeted egy. As a result, the proportion of shares
by large pension funds, such as CalPERS, relative to total assets of occupational pen-
increase significantly their performance. sion schemes ranges between 70% and
More recently, Del Guercio and Hawkins 80% while defined contribution pension
(1999) show that the monitoring effective- plans usually hold no more than 25–30%
ness depends on the investment strategies of shares (NAPF, 1996b). In 1997 the
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 225
overall value of individual pension schemes institutional activism are weaker in the UK
assets was £190 billion (ABI, 1998) because active shareholders that hold a
while occupational pension funds assets “block of shares” are not subject to any
(including insured schemes) reached £635 filing requirements, such as the 13D Form
billion.11 with the SEC in the US, and they cannot be
The proportion of UK pension funds sued for breaching any duty of disclosure of
assets invested in equities is the highest their plans or proposals.This rule applies in
amongst OECD countries (Davis, 1995). the US to shareholders who act together
For example, in 1993, 78% of assets were (e.g., in the case of co-ordinated activism),
invested in equities, 12% in fixed income and to investors who, individually or jointly
securities and the remaining in cash and hold 5% or more of a firm’s equity (Black,
property (Business Monitor, 1997). This 1998). Moreover, unlike in the US, UK
trend is likely to reflect the overall invest- pension funds have no legal duty to vote.
ments of all pension funds as Blake (1995) The pension fund industry is highly
shows that asset structure is not signifi- concentrated. For example, in 1994, the
cantly associated with the size of pension largest five in-house managed occupational
funds assets. The preference of equities pension funds managed assets worth £65.8
over fixed income securities can be related billion, 14.8% of all occupational pension
to the tax-exempt status of pension funds funds assets (NAPF, 1996a) and British
who, like charities, are not subject to Telecommunications, accounted for £17.2
capital gains tax and claim back the tax billion. The largest 68 schemes, whose
credit, referred to as the advanced corpora- assets value exceeds £1 billion in 1995,
tion tax, when they receive dividends. accounted for 57.3% of all occupational
Lasfer (1996) shows that this tax credit pension funds assets (Pension Funds and
discriminates in favor of dividends relative their Advisers, 1996). The industry of fund
to capital gains. managers is also highly concentrated. At
In contrast, US pension funds invest a the end of 1996, the top 20 segregated
lower proportion of their assets in equities. fund managers managed assets worth
For example, in 1990, out of the total £285.7 billion on behalf of occupational
assets of $2,491 billion, 38.6% are pension funds and, as a whole, managed
invested in equities (Charkham, 1995). assets of some £1029.2 billion (Financial
Davis (1995) shows that the proportion of Times, 1997a).
assets invested in equities has not changed Over the last three decades, the
over the 1970–1990 period. However, these aggregate share ownership in the UK has
investments are not evenly distributed but changed substantially. While in 1963
concentrated mainly in large firms individuals were the main shareholders
(Charkham, 1995; Stapledon, 1996; with 58.7% of all UK listed equities and
Brancato, 1997). pension funds held only 7%, by 1993
In managing these assets, the UK pension funds stakes increased to 34.7%
pension funds are subject to trust law and (London Stock Exchange, 1995). In
implicitly follow the prudent-man concept. contrast, in the US, individuals held 50% in
There is no explicit prudent-man rule and 1990 followed by pension funds with
the pension trust law is very flexible. 20.1%, increasing to 25.4% in 1995
However, the duty of prudence to trustees (Prowse, 1994; Brancato, 1997).
can be interpreted as requiring the pension We use our sample firms (detailed
funds money to be invested for the sole below) to analyze further the ownership
benefit of the beneficiaries. The recent leg- structure of UK companies. The results,
islation for pension fund management reported in Figure 9.1, show that managers
(e.g., the 1995 Pension Fund Act) is simi- and block shareholders (including occupa-
lar to the American Employee Retirement tional pension funds) own 53% of equity,
Scheme Act with a basic view towards implying a rather concentrated ownership
prudence. The legal barriers against structure. However, within these blocks, no
226 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Dispersed Managers +
shareholders blockholders
46.98% 53.20%
Managers Blockholders
26.91% 73.09%
Figure 9.1 Structure of block share ownership in our sample firms. Dispersed shareholders are
defined as those owners who individually own less than 3% of ordinary shares, excluding
managers. Blockholders are owners who individually own at least 3% of ordinary
shares (excluding managers). Non-institutional blockholdings are block shareholdings
attributable to individuals and families, non-financial companies, public bodies, foreign
investors and nominees. Institutional blockholdings refer to block shareholdings held by
financial institutions. Internally managed occupational pension funds are those pension
funds whose assets are, wholly or in part, managed directly by the fund’s trustees.
Externally managed pension funds holdings refer to block shareholdings attributable to
the largest 20 segregated pension fund managers, times the incidence of pension fund
assets to the total asset managed by the segregated fund. Managed funds (for clients
other than OPFs) refer to the residual proportion of block shareholdings managed by
the top 20 segregated pension fund managers, once accounted for the assets managed on
behalf of occupational pension funds (previous variable). Finally, all other institutional
holdings are computed as residual item, and refer to block shareholdings held by
merchant banks, unit and investment trusts, insurance companies, venture capital firms,
and all other financial institutions. All percentages reported are computed assuming the
relative upper line to be equal to 100%.
single class of shareholders clearly dominates funds account for 14% of all financial
the others. Financial blockholders account blockholdings as they hold an average of
for 62% of block shareholdings by holding 3% of equity. Externally managed occupa-
24% of our firms’ equity, while the remain- tional pension funds account for 9% by
ing 38% are attributable to individuals and holding 2% of equity. The assets managed
families, non-financial companies, public by external pension fund managers on
authorities, foreign investors and nominees. behalf of clients other than pension funds
Internally managed occupational pension represent 28% of these stakes, i.e., 7% of
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 227
equity. Finally, all other financial institu- nominating committee which is responsible
tions (including merchant banks, insurance for making recommendations for board
companies, unit and investment trusts) membership, they should be the sole or
wholly account for 48% of institutional majority members of the remuneration
holdings as they hold about 12% of equity. committee which makes recommendations
to the board on the pay of executive direc-
2.2.2. The UK corporate governance tors, and of the audit committee whose
system function is to advise on the appointment of
auditors, to insure the integrity of the com-
The importance of corporate governance in pany’s financial statements and to discuss
the UK has been emphasized by recent with the auditors any problems arising
concerns about the way in which remuner- during the course of the audit.13
ation packages for senior executives have The report has also highlighted the
been determined, the spectacular collapse responsibilities of institutional investors
of a number of large companies and by the such as pension funds and suggests that such
fraudulent use of the pension fund of institutions should be encouraged to make
Mirror Group Newspapers to finance an greater use of their voting rights and to seek
illegal scheme for supporting the share contacts with companies at a senior executive
price of Maxwell Communications level. In particular, they should monitor
Corporation. These cases have highlighted boards where there is a concentration of
instances where directors do not act in the power in the hands of the chief executive,
best interest of shareholders. seek to promote the influence of non-executive
In the train of these and other scandals, directors and they are expected to bring
the Committee on the Financial Aspects of about changes in underperforming compa-
Corporate Governance (referred to as the nies rather than dispose of their shares.
Cadbury Committee after its chairman) Although the report recognizes that closer
was set up to look at the changes needed in relations with managers can result in these
corporate governance in the UK and pub- investors gaining price sensitive information
lished a report in December 1992 which makes them insiders, it does not go so
(Cadbury, 1992). The report is similar to far as to recommend the formation of share-
the Statement on Corporate Governance holders’ committees and the participation of
released in September 1997 in the US by shareholders in the appointment of directors
the Business Roundtable, an association of and auditors.
CEOs of large companies. At the heart of This emphasis on institutions is driven by
the report is the Code of Best Practice the size of their holdings but also by the
which details the role and composition of fact that these institutions, such as pension
the board of directors, the appointment funds, do not target companies and they
of non-executive directors, the disclosure of rarely caste their vote at the annual general
the remuneration of executive directors and meetings, making them the object of public
the renewal of their contracts, and the way criticism (e.g., Mallin, 1997). This passive
companies should report and audit their stance has not changed even after the
accounts.The main recommendation is that advent of the Cadbury Report. More
the offices of the chairman and the chief recently, commenting on the first major cor-
executive officer should be separated to porate governance action by pension funds,
prevent excessive concentration of power in namely the ousting of the head of Mirror
boardrooms and that companies should Group, a media company which has under-
appoint independent non-executive direc- performed the market by 34% over the last
tors with high caliber so that their views 5 years, the Financial Times (1999) wrote:
will carry weight in board discussions.12
The code defines the various roles non- Unlike in the US, investor mutinies in
executive directors should play. For example, Britain are still a relatively rare event.
they are to be in a majority on the Complacency, reflected in the dismal
228 EQUITY OWNERSHIP STRUCTURE AND CONTROL
they do not want to gain access to price invested in these companies.We expect large
sensitive information or if they themselves pension funds and those with significant
are subject to agency problems. In addition, commitments to have a higher incentive to
given that these funds are defined benefits monitor companies in which they hold large
schemes, they are likely to be indexed and stakes. In the long run, companies in which
passive to minimize their management risk, occupational pension fund carry on holding
transaction costs and to fit the needs of large stakes are expected to outperform
long-term pension investors (Tomlinson, their industry counterparts.
1998). Del Guercio and Hawkins (1999)
argue that such passive management style
will lead pension funds to monitor by 3. SAMPLE CONSTRUCTION AND
promoting spill-over effects that boost the DEFINITION OF PROXY VARIABLES
performance of the stock market overall
rather than specific stocks. In our analysis We search all the 1640 non-financial
we focus on internally-managed pension companies quoted in the London Stock
funds to test whether size and passive Exchange in 1995–1996 for those where
management styles define the level of mon- occupational pension funds hold large
itoring activity but we control also for the stakes. We exclude financial companies
monitoring roles of external pension funds because of the specificity of some of their
and other blockholders. ratios such as leverage, which cannot be
The monitoring activity can take the related to the level of their risk and/or
form of selecting the board structure resolution of their agency conflicts.To avoid
and/or improving performance. The moni- survivorship bias, our sample includes all
toring of the board relates to the size of the companies for which the relevant data is
board and its composition so that it available even if they are currently
becomes more accountable to the share- extinct.17
holders (Cadbury, 1992).15 Jensen (1993) We started by gathering information on
argues that as the size of the board shareholding by category from the annual
increases, its ability to control manage- reports but we find that only a handful of
ment decreases and the communication companies disclose this information in their
and co-ordination problems increase. Analysis of Ordinary Shareholdings section
Consistent with this proposition, Yermack of their accounts. This data would have
(1996) and Eisenberg et al. (1998) find a been ideal but it would have given us only
negative correlation between board size the aggregate holding of pension funds
and firm performance.Thus, if occupational without distinguishing between funds that
pension funds are effective monitors, they are internally managed from those that are
are expected to mitigate this agency con- externally managed. Instead, we rely on any
flict by restricting the size of the board. At disclosed holding above 3% threshold in
the same time, they will lead companies to the accounts and reported in Extel
adopt the Code of Best Practice defined by Financial18 and define these as occupational
Cadbury (1992), i.e., to split the roles of pension funds holdings or blockholders.
chairman and chief executive officer, to All the holdings of directors are disclosed
have a high proportion of non-executive even if they are zero (Company Act 1985).
directors and to restrain executive pay.16 Although this cut off point of 3% is
The monitoring of firm performance constraining, it is the only data available.
implies that companies in which occupa- We, nevertheless, posit that the holding of
tional pension funds hold large stakes have 3% or above is significant to warrant
a higher value than widely held companies monitoring and to allow us to test directly
and/or companies held by other blockholders. the Diamond (1984) and Admati et al.
This level of monitoring is a function of the (1994) propositions. We searched for the
size of the pension funds and the relative keywords pension, pension funds, retire-
importance of the amount of assets ment, superannuation and superannuation
230 EQUITY OWNERSHIP STRUCTURE AND CONTROL
schemes and for known pension funds, such As in Agrawal and Knoeber (1996), we
as Hermes, to define the holdings of compute Tobin’s Q as the ratio of market
occupational pension funds. We identified value of equity plus book value of debt over
289 individual companies (18% of our total assets to measure firm value. The
total sample) with at least one occupational results are simulated using other measures
pension fund holding above 3%, and 356 of firm value such as market-to-book, M/B,
large stakes held by 99 individual occupa- and market-to-sales, M/S, ratios (Lins and
tional pension funds. We split the other Servaes, 1999). These variables are, how-
major disclosed holdings (other than man- ever, ambiguous measures of value-added
agerial holdings) into externally managed by pension funds investments, since they
pension funds, institutional and non- can also capture the value of future invest-
institutional blockholding depending on the ment opportunities. As in Yermack (1996),
identity of the shareholders. We compare we control for growth opportunities by
the performance of our test firms against using P/E ratio. In addition, we use various
all other remaining companies and against accounting rates of returns and a one year
a control group of companies with similar share price return, Ri, t 12 to t’ to measure
size and industry characteristics. We performance and market value of equity,
collect all the relevant accounting and ME, or total assets, TA, to control for size.
financial data from Extel Financial and We follow previous literature (e.g., Lang
from each company’s accounts. et al., 1996) and use both the market value
Table 9.1 reports the proxy variables of leverage, Mlev, defined as long-term debt
used to test our hypotheses. We use five over market value of equity plus long-term
definitions of pension fund holdings: (i) the debt, and the book value of leverage, Blev,
first largest stake held by pension funds, defined as long-term debt over shareholders
LPF; (ii) the sum of all pension fund funds plus long-term debt, to test for the
holdings, TPF, to analyze pension funds’ monitoring role of debtholders. We use
individual and collective monitoring roles; Split (a dummy variable equal to 1 if the
(iii) the pension fund incidence, IPF, the role of chairman and CEO are differenti-
ratio of pension fund investment in our test ated), Nechair (a dummy equal to 1 if the
sample over their total assets to assess the position of chairman is covered by a non-
magnitude of such investment in their executive director), number of directors,
portfolio;19 (iv) the number of occupational #DIR, the proportion of non-executive
pension funds in our test firms, NPF, and directors in the board, %NED, and the rel-
(v) the logged value of the largest pension ative remuneration of directors, Dir
fund’s asset, LNPFA. Directors’ ownership, rem./Sal, to assess the pension fund role in
Dir, is used to control for the managerial monitoring board composition and com-
entrenchment hypothesis.20 Any other pensation policy. Finally, payout ratio is
large stake, Block, is used to control for the used to test the hypothesis that pension
monitoring role of blockholders. This funds prefer to invest in companies that
variable is split into institutional block pay high dividends.
shareholdings, IBlock, which includes all
financial institutions’ stakes (excluding
internally- and externally-managed occupa- 4. EMPIRICAL RESULTS
tional pension funds’ stakes), externally-
managed pension funds, EPFM, i.e., the 4.1. Characteristics of occupation
stakes held by the largest 20 external pension funds holdings
pension fund managers (either on behalf of
occupational pension funds and other Table 9.2, Panel A, reports the distribution
clients), and NIBlock, the holdings of of companies with pension funds holdings. Out
individuals and families, non-financial of 289 test companies, 175 (61%) reported
companies, public bodies, foreign investors an overall holding of between 3% and 6%,
and nominees.21 and 70 (24%) had between 6 and 10%.
Table 9.1 Description of proxy variables. Expected signs (Esign) are for the Logit regressions
Variables Definition Proxying for (Hypotheses) Esign
LPF Proportion of outstanding equity owned by largest occupational pension funds Pension fund commitment to monitoring
TPF The proportion of outstanding equity owned by all occupational pension funds Pension fund commitment to monitoring
IPF Occupational pension funds’ holdings in test companies relative to their assets Pension fund commitment to monitoring
NPF Number of occupational pension funds holding large stakes in our test companies Pension fund commitment to monitoring
LNPFA Log of the asset value of the pension fund with the largest stake Pension fund commitment to monitoring
Dir Proportion of outstanding equity owned by directors Management entrenchment
Block Proportion of outstanding equity owned by other large shareholders Blockholders’ incentive to monitor
EPFM Proportion of outstanding equity owned by the top 20 external fund managers External managers’ incentive to monitor
IBlock Proportion of outstanding equity owned by financial institutions, Incentive to monitor by other financial
other than internally and externally managed pension funds companies
NIBlock Proportion of outstanding equity owned by non-financial shareholders Blockholders’ incentive to monitor
Q Market value of equity plus book value of debt over total assets Firm value
M/B Market value of equity over shareholders’ funds Firm value
M/S Market value of equity plus book value of debt over sales Firm value
P/E Year-end share price over earnings per share Growth opportunities
ROA Profit before interest and tax over total assets Firm performance
ROE Earnings over shareholders’ funds Firm performance
ROS Profit before interest and tax over sales Firm performance
Ri, t 12 to t 1-year share price return, adjusted for stock issues and dividends Firm performance
ME Market value of equity at balance sheet date Firm size
TA Total assets Firm size
Mlev Long-term debt over long-term debt plus market value of equity Monitoring role of debtholders
Blev Long-term debt over long-term debt plus shareholders’ funds Monitoring role of debtholders
Dir rem./Sal Directors’ monetary pay (excl. options) over sales Monitoring directors’ pay
Nechair Dummy 1 if the role of Chairman is attributed to a non-executive director Monitoring board composition
Split Dummy 1 if the role of Chairman is split from that of CEO Monitoring board composition
#DIR Number of directors (both executives and non-executives) Monitoring board size
%NED Proportion of non-executive directors Monitoring board composition
Payout Ordinary dividends over earnings Monitoring of dividends
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 231
232 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 9.2 Distribution of pension funds holdings in test companies
The sample includes 289 test companies where occupational pension funds hold 3% or more
of ordinary shares and 356 individual pension fund block shareholdings. LPF is the proportion
of outstanding equity owned by largest occupational pension funds; TPF is the proportion of
outstanding equity owned by all identified pension funds; IPF is the ratio of occupational pension
funds holdings over their total assets; NPF is the number of occupational pension funds; EPFM is
the proportion of shares owned by the largest 20 external pension fund managers, either on behalf
of occupational pension funds or other clients; NC is number of companies; and NStakes is for number
of stakes; NPension funds is the total number of pension funds.
Thus, more than 84% of our test companies interest; 40 companies (14%) have two
have pension funds holdings between 3 and relevant pension funds, 8 (3%) have three
10%. In 36 companies (12%) pension pension funds, 1 company had four pension
funds hold between 10% and 20% and in funds and 2 reported five relevant pension
8 companies (3%) the holdings exceed funds holdings, a total of 356 stakes.
20%. Although the pension fund deeds Table 9.2, Panel C, reports the distribu-
often set up upper limits of the proportion tion of the 99 pension funds that hold these
of shares that a fund is allowed to hold in 356 stakes. The vast majority of these
an individual company, the law in force funds hold a small number of stakes: 68
does not establish any limitations. The only pension funds hold just one large stake, and
limit is that a pension fund cannot invest 18 pension funds hold less than five large
more than 10% of its assets value in the stakes. These 86 pension funds hold a total
sponsoring company’s shares. of 115 large stakes. Although not reported
Table 9.2, Panel B, shows that 238 com- in the table, we find that the magnitude of
panies (82%) have only one pension fund the holdings is positively correlated with
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 233
the size of individual occupational pension total, rather than the individual, incidence
funds. The nine occupational pension funds is likely to be high. In particular, the largest
that hold at least 10 large stakes, hold a fund, British Telecommunications/Post
total of 211 stakes (59% of our sample of Office pension fund, holds 53 large stakes,
stakes) and, with the exception of Mars implying an average incidence of 7.95%
Pension, the market value of their managed (53 0.15%). Thus, while, the aggregate
assets exceeds £4 billion. British blocks may represent, individually, a small
Telecommunications-Post Office Pensions, proportion of pension fund assets, for each
the largest occupational pension fund in pension fund the total stake may be sub-
the UK, holds 53 stakes. stantial, suggesting that their monitoring
Table 9.2, Panel D, reports the descrip- role should be beneficial.
tive statistics of occupational pension Overall, the results reported in Table 9.2
funds holdings and their incidence. On are striking and suggest that, despite their
average, occupational pension funds hold aggregate holdings, individually, internally-
individually 6% and collectively 6.96% of managed occupational pension funds hold
our test firms equities. Although these large stakes in a small number of compa-
holdings range between 3 and 57%, the nies, most of these stakes are between 3
vast majority of our companies reported and 10%, the vast majority of our test
holdings of between 3 and 6% (Panel A). firms report only one occupational pension
These holdings are relatively large comparing fund holding and these holdings represent a
to those used in previous studies. For relatively small proportion of pension
example, the average stake of institutional funds’ assets.
investors in target firms is 0.3% in Wahal Table 9.2, Panel D, shows that external
(1996) and between 0.4% and 2.3% in pension fund managers, EPFM, as a whole,
Del Guercio and Hawkins (1999). hold, on average, 8.99% (median 4.79%)
To assess the impact of these stakes on of our sample firms’ equity, but these hold-
each individual pension fund assets, we ings are not statistically different from
compute Incidence on Pension Funds those of internally managed occupational
(IPF), the ratio of these investments over pension funds (t-statistic of the difference
total assets of each pension fund. As stated in means 1.07). However, Financial
above, the data on the portfolio of the Times (1997a) reports that only about
assets managed is not available for all pen- 25% of these funds’ assets are managed on
sion funds. The denominator of this ratio behalf of occupational pension funds. This
which includes other investments in fixed implies that the large stakes held by these
income securities, property, overseas equity external managers on behalf of occupa-
and cash, is likely to be larger than pension tional pension funds represent on average
fund equity investment. Thus, this ratio will roughly 2.25% (8.99% * 0.25) of firms’
understate the actual incidence level.22 equity, assuming that these funds invest in
Keeping this drawback in mind, Table 9.2, the same way pension funds assets and
Panel D shows that, on average, the invest- other assets they manage. This suggests
ment of pension funds in our test compa- that the impact of these investors is likely
nies represents 0.15% of their total assets, to be marginal. Unfortunately, we do not
ranging between 0.01 to 14.75%. These have data on the way these stakes are allo-
proportions are larger than those reported cated among the different clients, but we
by Del Guercio and Hawkins (1999) who have included them in our analysis to
show that pension funds holdings represent assess the extent to which externally man-
between 0.17 and 0.34% of their invested aged pension funds monitor companies.
portfolio. However, given that, in our sam- In Table 9.3, Panel A and B, we report
ple, 211 blocks out of 356 are held by only changes in the holdings of occupational
nine funds, the combination of the pension pension funds in our test firms over the
fund incidence statistics with this high con- 1992–1996 period and the t-statistics of
centration of holdings implies that the differences in means and medians. Data on
234 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 9.3 Annual changes in occupational pension fund holdings and list of occupational pension
funds investing in their own companies
pension fund stock ownership in 1992 is confirm that pension fund holding did not
taken from the Hambros Company Guide change over the 1992–1996 period. The
and from the London Stock Exchange results are consistent with WM (1996)
Yearbook,23 since Extel Financial provides findings and suggest that pension funds are
data only for the latest year. The table long-term investors and that our analysis is
shows that 241 companies (83% of our not sample dependent.
test companies) reported large pension We check whether pension funds invest in
funds holdings in 1992 and 1996. On their own companies where they may have
average, occupational pension funds held different objectives and monitoring roles
individually 6.21% of shares in 1992, than when they invest in other firms.
5.93% in 1995 and 6.09% in 1996. The Table 9.3, Panel C, lists the 11 companies
difference in means and in median across (3.8% of our test firms) with such stakes.
these years is not statistically significant. The number of these companies and the
We repeat the exercise using total pension magnitude of their holdings are not substantial
fund holdings and number of pension funds and the inclusion of these companies in our
and the results (not reported in Table 9.3) test sample did not alter our results.
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 235
Panel A: Size and other variables relative to all companies in the UK (NTest 586;
NControl 2702)
Notes:*** Significant at 0.01 levels.** Significant at 0.05 levels.* Significant at 0.10 levels.
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 237
shows that, our test firms do not exhibit Column (6), Table 9.5, controls for
lower directors’ remuneration than the differences in other monitoring mecha-
control firms.These results are striking and nisms to account for the fact that these
suggest that pension funds large holdings various mechanisms may be substitutes.
do not increase the likelihood of compliance The results show that the main difference
with the Cadbury (1992) and Greenbury between our test and control firms is the
(1995) recommendations. measure of firm value, Q, suggesting that
Table 9.4, Panel B, shows that our test our test firms have lower value than the
firms pay an average of 32% of their control firms.25 The coefficient of block-
earnings as dividends. This is not statisti- holding is also negative and significant at
cally different from the average payout the 10% confidence level, suggesting that,
ratio of 33% of our control firms. The after controlling for all other differences
results are not consistent with the short- between our test and control firms, the test
termism arguments and indicate that companies have lower blockholding than
occupational pension funds do not our control firms. No single pension fund-
necessarily demand high payouts from monitoring variable is significant at any
companies in which they hold large stakes. confidence level. These results cast doubt
Following Jensen (1986) arguments, the on the monitoring role of pension funds in
results also imply that pension funds are the UK.26
not monitoring these companies as they do In Table 9.6 we analyze the causality in
not make them disgorge the free cash the relationship between board structure
flow. The results are, nevertheless, and occupational pension funds holdings.
consistent with previous studies who do We run a set of regressions where the
not find dividend tax clientele in the UK various dependent variables which proxy
(e.g., Lasfer, 1996). for board structure are measured in the
1996–1997 financial year and the explana-
4.3. Do pension funds affect firm tory variables such as pension funds holdings
value and board structure? are in 1995–1996 financial year. We
hypothesize that pension funds require
Table 9.5 reports the results of the Logit companies in which they hold large stakes
regressions where the various agency in 1995–1996 to adopt the Cadbury’s
variables are considered simultaneously. (1992) board structure in 1996–1997,
The dependent variable is equal to 1 if i.e., to split the roles of chairman and chief
company i is in the test sample and 0 if it executive officer, to appoint a non-executive
is part of the control sample. The table director as a chairman and to have a large
shows that, with the exception of the vari- proportion of non-executive directors on
ables that proxy for firm value that are the board. In addition, we follow
significantly lower for our test firms, there McConnell and Servaes (1995) and focus
is no statistical difference between our test separately on low and high growth firms
and control firms. As shown in Table 9.4, using E/P ratio as a proxy for growth
companies in which occupational pension opportunities. Firms with E/P ratio above
funds hold large stakes are not more (equal or below) the median are classified
profitable and do not pay higher dividends as low (high) growth. We expect a positive
than the control firms. In addition, our test relationship between pension funds hold-
firms are not more likely to split the roles ings and the adoption of the Cadbury
of chairman and chief executive officer, (1992) recommendations on the composi-
have less directors or more non-executive tion of the board of low growth companies,
directors than our control group.These two which are more likely to suffer from the
last issues were the main focus of the free cash flow problem.
recommendations of Cadbury (1992), The first three columns of Table 9.6
which relied on pension funds for their report the results of the Logit regression
implementation. where the dependent variable is equal to 1
Table 9.5 Logit regressions of the probability that pension fund holdings exceeds 3% of shares 95–96
The dependent variable is equal to 1 for companies that reported pension fund holding above 3% in 1995–1996 and to 0 for industry and size matched
control firms. M/S is market value of equity plus book value of debt over sales; Q is the ratio of market value of equity plus book value of debt over total
assets; Blev is the ratio of long-term debt over long-term debt plus shareholders funds; Ri,t12 to t is a 1-year share price return; Dir rem./Sal is the ratio
of directors remuneration over sales; Split is a dummy variable equals to 1 if the roles of chairman and CEO are split; #DIR is the number of directors on
the board; %NED is the proportion of non-executive directors in the board; Payout is the ratio of dividends over earnings; Dir is the proportion of
outstanding equity held by directors; Block is the proportion of outstanding equity capital held by blockholders other than directors and occupational
pension funds; 2, the chi-squared, is used to test that all slopes of the Logit regression are zero by comparing the restricted and the unrestricted log
likelihoods. t-values are in parentheses.
Constant 0.71*** (6.64) 0.61*** (6.69) 0.52*** (31.91) 0.52*** (32.16) 0.47* (1.84) 0.79*** (6.93)
Split 0.014 (0.27) 0.017 (0.31) 0.005 (0.09)
#DIR 0.011 (1.18) 0.016 (1.58) 0.008 (0.70)
%NED 0.04 (0.37) 0.042 (0.36) 0.02 (0.19)
Dir 0.003 (0.28) 0.005 (0.53) 0.00 (0.00)
238 EQUITY OWNERSHIP STRUCTURE AND CONTROL
if the roles of CEO and Chairman are split externally-managed pension fund monitoring
and zero otherwise. The results show that of the board structure. At the same time,
the coefficients of the pension fund vari- board structure is unrelated to market
ables are positive but not significant except performance, suggesting that the recent
for pension fund size (LNPFA). When we trend towards the adoption of the
split our sample into high and low growth Cadbury’s prescriptions was not related to
companies none of the pension funds vari- the presence of agency conflicts, but rather
ables is significant. We tested for possible dictated by some “need of visibility” by
multicollinearity problem by running the companies. Franks et al. (1998) also
regressions with each single variable. We report a similar relationship. We do, how-
find (but do not report) that the holdings of ever, report some evidence consistent with
directors, Dir, and the coefficient of non- monitoring role of other than pension funds
institutional block shareholders, NIBlock, institutional shareholders and of debtholders,
are negative and significant. None of the at least with regards to the appointment of
occupational pension funds variables is sig- non-executive directors within low growth
nificant. We find that the coefficient of the firms. There is also some evidence that
external pension fund managers variable, managerial ownership is used to entrench
EPFM, is positive and significant for the the position of incumbent managers as the
whole sample, though it is not when the coefficient of directors’ ownership is, in
sub-samples of high and low growth firms most cases, negative and significant.
are separated. Finally, our results are consistent with the
Columns 4 to 6 provide the results of the hypothesis that institutional blockholders
Logit regressions where the dependent vari- (other than pension funds) lead high
able is equal to 1 if the company has growth firms to appoint a non-executive
appointed a non-executive director as chairman. We simulated our results using
chairman and zero otherwise. Here again growth in profits, market returns and Q as
none of the coefficients of the pension fund proxies for growth opportunities and/or
variables are significant. However, when we presence of agency conflicts. The results
run the regressions with a single independent are qualitatively similar to those reported
variable, we find that the coefficient of in Table 9.6.
pension funds incidence, IPF, is positive and
significant (1.36 with t 1.71) for the 4.4. Pension funds investments and
whole sample. For the low-Q companies, we firm value
find that the coefficient of total pension
funds,TPF, is positive and significant (0.06 In this section we focus only on our test
and t 1.93).The coefficients of directors companies and test for the relationship
ownership, Dir, and non-institutional between firm value and ownership structure.
blockholders, NIBlock, are negative and Table 9.7 reports the Pearson correlation
significant while those of institutional matrix.The correlation between the various
blockholders and external fund managers measures of occupational pension fund
are positive and significant. holdings and firm value is, in most cases,
The last three columns present the weak and negative. An exception is
results of the OLS regressions where the represented by the correlation between the
dependent variable is the proportion of market-to-sales ratio and the pension fund
non-executive directors in the board. The incidence variable. Similarly the correlation
results show that none of the pension fund between firm value, blockholding and direc-
variables is significant. When the regres- tors holdings is not significant. However,
sions are run separately, we find that only ownership variables are negatively corre-
size, Ln(TA), and leverage, Blev, are posi- lated with size as measured by total assets,
tive and significant. suggesting that occupational pension
The overall results do not provide funds, directors, institutional blockholders
strong support for the occupational and (other than pension fund managers) and
Table 9.6 Determinants of board structure for low and high growth firms
In each regression the dependent variable is computed for the period 1996–1997, while all independent variables are 1-year lagged (i.e., 1995–1996).The
regressions are run separately for all 250 companies, All, low-growth companies (119), and high growth companies (131). Growth opportunities are
measured using the median E/P ratio. The first six columns are Logit regressions where Split or Nechair are equal to 1 if the company has split the roles
of its chairman and chief executive officer or has a non-executive director as a chairman, and zero otherwise. In these regressions the Pseudo-R2 measures
the goodness of fit. In the last three columns we run OLS regression with the proportion of non-executive on the board %NED as the dependent variable.
TPF is the proportion of outstanding equity owned by all identified pension funds; IPF is the ratio of occupational pension funds holdings over their total
240 EQUITY OWNERSHIP STRUCTURE AND CONTROL
assets; LNPFA is the log of the asset value of the pension fund; Dir is the proportion of outstanding equity owned by the directors; EPFM is the proportion
of equity held by external pension fund managers (either on behalf of pension funds or other clients); IBlock represents the proportion of equity overall held by
institutional block shareholders other than (both internally and externally managed) pension funds; NIBlock is the proportion of equity held by non-institutional
blockholders; Ri,t – 12 to t is a 1-year stock return.
Constant 2.974 0.611 6.447 0.091 0.165 0.121 0.163 0.213 0.048
(0.68) (1.11) (1.11) (0.02) (0.03) (0.02) (0.60) (0.50) (0.13)
Dir 0.004 0.019 0.002 0.024** 0.020 0.019 0.0002 0.0001 0.0004
(0.44) (0.88) (0.19) (2.17) (1.42) (1.28) (0.24) (0.09) (0.46)
IBlock 0.030 0.010 0.028 0.039** 0.015 0.067** 0.0002 0.001 0.002
(1.13) (0.19) (0.77) (2.34) (0.74) (2.06) (0.21) (0.74) (1.25)
NIBlock 0.011 0.068* 0.000 0.017 0.025 0.003 0.001 0.001 0.00003
(0.96) (1.79) (0.03) (1.46) (1.44) (0.18) (0.73) (1.00) (0.03)
TPF 0.02 0.10 0.03 0.05 0.03 0.05 0.002 0.001 0.003
(0.38) (0.86) (0.51) (1.38) (0.71) (1.03) (0.79) (0.36) (1.9)
LNPFA 0.53** 0.06 0.26 0.01 0.13 0.05 0.01 (0.01) 0.005
(2.15) (0.12) (0.94) (0.01) (0.50) (0.17) (0.64) (0.01) (0.27)
IPF 1.48 1.46 0.75 0.74 1.88 1.05 0.0005 0.001 0.10
(1.26) (0.46) (0.45) (0.94) (1.13) (0.77) (0.04) (0.10) (1.55)
EPFM 0.016 0.259 0.002 0.004 0.006 0.001 0.0002 0.0001 0.001
(0.43) (1.10) (0.05) (0.25) (0.28) (0.05) (0.23) (0.04) (0.35)
Blev 6.25** 0.63 1.54 1.56 0.11 1.18 0.10* 0.08 0.17
(2.17) (0.32) (0.63) (1.61) (0.10) (0.78) (1.70) (0.99) (1.61)
Ri, t12 to t 0.03 0.83 0.11 0.21 0.02 0.34 0.01 0.02 0.005
(0.08) (1.23) (0.07) (0.82) (0.06) (0.94) (0.76) (0.83) (0.18)
LN(TA) 0.47 0.09 0.07 0.02 0.14 0.004 0.02** 0.02 0.01
(1.56) (0.24) (0.18) (0.12) (0.68) (0.01) (2.19) (1.40) (0.76)
Pseudo-R2 7.29% 15.92% 4.87% 15.27% 11.98% 19.82% 0.95% 0.00% 3.17%
or R2
***
Significant at the 0.01 level.
**
Significant at the 0.05 level.
*
Significant at the 0.10 level.
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 241
Table 9.7 Pearson correlation coefficients between the variables used
The sample includes all companies that reported occupational pension fund holding above 3%. The dependent variables Q and M/S are measured
in 1996–1997 financial year. LPF is the proportion of outstanding equity owned by largest occupational pension funds. All the variables are measured in
1995–1996 financial year.TPF is the proportion of outstanding equity owned by all identified pension funds; IPF is the ratio of occupational pension funds
holdings over their total assets; NPF is the number of occupational pension funds; Dir is the proportion of outstanding equity owned by the directors; EPFM
is the proportion of equity held by external pension fund managers (either on behalf of pension funds or other clients); IBlock represents the proportion
of equity overall held by institutional block shareholders other than (both internally and externally managed) pension funds; NIBlock is the proportion of
equity held by non-institutional blockholders; In(TA) is the log of total assets; Blev is the ratio of long-term debt over long-term debt plus shareholders
funds; #DIR is the number of directors on the board; Split is a dummy variable equals to 1 if the roles of chairman and CEO are split; %NED is the
proportion of non-executive directors in the board; P/E is the price–earnings ratio at the balance sheet date.
Q M/S TPF IPF NPF LNPFA EPFM Dir Dir2 IBlock NIBlock ln(TA) Blev No.Dir Split %NED
M/S 0.59
TPF 0.18 0.07
IPF 0.02 0.31 0.03
242 EQUITY OWNERSHIP STRUCTURE AND CONTROL
non-institutional blockholders hold large funds holdings (3). Although the coefficient
stakes mainly in small firms. However, the of pension fund incidence, IPF, and the size
fund manager variable is positively related of pension fund, LNPFA, are positively
to size. Consistent with previous evidence related to firm value, they are not
(e.g., Lasfer, 1995), leverage is positively significant. The coefficient of external
related to firm size but negatively pension fund managers’ stakes is negatively,
correlated with firm value. The number of though not significantly, related to firm value.
directors is positively correlated with In contrast to previous studies (e.g.,
occupational pension fund, externally- Yermack, 1996, and Eisenberg et al.,
managed pension funds, non-institutional 1998) we report a positive relationship
blockholding, leverage and size, while it is between firm value and the number of
negatively correlated with institutional directors. The difference in the results
blockholders and directors’ ownership. could be due to the fact that our companies
The split dummy variable is positively are relatively middle-sized compared to the
related to leverage, block ownership, size sample of small companies of Eisenberg
and proportion of non-executive directors et al. (1998) and that of large companies
in the board, but negatively related to of Yermack (1996). Finally, the coefficients
directors’ holdings. This suggests that the of the holdings of directors variable and its
larger the company and the higher squared value are not significant suggesting
the debt–equity ratio, blockholding and the that there is no linear or non-linear
proportion of non-executive directors in the relationship between value and managerial
board, the higher its propensity to split ownership. These results are not consistent
the roles of chairman and CEO. However, with the findings of McConnell and Servaes
none of the occupational pension fund (1990; 1995).
measure is statistically related to the split We simulate these results using market
dummy, implying that pension funds, value of the firm over sales as a proxy for
individually or collectively, do not push firm value (Lins and Servaes, 1999). The
companies to split the roles of chairman results, reported in (4) and (5), show a
and chief executive officer. Finally, directors’ positive relationship between value and
holdings are negatively related to the pension fund incidence, but a negative
proportion of non-executive directors in the relationship with the number of pension
board, suggesting that such holdings funds.The coefficient of externally-managed
exacerbate the potential agency conflicts pension funds is negative but not significant
between the board and the management. while that of institutional block ownership
In Table 9.8, columns (1) to (3), we is positive and significant. We also simulate
report the results of the regressions of firm our results by using sales as a measure of
value, as measured by Tobin’s Q in size and capital expenditure over total
1996/97, against various measures of assets as a proxy for growth opportunities.
occupational pension funds holdings in the The results, not reported for space reasons,
1995/96 financial year. These results show are qualitatively similar to these reported
that, with the exception of total pension in Table 9.8. Overall, our results suggest
funds stakes, TPF, none of the various that pension funds do not add value to
measures of pension funds holdings explain companies in which they hold large stakes.
firm value. The total pension funds stakes
variable, TPF, is actually, negative and 4.5. Effects of pension fund holdings
significant, suggesting that pension funds on firms’ long-term performance
collectively destroy value. Even after
controlling for other monitoring mechanisms We analyze the long-term performance of
documented in the previous literature (e.g., our test firms by comparing the changes in
Agrawal and Knoeber, 1996; Yermack, accounting and stock price performance
1996), such as size and P/E, firm value is over the sample periods 1994–1995 and
still negatively related to total pension 1996–1997. As in Karpoff et al. (1996)
Table 9.8 Relationship between firm value and pension fund holdings
The sample includes all companies that reported occupational pension fund holding above 3%. In columns (1) to (3) we use Q, the ratio of market value
of equity plus book value of debt over total assets, as the dependent variable. In column (4) and (5) the results are simulated using the market value of
equity to sales as dependent variable. The dependent variables are measured in 1996–1997 while the independent variables are measured in 1995–1996.
TPF is the proportion of outstanding equity owned by all identified pension funds; IPF is the ratio of occupational pension funds holdings over their total
assets; LNPFA is the log of the market value of the largest pension fund’s asset; NPF is the number of occupational pension funds; EPFM is the propor-
tion of equity held by external pension fund managers (either on behalf of pension funds or other clients); Dir is the proportion of outstanding equity owned
by the directors; Dir2 is its squared value; IBlock represents the proportion of equity overall held by institutional block shareholders other than (both inter-
nally and externally managed) pension funds; NIBlock is the proportion of equity held by non-institutional blockholders; #DIR is the number of directors;
Split is a dummy variable equals to 1 if the roles of chairman and CEO are split; %NED is the proportion of non-executive directors; P/E is the price–earnings
ratio at the balance sheet date. t-values are in parentheses.
Q M/S
(1) (2) (3) (4) (5)
Constant 0.97 (0.64) 0.62 (0.39) 0.57 (0.36) 0.07 (0.04) 0.97*** (2.80)
TPF 0.033*** (2.60) 0.042*** (3.07) 0.049** (2.57) 0.001 (0.06)
IPF 0.034 (0.46) 0.083 (1.34) 0.086 (1.38) 0.36*** (4.50) 0.372*** (5.41)
LNPFA 0.110 (1.62) 0.092 (1.46) 0.095 (1.50) 0.007 (0.09)
NPF 0.09 (0.56)
244 EQUITY OWNERSHIP STRUCTURE AND CONTROL
and Del Guercio and Hawkins (1999), we company selected by identifying all firms
investigate whether companies in which with a market value of equity between
pension funds hold large stakes rebound 70% and 130% of that of the test
more quickly from poor performance or company. Then, from this set of firms, we
maintain their good performance over a select the company with the market-to-
longer time period. We use return on book ratio closest to that of the test firm.
assets (ROA) as a proxy for accounting We use the return of this company, rc*,n, as
performance. We follow Wahal (1996) and a benchmark.
compute the 2-year industry-adjusted This measure of performance accounts
Cumulative Abnormal Return On Assets for the fact that occupational pension
(CAROA) as funds are long-term investors, as reported
in Table 9.3, and does not suffer from
(1 ROA )
T
cumulating biases observed in arithmetic
CAROAi,t,T i,n
nt mean returns (Conrad and Kaul, 1993;
(1 ROA
T
Wahal, 1996; Barber and Lyon, 1997). We
s,n) eliminate survivorship bias by including
nt
dead companies in our sample. Out of our
where ROAi,t is company i return on assets 289 test companies, 34 firms (11.76%)
in year t and computed as are excluded because they went public after
1995 and we could not compute the mar-
Profit Before Interest and Tax . ket performance for the first period.
ROAi,t The results reported in Table 9.9 show
Book Value of Total Assests
that the industry-adjusted return on assets
ROAs,t is the industry s median return on (CAROA) has not changed significantly
assets in year t. over the two sub-periods. Companies in
We simulate our results by computing in which occupational pension funds hold
the same way Cumulative Abnormal large stakes underperform the industry
Return On Equity (CAROE) and average (though not significantly) by
Cumulative Abnormal Return On Sales 3.38% in 1994–1995 and by 1.50% in
(CAROS). All measures of performance are 1996–1997. As in Wahal (1996), we split
adjusted by subtracting the industry the sample into overperformers and
median. underperformers. Companies with CAROA
The abnormal performance is also evalu- below (above) the median are classified as
ated using stock market returns.We compute underperformers (overperformers).27 The
the Share Price Return (SPR) from a abnormal performance of the overperforming
buy-and-hold strategy over the sub-periods companies decreased from 11.62% to
1994–1995 and 1996–1997 as follows: 6.44% and that of the underperformers
has increased from 18.5% to 9.51%.
{1 r }
T
The results are consistent with Wahal
SPR i,t,T i,n
nt (1996) and indicate that both the
{1 r
T
under-and overperformers experience a
c*,n} convergence to industry means. The results
nt
are not qualitatively affected by the use of
where other measures of performance such as
industry-adjusted return on equity (Panel B)
Pi,n or industry-adjusted return on sales
ri,n 1.
Pi,n1 (Panel C).
Panel D and E report the changes in firm
We adopt the Barber and Lyon (1997) value as measured by (the industry-
methodology, and control for size (market adjusted) Tobin’s Q and changes in abnormal
capitalization) and market-to-book. We returns. For the full sample, the industry-
match each test company with a control adjusted Q has decreased from 0.163 to
246 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Table 9.9 Pension fund holdings and long term accounting and stock price performance
The sample includes all companies that displayed (at least) one relevant pension fund holding.
Industry-adjusted return on assets (CAROA) is the return on assets of each company in the sample
less the median return on asset of the industry. Industry-adjusted return on equity and return on
sales and Q are computed in a similar way. The control sample includes firms taken from the same
SEC industry whose 1994–1995 performance is the closest to our test firm. Panel E reports the
results based on the size and market-to-book adjusted share price returns using the Barber
and Lyons (1997) methodology. The sample is dividend into a sample of underperformers and
overperformers based on the median performance of the firm in 1994–1995. For example, if the
industry-adjusted return on assets for a firm is lower (higher) than the median, the firm is classified
as an underperformer (overperformer).
t-statistic Mann–
1994–1995 1996–1997 difference Whitney
N Mean Median Mean Median in means p-value
0.072.28 The decrease in both the average to 0.48. The decrease in the mean median
and median firm value is statistically are statistically significant. In contrast, the
significant. The results suggest that, over average and the median Tobin’s Q of the
time, the value of companies in which occu- underperforming companies did not change
pational pension funds hold large stakes significantly. These companies have under-
decreases. As in the above panels, we split performed in 1994–1995 period and car-
the sample into overperformers and ried on underperforming in the 1996–1997
underperformers. Overperforming (under- period. These results are striking as they
performing) companies are companies with imply that companies in which occupa-
Tobin’s Q higher (lower) than the median. tional pension funds hold large stakes
The overperforming companies have done decrease in value through time, and those
worst over the two sample periods. Their that are already underperforming do not
average value has decreased from 0.67 improve. In sum, it appears that pension
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 247
funds are passive investors: they do not companies in which they hold large stakes.
make companies in which they hold large Our results cast doubt on the monitoring
stakes improve their performance and they role of pension funds which are considered
do not sell their holdings in companies that theoretically, on the one hand, to be the
are underperforming. main promoters of corporate governance in
Panel E reports the results based on the the UK, and, on the other hand, to be short-
size and market-to-book adjusted share termist and dictate their rules to companies.
price returns using the Barber and Lyon At the same time, we show that, despite
(1997) methodology. For the full sample, the relatively poor performance of the
companies have decreased their average companies in which they invest, occupa-
performance from 0.49% to 3.2% and tional pension funds do not opt for the ‘exit’
the median performance from 3.85% to strategy. We show that there is no apparent
8.76%. However, this decrease is not specific relationship between the funds and
statistically significant. The stock price the companies in which they invest and
performance of the sub-sample of the over- these companies are not illiquid to make it
performing firms declined significantly impossible to exit. One possibility is that
from 68.22% to 6.83% (t-statistic of pension funds choose to invest in low Q
the difference in means 6.62), while firms to benefit from return reversals on
underperforming companies increased their these securities. However, the lack of data
performance from 67.76% to 0.45% on the investment styles of these funds does
(t-statistic of the difference in not allow us to explore this issue further. At
means 7.01). As in Wahal (1996), our this stage, our results suggest that, once
results are consistent with mean reversion ‘locked in’ pension funds find it difficult and
hypothesis and cannot be attributed to costly to monitor or to sell their holdings for
pension funds investments.29 fear of selling at a discount or to convey
information to the market.
Our results may not come as a surprise.
5. DISCUSSION AND CONCLUSIONS There is a large debate in the UK about the
lack of monitoring by pension funds and it
In this paper we analyze the performance is only recently that the National
of companies in which occupational Association of Pension Funds considered
pension funds hold large stakes and test seriously this issue (NAPF, 1996b). In
the hypotheses that, because of their addition, occupational pension funds may
size, structure and objectives, these funds not have a material effect on the perform-
should be effective monitors of UK ance of companies in which they hold large
companies. stakes because these blocks tend to repre-
We compare the financial performance sent relatively small fractions of the total
of companies in which occupation pension values of the funds’ assets. The mean
funds hold more than 3% of the issued (median) block represents only 0.15%
share capital against all other non-financial (0.03%) of the typical funds assets, and
and industry and size-adjusted control with such a small fraction, it is not clear
groups. We show that our test firms are that the gains to the funds of expending
small and have low value. These companies effort monitoring the sample firms would
are also not likely to be more efficient justify the costs. Thus, our results are
and/or to pay higher dividends than the consistent with Admati et al. (1994)
control group. However, our results show proposition that, given that monitoring is
that these holdings constitute a small pro- costly, these funds will not monitor as this
portion of the occupational pension funds is not likely to result in a modification in
assets, suggesting that most of their funds the firms payoff structure and will not lead
are invested in large companies where they to net gains. They may also refrain from
do not hold large stakes. We report that intervening publicly for fear of drawing to
pension funds do not add value to the public attention the difficulties the company
248 EQUITY OWNERSHIP STRUCTURE AND CONTROL
is facing and/or trading on insider information. 3 See Black (1998) and Karpoff (1998) for a
However, our results could also imply that survey of the shareholder activism literature.
pension funds are passive investors, investing 4 Hutton (1995) argues that “pension
most of their funds in the index and these funds . . . have become classic absentee land-
lords, exerting power without responsibility and
investments we analyzed in this paper are
making exacting demands upon companies with-
peripheral. out recognizing their reciprocal obligations as
While our results are not consistent with owners” (p. 304).
the monitoring role of occupational pension 5 For example, Drucker (1976) stipulates that
funds, our study does not fill all the gaps in “pension funds are not ‘owners’, they are
the literature. Our sample includes only investors. They do not want control . . . If they
companies in which occupational pension do not like a company or its management, their
funds hold more than 3% of the issued duty is to sell the stock” (p. 82). More recently,
share capital. A more desegregated and Porter (1997) argues that institutional
comprehensive data is not available. investors, despite their substantial aggregate
holdings, do not sit on corporate boards and
Similarly, because of data unavailability, we
have virtually no real influence on managements
have not addressed the question of whether behavior because they invest nearly all their
occupational pension funds sponsor initia- assets in index funds rather than directly in
tives, whether they meet with companies companies. Short and Keasey (1997) suggest
and whether there is a co-operation that once pension funds are locked in, it is costly
between various funds. The extent to which to get involved in monitoring and they cannot
these additional factors will strengthen or exit in case they are considered to trade on
alter our analysis is a matter of further insider information. Murphy and Van Nuys
investigation. (1994) argue that pension funds are run by indi-
viduals who do not have the proper incentives to
maximize fund value.
6 Cadbury (1992) notes that “Because of
ACKNOWLEDGEMENTS their collective stake, we look to the institutions
in particular, with the backing on Institutional
We are grateful to Yakov Amihud, Shareholders’ Committee, to use their influence
Francesco Cesarini, John McConnell, as owners to ensure that the companies in which
Annette Poulsen (the editor), Jonathan they have invested comply with the Code” (para.
6.16).The National Associate of Pension Funds
Sokobin, Geof Stapledon, Sudi
also endorses such recommendations (NAPF,
Sudarsanam, an anonymous referee, and 1996b).
seminar participants at Brunel University 7 A number of studies document the free cash
and the 1998 European Financial flow problem in the UK. For example, Franks
Management Association meeting in and Mayer (1994) find that UK companies pay
Lisbon for valuable insights and helpful high dividends relative to German companies;
comments. Faccio acknowledges a research Lasfer (1995) shows that debt mitigates the
grant from City University Business free cash flow problem; Lasfer (1997) provides
School. All errors are our responsibility. evidence that firms with free cash flow problems
pay scrip, rather than cash, dividends. The
reports of Cadbury (1992) and Greenbury
(1995) are a manifestation of the previous
NOTES debate on the various wider corporate gover-
nance issues detailed in Charkham (1995),
* Corresponding author. Tel.: 44-0-171- Stapledon (1996) and Keasey et al. (1997).
477-8634; fax: 44-0-171-477-8648; e-mail: 8 Companies Act 1995, Sections 198 and 199
m.a.lasfer@city.ac.uk requires UK companies to disclose in their
1 Tel.: 39-02-7234-2436; e-mail: m.faccio@ accounts the name of any investor who holds
iol.it. 3% or more of the issued share capital.
2 Other studies suggest that the relationship 9 See Chew (1997) for a collection of papers
between ownership, such as the fraction of dealing with these two corporate governance
shares held by insiders and performance is not systems.
linear but roof-shaped (e.g., Morek et al., 1988; 10 See Black (1998) and Karpoff (1998)
Stulz, 1988; McConnell and Servaes, 1990). for a survey.
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 249
11 Similarly, in the US, the value of DC plans 19 The ideal would be to exclude from the
in 1993 was $1068 billion compared to $1248 denominator of this ration other assets such as
billion for DB plans. However, DC schemes are property, cash and fixed income securities, but
growing at much faster rates of 19% per year, the desegregated data on equity investment is
compared to 14% for DB plans (Jepson, 1998). not available.
12 The report states that no “one individual 20 This variable includes directors holdings
has unfetted powers of decision. Where the chair- but excludes those of officers. UK quoted com-
man is also chief executive, it is essential that panies are required to disclose in their financial
there should be a strong and independent element statements the proportion of shares held
on the board, with a recognised senior member”. directly and indirectly by executive and non-
13 The Code of Best Practice No 4.3 executive directors (Companies Act 1985).
recommended an audit committee of at least However, no similar disclosure applies to offi-
three non-executive directors with written terms cers. This legal disclosure requirement means
of reference and No 4.30 recommends the that we had to define managerial ownership as
institution of a nomination committee as an ownership by members of the board of direc-
internal committee within the board. This tors. Although this definition is consistent with
committee should be composed of majority of that of Morck et al. (1988), it differs from that
non-executive directors and chaired by the of McConnell and Servaes (1990) and Denis
chairman of the board. and Sarin (1999) as we do not include shares
14 The 99 in-house managed pension funds owned by corporate officers not members of
included in our study manage assets for some the board.
£150 billion, which correspond to 26.8% of all 21 Denis and Kruse (1999) and Denis and
occupational pension funds’ assets. Sarin (1999) distinguish between unaffiliated
15 The report summarizes the functions of the and affiliated block shareholdings. Unaffiliated
board as follows: “The responsibilities of the blockholders would be expected to perform an
board include setting up the companys strategic important monitoring role and make firms com-
aims, providing the leadership to put them into ply with the Code of Best Practice, while affili-
effect, supervising the management of the ated blockholders are more likely to be sensitive
business and reporting to shareholders on their to managerial pressure or pursue goals other
stewardship”. As to the financial aspects of than share price maximization (Brickley et al.,
corporate governance, the report mentions: 1988, 1994). In this perspective, our internally
“The way in which boards set financial policy managed occupational pension funds are clearly
and oversee its implementation, including the “unaffiliated shareholders”. Since our data
use of financial controls, and the process does not generally allow us to check for the
whereby they report on the activities and intensity of the relationships between firms and
progress of the company to the shareholders”. shareholders, we consider together affiliated
16 See John and Senbet (1998) for an exten- and unaffiliated shareholders in the category of
sive survey of the monitoring role of corporate blockholders. However, we distinguish between
board of directors and Stapledon (1996, pp. institutional and non-institutional blockholders,
138–153), for the monitoring role of non- because these two categories of investors were
executive directors. shown to behave differently, i.e., when sponsor-
17 The choice of 1995–1996 sample period is ing initiatives (Jarrell and Poulsen, 1987;
driven by data availability.The data on sharehold- Karpoff et al., 1996).
ing is inserted manually because Extel Financial 22 One possibility of overcoming this problem
provides only the latest data on shareholding in of lack of data of pension fund assets invested in
text format and this data is not available in UK equities would be to assume that all pension
machine-readable form. Other similar studies use funds are homogeneous in their investment
also short time period (see Karpoff, 1998, for a styles, i.e., the aggregate distribution of pension
review).We report below that the vast majority of fund assets would apply to each fund. They
companies (83% of our test firms) had large would therefore hold an average of 53% of
pension funds holdings in both 1992 and 1996 their assets in UK equities and the remaining
periods and the magnitude of their holdings has 47% in fixed income securities, overseas equi-
not changed significantly.Thus, our results are not ties, property and cash, as reported in NAPF
likely to be sample-period dependent. (1996a). However, we consider that this
18 Extel Financial is a financial database assumption is not likely to hold as the average
microsystem. The database provides accounting holding is not constant through time.
as well as financial and reference data for all 23 Hambros Company Guide is a quarterly
UK companies and many international companies. publication, which gives a summary financial
250 EQUITY OWNERSHIP STRUCTURE AND CONTROL
data and shareholdings above 3% of UK Journal of Financial and Quantitative
companies. The Stock Exchange Yearbook is an Analysis 31, 377–397.
annual publication of the London Stock Agrawal, A., Mandelker, G.N., 1990. Large
Exchange. It provides a summary data of the shareholders and the monitoring of
activities, shareholdings and performance of UK managers: The case of antitakeover charter
listed companies.
amendments. Journal of Financial and
24 We exclude 58 companies with negative
book value of equity due to goodwill write-offs Quantitative Analysis 25, 143–162.
when we use the market-to-book ratio, book Ayres, I., Cramton, P., 1993. An agency per-
leverage and return on equity. The inclusion of spective on relational investing. Working
these companies did not, however, alter our Paper, Stanford University, Stanford, CA.
reported results. Barber, B.M., Lyon, J.D., 1997. Detecting long
25 The negative relationship between pension run abnormal stock returns: The empirical
fund investments and Q has been widely power and specification of test-statistics.
documented in the investment literature. For Journal of Financial Economics 43, 341–372.
example, Lakonishok et al. (1994) show that Bethel, J.E., Liebeskind, J.P., Opler, T., 1998.
pension funds invest in glamour stocks (low
Block share repurchases and corporate per-
book-to-market firms), which generally under-
perform value stocks (high book-to-market formance. Journal of Finance 53, 605–635.
firms) because the previous success of the Black, B.S., 1998. Shareholder activism and
glamour stocks helps institutions justify their corporate governance in the United States.
portfolio selection to their investors. In: Newman, P. (Ed.), The Palgrave
26 The issue of director’s pay system has just Dictionary of Economics and the Law.
been considered by the National Association of Blake, D., 1995. Pension Schemes and Pension
Pension Funds through their call to vote on Funds in the United Kingdom. Clarendon
boardroom pay (Financial Times, 1997b). Press, Oxford.
27 We have also split the sample into over- Brancato, C.K., 1997. Institutional Investors
and underperformers on the basis of positive
and Corporate Governance: Best Practice for
and negative CAROA or other variables used
below but the results did not change. Increasing Corporate Value. McGraw Hill,
28 For both 1994–1995 and 1996–1997, the Chicago.
Q ratio is computed as arithmetic average. Brickley, J.A., Lease, R.C., Smith, C.W. Jr.,
29 We obtained similar results either by using 1988. Ownership structure and voting on
the industry median return as benchmark and, antitakeover amendments. Journal of
by matching our test firms by industry and prior Financial Economics 20, 267–291.
performance. Also, for all accounting measures Brickley, J.A., Lease, R.C., Smith, C.W. Jr.,
of performance, as well as firm value, the results 1994. Corporate voting: Evidence from
were qualitatively similar when we compared charter amendment proposals. Journal of
the industry adjusted performance of 1990–92
Corporate Finance 1, 5–30.
to that of 1993–95. Similar results are obtained
when we concentrate on survived companies, Business Monitor, 1997. Central Statistical
and exclude IPO firms. Office, London.
Cadbury, A., 1992. Report of the Committee on
the Financial Aspects of Corporate
Governance. Gee, London.
REFERENCES Carleton, W.T., Nelson, J.M., Weisbach, M.S.,
1998. The influence of institutions on corpo-
ABI, 1998. Research update: How much in rate governance through private negotiations:
funded pensions in 1996? Insurance Trends Evidence from TIAA-CREF. The Journal of
16, 27–29. Finance 53, 1335–1362.
Admati, A.R., Pfleiderer, P., Zechner, J., 1994. Charkham, J., 1995. Keeping Good Company: A
Large shareholder activism, risk sharing, and Study of Corporate Governance in Five
financial market equilibrium. Journal of Countries. Oxford University Press, Oxford.
Political Economy 102, 1097–1130. Chew, D.H. (Ed.), 1997. Studies in International
Agrawal, A., Knoeber, C.R., 1996. Firm per- Corporate Finance and Governance Systems:
formance and mechanisms to control agency A Comparison of the US, Japan and Europe.
problems between managers and shareholders. Oxford University Press, New York.
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 251
Coffee, J., 1991. Liquidity versus control: The France. Journal of Applied Corporate Finance
institutional investor as corporate monitor. 9, 30–45.
Columbia Law Review 91, 1277–1368. Franks, J.R., Mayer, C., Renneboog, L., 1998.
Conrad, J., Kaul, G., 1993. Long-term market Who disciplines bad managers? Working
overreaction or biases in computed returns? Paper, London Business School, London.
Journal of Finance 48, 39–64. Gordon, L.A., Pound, J., 1993. Information,
Cotter, J.F., Shivdasani, A., Zenner, M., 1997. Do ownership structure, and shareholder voting:
independent directors enhance target share- Evidence from shareholder-sponsored corpo-
holder wealth during tender offers? Journal rate governance proposals. Journal of
of Financial Economics 43, 195–218. Finance 48, 697–718.
Davis, E.P., 1995. Pension Funds. Retirement- Greenbury, R., 1995. Directors Remuneration:
Income, Security and Capital Markets. An Report of a Study Group Chaired by Sir
International Perspective. Clarendon Press, Richard Greenbury. Gee, London.
Oxford. Hart, O., Holmstrom, B., 1987. The theory of
Del Guercio, D., Hawkins, J., 1999. The motiva- contracts. In: Bewley, T. (Ed.), Advances in
tion and impact of pension fund activism. Economic Theory. Cambridge University
Journal of Financial Economics 52, Press, Cambridge, pp. 71–155.
293–340. Hart, O., 1995. Firms, Contracts, and Financial
Demsetz, H., Lehn, K., 1985. The structure of Structure. Oxford University Press, Oxford.
corporate ownership: causes and conse- Hutton, W., 1995. The State We Are In. Vintage
quences. Journal of Political Economy 93, Press, London.
1155–1177. Jarrell, G.A., Poulsen, A.B., 1987. Shark repel-
Denis, D.J., Kruse,T.A., 1999. Managerial disci- lents and stock prices: The effects of anti-
pline and corporate restructuring following takeover amendments since 1980. Journal of
performance declines. Working Paper, Purdue Financial Economics 19, 127–168.
University, West Lafayette, IN. Jensen, M.C., 1986. Agency costs of free cash
Denis, D.J., Sarin, A., 1999. Ownership and flow, corporate finance, and takeovers.
board structures in publicly traded corpora- American Economic Review 76, 323–329.
tions. Journal of Financial Economics 52, Jensen, M.C., 1993. The modern industrial
187–223. revolution, exit, and the failure of internal
Diamond, D.W., 1984. Financial intermediation control systems. Journal of Finance 48,
and delegated monitoring. Review of 831–880.
Economic Studies 51, 393–414. Jepson, J., 1998. US Defined Benefits and
Drucker, P.F., 1976. The Unseen Revolution: Defined Contribution: Comparison and
How Pension Fund Socialism Came to Implications. Paper presented at the 1998
America. Heinemann, London. NAPF Conference, Eastbourne, UK.
Eisenberg, T., Sundgren, S., Wells, M.T., 1998. John, K., Senbet, L.W., 1998. Corporate gover-
Large board size and decreasing firm value in nance and board effectiveness. Journal of
small firms. Journal of Financial Economics Banking and Finance 22, 371–403.
48, 35–54. Kahn, C., Winton, A., 1998. Ownership struc-
Financial Times, 1997. FT Survey on Pension ture, speculation and shareholder interven-
Fund Investment. May 9, 1997. tion. Journal of Finance 53, 99–129.
Financial Times, 1997. Pension Funds to Call Kang, J.K., Shivdasani, A., 1995. Firm perform-
for Votes on Boardroom Pay: NAPF Will ance, corporate governance, and top executive
Oppose Hampel Committees Findings. turnover in Japan. Journal of Financial
September 29, 1997, p. 20. Economics 38, 29–58.
Financial Times, 1999. Fighting Fit. January Kaplan, S.N., Minton, B.A., 1994.
29, 1999, p. 17. Appointments of outsiders to Japanese
Franks, J.R., Mayer, C., 1994. The ownership Boards: Determinants and implications for
and control of German corporations. Working managers. Journal of Financial Economics
Paper, London Business School, London. 36, 225–258.
Franks, J.R., Mayer, C., 1997. Corporate owner- Karpoff, J.M., 1998. Does Shareholder
ship and control in the UK, Germany, and Activism Work? A Survey of Empirical
252 EQUITY OWNERSHIP STRUCTURE AND CONTROL
Findings. Working Paper, University of Minns, R., 1980. Pension Funds and British
Washington, Seattle, WA. Capitalism: The Ownership and Control of
Karpoff, J.M., Malatesta, P.H., Walkling, R.A., Shareholders. Heinemann, London.
1996. Corporate governance and shareholder Morck, R., Shleifer, A., Vishny, R.W., 1988.
initiatives: empirical evidence. Journal of Management ownership and market valua-
Financial Economics 42, 365–395. tion: An empirical analysis. Journal of
Keasey, K.S., Thompson, S., Wright, M., 1997. Financial Economics 20, 293–316.
Corporate Governance: Economic, Murphy, K., Van Nuys, K., 1994. State pension
Management, and Financial Issues. Oxford fund shareholder activism. Working Paper,
University Press, Oxford. Harvard Business School, Boston, MA.
Lakonishok, J., Shleifer, A., Vishny, R.W., 1994. NAPF, 1996. Year Book. National Association
Contrarian investment, extrapolation, and of Pension Funds, London.
risk. Journal of Finance 49, 1541–1578. NAPF, 1996. Good Corporate Governance.
Lang, L.H.P., Ofek, E., Stulz, R.M., 1996. National Association of Pension Funds,
Leverage, investment, and firm growth. London.
Journal of Financial Economics 40, 3–29. Nesbitt, S.L., 1994. Long-term rewards from
Lasfer, M.A., 1995. Agency costs, taxes and shareholder activism: a study of the
debt: The UK evidence. European Financial ‘CalPERS effect’. Journal of Applied
Management 1, 265–285. Corporate Finance 6, 75–80.
Lasfer, M.A., 1996.Taxes and dividends:The UK Pension Funds and their Advisers, 1996.
evidence. Journal of Banking and Finance 20, London.
455–472. Porter, M.E., 1997. Capital choices: Changing
Lasfer, M.A., 1997. On the motivation for pay- the way American invests in industry. In:
ing script dividends. Financial Management Chew, D.H. (Ed.), Studies in International
26, 62–80. Corporate Finance and Governance
Lin, T.H., 1996. The certification of large block Systems: A Comparison of the US, Japan
shareholders in initial public offerings: The and Europe. Oxford University Press, New
case of venture capitalists. Quarterly Journal York, pp. 5–17.
of Business and Economics 35, 55–65. Prowse, S., 1994. Corporate governance in and
Lins, K., Servaes, H., 1999. International evi- international perspective: A survey of corpo-
dence on the value of corporate diversifica- rate control mechanisms amongst large firms
tion. Journal of Finance 54, 2215–2239. in the United States, the United Kingdom,
London Stock Exchange, 1995. Survey on Japan and Germany. BIS Economic Papers,
Share Ownership, Stock Exchange, London. No. 41, Bank of International Settlements,
Mallin, C.A., 1997. Investors voting rights. In: Basel.
Keasey, K., Wright, M. (Eds.), Corporate Romano, R., 1994. Public pension fund activism
Governance: Responsibilities, Risks and in corporate governance reconsidered.
Remuneration. Wiley: West Sussex. Columbia Law Review 93, 795–853.
Maug, E., 1998. Large shareholders as moni- Shleifer, A., Vishny, R.W., 1986. Large share-
tors: is there a trade-off between liquidity and holders and corporate control. Journal of
control? Journal of Finance 53, 65–98. Political Economy 94, 461–488.
McConnell, J.J., Servaes, H., 1990. Additional Shleifer, A., Vishny, R.W., 1997. A survey of cor-
evidence on equity ownership and corporate porate governance. Journal of Finance 52,
value. Journal of Financial Economics 27, 737–783.
595–612. Shome, D.K., Singh, S., 1995. Firm value and
McConnell, J.J., Servaes, H., 1995. Equity own- external blockholdings. Financial
ership and the two faces of debt. Journal of Management 24, 3–14.
Financial Economics 39, 131–157. Short, H., Keasey, K., 1997. Institutional share-
Mikkelson, W.H., Ruback, R.S., 1985. An holders and corporate governance in the
empirical analysis of the interfirm equity United Kingdom. In: Keasey, K.S., Thompson,
investment process. Journal of Financial S., Wright, M. (Eds.), Corporate Governance:
Economics 14, 523–553. Economic, Management, and Financial
DO OCCUPATIONAL PENSION FUNDS MONITOR COMPANIES? 253
Issues. Oxford University Press, Oxford, Sudarsanam, S., 1996. Large shareholders,
18–53. takeovers and target valuation. Journal of
Smith, M.P., 1996. Shareholder activism by Business Finance and Accounting 23,
institutional investors: evidence from 295–314.
CalPERS. Journal of Finance 51, 227–252. Tomlinson, L., 1998. Should the Future Be All
Stapledon, G.P., 1996. Institutional Passive? Paper presented at the 1998 NAPF
Shareholders and Corporate Governance. Conference, Eastbourne, UK.
Clarendon Press, Oxford. Wahal, S., 1996. Pension fund activism and
Strickland, D., Wiles, K.W., Zenner, M., 1996. A firm performance. Journal of Financial and
requiem for the USA: Is small shareholder Quantitative Analysis 31, 1–23.
monitoring effective? Journal of Financial WM, 1996. UK Pension Fund Industry Results.
Economics 40, 319–338. WM, Edinburgh (various issues 1994–1996).
Stulz, R.M., 1988. Managerial control of voting Yermack, D., 1996. Higher market valuation of
rights: Financing policies and the market for companies with a small board of directors.
corporate control. Journal of Financial Journal of Financial Economics 40,
Economics 20, 25–54. 185–211.
Part 3
Corporate governance,
underperformance and
management turnover
INTRODUCTION
In the latter case, some of the mechanisms may be redundant although, ex ante,
determining which ones are redundant may be difficult. In the second paper in this part,
Luc Renneboog (Ch.11) examines how corporate control is exerted in companies listed on
the Brussels Stock Exchange. There are several alternative corporate governance mecha-
nisms, which may play a role in disciplining poorly performing management: block-holders
(holding companies, industrial companies, families and institutions), the market for partial
control, creditors, and board composition. Even if there is redundancy of substitute forms
of discipline, some mechanisms may dominate. It is found that top managerial turnover is
strongly related to poor performance measured by stock returns, accounting earnings in
relation to industry peers and dividend cuts and omissions. Tobit models reveal that there
is little relation between ownership and managerial replacement, although industrial
companies resort to disciplinary actions when performance is poor. When industrial
companies increase their share stake or acquire a new stake in a poorly performing company,
there is evidence of an increase in executive board turnover, which suggests a partial
market for control.There is little relation between changes in ownership concentration held
by institutions and holding companies, and disciplining. Still, increased disciplining also
follows high leverage and decreasing solvency and liquidity variables, as are a high propor-
tion of non-executive directors and the separation of the functions of CEO and chairman.
The third paper by Julian Franks, Colin Mayer and Luc Renneboog (Ch.12) follows a
similar line of empirical enquiry into the relative effectiveness of different disciplinary
devices but this time in the UK context and the focus is on financially distressed firms.
Economic theory points to five parties disciplining management of poorly performing
firms: holders of large share blocks, acquirers of new blocks, bidders in takeovers,
non-executive directors, and investors during periods of financial distress. This paper
reports a comparative evaluation of the role of these different parties in disciplining man-
agement. The authors find that, in the United Kingdom, most parties, including holders of
substantial share blocks, exert little disciplining and that some, for example, inside holders
of share blocks and boards dominated by non-executive directors, actually impede it.
Bidders replace a high proportion of management of companies acquired in takeovers
but do not target poorly performing management. In contrast, during periods of financial
constraints prompting distressed rights issues and capital restructuring, investors focus
control on poorly performing companies. These results stand in contrast to the United
States, where there is little evidence of a role for new equity issues but non-executive
directors and acquirers of share blocks perform a disciplinary function.The different gov-
ernance outcomes are attributed to differences in minority investor protection in two
countries with supposedly similar common law systems. This paper makes the interesting
link between the relative effectiveness of different corporate control mechanisms and the
larger legal environment in which these mechanisms operate.
In 1992, the Cadbury Committee issued the Code of Best Practice (the Code). It rec-
ommends that boards of UK corporations include at least three outside (i.e. independent)
directors and that the positions of chairman and CEO be held by different individuals (the
non-duality principle). The underlying presumption is that these recommendations would
lead to improved board oversight. Jay Dahya, John McConnel and Nick Travlos (Ch.13)
empirically investigate the impact of the Code by comparing the relationship between
CEO turnover and corporate performance before and after the Code. CEO turnover
increased following issuance of the Code; the negative relationship between CEO turnover
and performance became stronger following the Code’s issuance; and the increase in sen-
sitivity of turnover to performance was concentrated among firms that adopted the Code.
This Code is now part of the listing requirements on the London Stock Exchange on the
‘comply or explain’ (non-compliance) basis. Dahya et al.’s study is an interesting example
of empirical methodology for evaluating externally imposed corporate governance regimes
in their impact on corporate behaviour and performance.
Chapter 10
Paolo F. Volpin
GOVERNANCE WITH POOR
INVESTOR PROTECTION:
EVIDENCE FROM TOP EXECUTIVE
TURNOVER IN ITALY*
Source: Journal of Financial Economics, 64(1) (2002): 61–90.
ABSTRACT
This paper studies the determinants of executive turnover and firm valuation as a function of
ownership and control structure in Italy, a country that features low legal protection for
investors, firms with controlling shareholders, and pyramidal groups. The results suggest that
there is poor governance, as measured by a low sensitivity of turnover to performance and a
low Q ratio, when (i) the controlling shareholders are also top executives, (ii) the control is fully
in the hands of one shareholder and is not shared by a set of core shareholders, and (iii) the
controlling shareholders own less than 50% of the firm’s cash-flow rights.
interest between controlling and minority between voting and cash flow rights (for an
shareholders) is likely resolved in favor of example of pyramid, see Figure 1).
the second one, leading to a magnification The first finding in the analysis is that
rather than a reduction of the agency problem. controlling shareholders are entrenched.
This paper provides direct evidence on the Indeed, the probability of turnover and its
potential costs of pyramidal groups and sensitivity to performance are significantly
nonvoting shares by analyzing executive lower for top executives who belong to the
turnover and market valuation in Italian family of the controlling shareholder than
listed companies. for other executives. Second, the larger the
Italy represents an ideal setting to fraction of cash-flow rights owned by the
address these issues because it features controlling shareholder, the more sensitive
weak legal protection of creditors and turnover is to performance. This result
shareholders, inefficient law enforcement, suggests that incentives matter and that
high ownership concentration, and an abun- governance improves when the controlling
dance of pyramidal groups and nonvoting shareholder internalizes the consequences
shares. There is suggestive evidence that of his actions. The third finding is that
the size of the private benefits of control is turnover is more sensitive to performance
particularly large in Italy. Zingales (1994) when a voting syndicate controls the firm.
finds an average voting premium of 82% in A voting syndicate is a coalition of relevant
Italian companies with dual-class shares, shareholders who sign a binding agreement
while the average voting premium in the to vote together for a few years. About 15%
US is 10%, in the UK is 13%, in Canada of the companies in the sample have a voting
23%, and in Switzerland is 27%. syndicate. These coalitions help the largest
The existing literature indicates two shareholder to control a company when his
strategies to assess the effectiveness of a stake would not be large enough to do so by
corporate governance system.The first one, himself. This result suggests that turnover
following Kaplan (1994a) and Coffee becomes more sensitive to performance
(1999), is to test whether executive when control is, to some extent, contestable,
turnover increases as a firm’s performance as in the case of a voting syndicate.
declines. The second one, derived from These findings are confirmed by the
Morck et al. (1988) and McConnell analysis of the firm’s Q ratio: Q is signifi-
and Servaes (1990), is to analyze the cantly smaller in firms where the control-
firm’s valuation in relation with similar ling shareholders are among the top
companies. executives, is significantly larger when con-
Accordingly, this paper will first study trol is partially contestable as in the case
the determinants of executive turnover in of a voting syndicate controlling the firm,
Italian publicly traded companies, by and increases with the fraction of cash-flow
focusing on how the ownership and control rights owned by the controlling shareholder.
structure of a firm affects the sensitivity of Within pyramidal groups, I find a signif-
the firm’s executive turnover to perform- icant lower Q ratio (between 13% and
ance. Then, it will evaluate the effect of 27%) in firms at the bottom of a pyramid.
these same factors on the firm’s Q ratio. This result is consistent with the argument
Both analyses are based on a large data that pyramids increase agency problems by
set, which covers all traded companies in creating a wedge between voting and cash-
Italy (banks and insurance companies flow rights. A possible explanation is that
excluded) during the period 1986–1997 good managers are promoted to a higher
and contains information on ownership, layer of the pyramidal group against the
board, and capital structures. For all com- interests of investors in the firms at the
panies in the sample, I was able to trace bottom of the pyramid. Indeed, the rela-
back the control chain, identify the ulti- tionship between turnover and performance
mate owner, and determine his ownership is weaker in pyramidal groups although the
stake in the company, distinguishing difference is not statistically significant.
GOVERNANCE WITH POOR INVESTOR PROTECTION 259
Regarding the role of large minority Hence, the basic empirical hypothesis to
shareholders, they do not seem to play an be tested in this paper is the following.
important monitoring role within the firm.
Basic Hypothesis. Top executive turnover is
Specifically, the results in this paper
negatively related to performance.
suggest that minority shareholders have a
If the basic hypothesis is verified on the
governance role only if their votes are
whole sample, the second step is to study
necessary to the controlling shareholder
whether there are significant differences
to control the firm, as is the case in voting
across firms. An important factor that may
syndicates. Except for their role in a voting
affect turnover is the ownership structure
syndicate, large minority shareholders do
of a firm.
not improve the firm governance since
Firms on the Italian stock market typi-
they do not increase the sensitivity of
cally have a clearly identifiable controlling
turnover to performance nor the firm’s
shareholder who controls at least 20% of
valuation.
the voting rights. Exceptions are banks and
Finally, turnover is much lower in the
insurance companies. However, these are
company at the top of a pyramid (6%)
excluded from the sample because their
than in its subsidiaries (16%). This result
accounting measures of performance are not
may be explained by the fact that the con-
directly comparable with the other firms.
trolling shareholders of the group sit as
Hence, Italian traded companies can be
executives of their holding companies and
classified into four large categories according
they are entrenched in control. They do so
to their ultimate owner: the state, a foreign
because the benefits of control are larger in
company, a set of banks, and one or more
the holding company, as suggested by the
Italian individuals (I define this last
finding that the voting premium in the holding
category as family-controlled firms). One
companies is significantly higher than in
may expect to find differences across these
the subsidiaries.
groups in the sensitivity of turnover to per-
The structure of the paper is as follows.
formance. For example, in state-controlled
Section 2 formulates the hypothesis to test
companies management turnover can be
by overviewing the literature and describing
more affected by political than economic
the Italian corporate governance system.
factors. Therefore, I first check whether
Section 3 describes the data set. Section 4
there are differences across these
contains and discusses the results on the
categories. However, in order to use a
determinants of top executive turnover and
homogeneous and large set of observations,
firm valuation. Section 5 extends the analysis
I focus the remainder of the analysis on the
to pyramidal groups and evaluates the role
set of firms that are controlled by Italian
of the market for corporate control. The
individuals.
conclusion is in Section 6.
Consistent with the result above, Denis legally binding only in the former case. It is
et al. (1997) show that the probability of important to notice that the degree of
top executive turnover (and the sensitivity entrenchment of the controlling share-
of turnover to performance) is negatively holder could be much lower if he or she
correlated with the ownership stake held needs a voting syndicate to control the
by officers. Most Italian traded companies firm. With a voting syndicate, the controlling
are majority owned. Moreover, the size of shareholder does not have a lock on control
the private benefits of control is extremely and control is partially contestable.
large in Italy, as shown by Zingales An example is given by the turnover in
(1994). It is possible that the controlling Olivetti in 1996 (Il Sole 24 Ore, September
shareholder is entrenched as a top execu- 4, 1996). Carlo De Benedetti, the long-time
tive against the interest of the other Chairman and President of the Board, was
shareholders, in order to preserve his ability the relative majority shareholder in Olivetti
to extract those benefits. Hence, the with 15% of the voting rights. Thanks to a
first main hypothesis to test is the voting syndicate he controlled another
entrenchment hypothesis. 25% of the votes. In January 1996, after
several years of very poor performance at
Entrenchment Hypothesis. Top executive Olivetti, the voting syndicate broke down,
turnover is lower and less sensitive to per- and in September of the same year De
formance if the controlling shareholder is Benedetti was forced to step down from all
an executive. executive roles in the company. This case
In Italy, the separation between owner- suggests that executive turnover is more
ship and control is enhanced by the wide- sensitive to performance if the controlling
spread use of traded pyramids and shareholder does not have absolute control
nonvoting shares. The sensitivity of over the company, that is, in the instances
turnover to performance can be proportion- where there exists a voting syndicate.
ate to the fraction of cash-flow rights However, it is also conceivable that voting
owned by the controlling shareholder. One syndicates sustain collusive agreements
immediate rationale for this hypothesis is among large families aiming at preserving
that monitoring the management may the stability of control. In this second case,
come at a cost. Hence, the higher the voting syndicates do not necessarily
fraction of cash-flow rights owned by increase turnover or the sensitivity of
the controlling shareholder, the larger the turnover to performance.
controlling shareholder’s incentive to Pagano and Roell (1998) argue from a
monitor the management. theoretical viewpoint that large minority
shareholders play a role in monitoring the
Incentive Hypothesis. Top executive
controlling shareholder. For the US, Denis et
turnover is more sensitive to performance if
al. (1997) show that the probability of top
the controlling shareholder owns a larger
executive turnover is positively correlated to
fraction of cash-flow rights.
the presence of an outside blockholder.
In about 15% of Italian traded compa-
nies, a coalition of important shareholders Outside Monitoring Hypothesis. Top execu-
helps the controlling party control the com- tive turnover is higher and more sensitive to
pany. These shareholders are kept together performance in companies with a large
by explicit agreements to vote together, minority shareholder and/or a voting
which are called voting syndicates (“sindacati syndicate.
di voto”). These agreements are publicly An alternative way to test the quality of
announced on national newspapers, last for the corporate governance within a firm is to
a fixed number of years (usually three) and look at the valuation of the firm. For the
can be renewed. A voting syndicate can US, Morck et al. (1988) and McConnell
decide on its actions either unanimously or and Servaes (1990) find a nonlinear rela-
by majority rule. According to law experts tionship between Q ratio and managerial
(see Galgano, 1997), these agreements are ownership. A similar approach is employed
GOVERNANCE WITH POOR INVESTOR PROTECTION 261
also in Yermack (1996) and in La Porta Since Italian firms typically have a
et al. (1999b). controlling shareholder, a sale of the
If the absence of sensitivity of turnover controlling stake is a simple proxy for a
to performance is an indicator of bad change in the firm’s ownership structure
governance, this should be reflected in the and should be associated with an increase
firm’s valuation. Hence, I will test the in top executive turnover. However, since the
following hypotheses on the Q ratio. sale can only happen with the consent of
the controlling shareholder, control is not
Entrenchment Hypothesis. Q is lower if the
contestable, and takeovers in Italy can have
executives are controlling shareholders.
a more limited disciplinary role than in the
Incentive Hypothesis. Q is higher if the con- US or UK.
trolling shareholder owns a larger fraction Gilson (1989), for the US, and Franks
of the cash-flow rights. et al. (2001), for the UK, find that turnover
is higher in firms at the onset of a financial
Outside Monitoring Hypothesis. Q is higher
crisis, when the firm’s creditors increase
in firms where there are outside blockhold-
their pressure on the management and seize
ers and/or a voting syndicate.
control. Transfer of the control to creditors
should then be associated with an increase in
2.2. Pyramidal groups top executive turnover in the Italian sample.
The impact of transfers of control as a
The above hypotheses will be tested in determinant of executive turnover will be
Section 4 of this paper and will help char- evaluated in Section 5.2.
acterize the relationship between executive
turnover and ownership structure in Italian
firms. However, as mentioned before, many 3. DESCRIPTION OF THE DATA SET
firms in Italy (more than half of the firms
traded on the Milan stock exchange) are The sample is collected from several issues
organized in pyramidal groups. Pyramids of of “Il Taccuino dell’Azionista”, an annual
traded firms magnify the separation publication edited by Il Sole 24 Ore. This
between ownership and control because source provides basic balance sheet data,
they allow the controlling shareholder of information about the ownership structure,
the holding company at the top of the and the names of the individuals sitting on
pyramid to control the companies in the the Board of Directors (“Consiglio di
pyramid by owning a small fraction of their Amministrazione”) of all companies traded
capital. By the same token, they magnify on the Italian stock market. Data cover a
the potential conflict of interest between period of 12 years, from 1986 to 1997.
controlling and minority shareholders. The From the set of all companies traded on the
impact of pyramidal groups on executive Milan stock exchange over that period,
turnover and firm valuation is studied in I exclude banks, insurance companies, and
Section 5.1. foreign companies because of different
accounting rules. Foreign companies are
2.3. Transfer of corporate control firms incorporated abroad, while foreign-
controlled companies are firms incorpo-
Finally, the paper will try to evaluate the rated in Italy even though owned by
relative role of external and internal gover- foreigners. The latter ones are included in
nance forces. Martin and McConnell the sample. I exclude all companies that
(1991), for the US, and Franks and Mayer were traded on the Milan stock exchange
(1996), for the UK, find that turnover for less than three years because I need at
increases following takeovers. Barclay and least three years of data to compute the
Holderness (1991), for the US, and Franks measures of performance and turnover.
et al. (2001), for the UK, find a similar Hence, the sample used in the regres-
increase in turnover following block trades. sions covers 205 firms and contains a total
262 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Table 10.1 Descriptive statistics: sample of all firms traded on the Milan Stock Exchange
The table reports the number of observations, the mean, median, standard deviation, minimum, and
maximum for some of the variables used in the analysis. Top executive turnover is a dummy variable
that takes value 1 in year t if at least half of the top executives are replaced between t and t 1.
The sample includes all companies traded on the Milan Stock Exchange, excluding banks, insurance,
and pure financial companies, in the years 1987 through 1996.The number of observations is 1,611.
Ownership structure
Fraction of cash-flow rights owned by 38.0 40.5 25.0 0.3 99.4
controlling shareholder (%)
Fraction of voting rights controlled by 56.4 53.8 14.7 20.1 100
controlling shareholder (%)
Board composition
Number of directors 10.4 10 3.3 3 25
Number of top executives 3.34 3 0.68 1 6
Turnover data
Fraction of top executives replaced in 16.9 0 25.3 0 100
a year (%)
Top executive turnover 0.14 0 0.35 0 1
of 1,611 observations. Table 10.1 reports ultimate owner is the state or a government
summary statistics on ownership structure, agency (216 observations). In 55 observations
board composition, and executive turnover the controlling party is a group of banks,
for this sample. All variables used in the since the company defaulted and the banks
analysis are defined in Appendix A. took control. The rest of the observations
Italian law limits the extent of cross- (1,234) have one or more private Italian
holdings to 2% of voting rights among citizens or an Italian family as the ultimate
traded companies. Moreover, controlled owner (I will call this last set family-
firms cannot exercise the voting rights controlled firms).
eventually owned in their parent company. In Table 10.1, two variables describe the
Hence, it is simple to identify the control firm ownership structure. The fraction of
chain once ownership data are available for cash-flow rights owned by the largest
all companies. An example of a common shareholder is defined as the product of the
control structure is represented in Figure fraction of voting rights along the controlling
10.1. The figure shows the structure of the path.This number is corrected for nonvoting
Pesenti group at the end of 1995. The shares by assuming that the ultimate owner
Pesenti family is defined as the controlling owns none of them. For example in Figure.
shareholder because an individual, Rosalia 10.1, the fraction of cash-flow rights owned
Radici Pesenti, owns the controlling stake by the Pesenti family in Italmobiliare is
in the holding company (44.87%). 29.4%, while the fraction of voting rights
The observations have been classified is 44.8% because about a third of
into four categories according to the Italmobiliare’ s equity is made up of nonvot-
information available on the controlling ing shares. The fraction of cash-flow rights
shareholder. A firm is classified as foreign- owned by the Pesenti family in Italcementi
controlled in a given year if the ultimate is 9.5%, which is the product of 29.4% (the
owner is a foreign company (106 observa- fraction of cash-flow rights directly owned
tions); and is defined as state-controlled if the by the family in the holding company
GOVERNANCE WITH POOR INVESTOR PROTECTION 263
Pesenti Family
Italmobiliare
Giampiero Pesenti (P and CD)
Figure 10.1 Structure of the Pesenti’s group as of 12/31/1995.The figure shows the ownership and
control structure of the traded companies controlled by the Pesenti family. Each box
represents a traded company. Inside each box are the name of the company and the
names of the top executives with their role on the board of directors: P President,
CD CEO (“Consigliere Delegato”), VP Vice-President. The arrow indicates the
direction of control. The numbers above each box represent the percentages of cash-
flow (CF) and voting (V) rights directly owned by the controlling party (individual or
company). In parenthesis are the fractions of cash-flow and voting rights directly or
indirectly owned by the ultimate owner (the Pesenti family).These are computed according
to the definitions reported in Appendix A.
Italmobiliare) and 32.4% (the fraction of exercise its voting rights on Italmobiliare.
cash-flow rights owned by Italmobiliare in This represents the burden of having an
Italcementi). The fraction of voting rights extra layer of control. Even if Italmobiliare
controlled by the controlling shareholder is owns 54.26% of the voting rights in
computed as the minimum share of voting Italcementi, the Pesenti family owns only
rights controlled by the controlling share- 44.87% of the voting rights in
holder along the control path. Italmobiliare. Hence, the effective voting
The rationale for this definition is power of the Pesenti family in Italcementi
explained as follows with the help of Figure is equal to 44.87%. The logic of this
10.1. Let us consider the second layer example simply generalizes to more complex
firm Italcementi and try to evaluate the control structures. In the case of a syndi-
voting rights of the Pesenti family in this cate among shareholders, the controlling
company. The Pesenti family controls party is assumed to control all the voting
Italcementi through the holding company rights that belong to the syndicate.
Italmobiliare. In order to exercise the As measure of executive turnover, I use
voting rights that Italmobiliare owns in the fraction of top executives replaced
Italcementi, the Pesenti family needs to within a year. This is preferable to the CEO
264 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
turnover because in Italian firms all top 99.4%. Disciplinary turnover (measured as
executives (Presidente, Vice-Presidente, top executive turnover) happens in 14% of
and Amministratore delegato) have similar the observations.The median number of top
executive power and there is no clear rank- executives in the sample is three, while the
ing in authority among them (especially median size of the board is ten directors.
between Presidente and Amministratore
delegato). Also, some firms do not have a
CEO; others have many CEOs. The data 4. EMPIRICAL ANALYSIS
have been cleaned of the cases of retire-
ment identifiable through LEXIS-NEXIS In this section, I will test each of the
(42 observations are relabeled as non- hypotheses discussed in Section 2, starting
turnover in this way). However, Italian com- with the relationship between turnover and
panies do not provide information about performance in the whole sample.
the age and tenure of their executives in the
annual report. Since I was able to collect 4.1. Turnover and performance
data on age and tenure for only a subset of
the executives, I exclude these two vari- In Table 10.2, the observations are classified
ables from the regressions.2 In order to according to the ultimate owner in family-,
reduce the biases due to potential mislabel- state-, foreign-, and bank-controlled firms.
ing of individual cases of turnover, in the The table shows that turnover is negatively
regressions I use as dependent variable a related to performance in the sample as a
dummy variable (Top executive turnover) whole. More precisely, the observations are
that takes value one in year t when at least sorted into quintiles according to the firm’s
half of the top executives are replaced past performance. I then compare turnover
between t and t 1. between the worst quintile and the best
Two alternative measures of perform- one. In Panel A, performance is proxied by
ance are used in the analysis: the change in the change in the ratio of EBIT over assets,
the ratio of earnings before interest and while in Panel B is measured by the stock
taxes (EBIT) and total assets between return.
t 1 and t and the annual stock return In the whole sample, the probability of
between t 1 and t. The former is the top executive turnover is significantly lower
measure of performance used in Denis and if performance is good than if is bad. In
Kruse (2000). The latter one, the stock Panel A, the average turnover increases
return, is the standard measure of perform- from 14.5% in best performing firms to
ance used in studies of executive turnover 22.5% in worst performing ones. Similarly,
since the contributions by Warner et al. in Panel B, turnover increases from 14.5%
(1988) and Weisbach (1988). In unre- in best performing firms to 19.3% in worst
ported regressions I performed the performing ones. In Panel A, the average
analysis with the addition of lagged turnover decreases monotonically as
measures of performance without finding performance improves across quintiles with
any difference. the significant jump being the one
As illustrated in Table 10.1, all compa- between the first and second quintiles. In
nies in the sample have a controlling share- Panel B, the average turnover is not monot-
holder, that is a shareholder who controls a onic and the third and fourth quintiles are
fraction of voting rights larger than 20%. not significantly different from the first
The fraction of voting rights held by the one, as far as turnover is concerned.
controlling shareholder is on average 56%, When examining turnover across types of
while it ranges between 20.1% and 100%. ultimate owner, one finds a negative and
There is a large variability in the fraction of significant relationship between turnover and
cash-flow rights owned by the controlling performance according to either perform-
shareholder. The average is 40%, while the ance measure only for family-controlled
minimum is 0.3% and the maximum firms. It is interesting to notice that in
GOVERNANCE WITH POOR INVESTOR PROTECTION 265
Table 10.2 Turnover and performance: family-, state-, foreign-, and bank-controlled firms
The table reports the average fraction of top executives replaced by quintiles of performance. The
observations are sorted in five classes according to their past performance (1 low, 2–5 high).
The average turnover is compared between the firms with low performance and those with high
performance using a two-tailed t-test. The observations are also classified according to the type of
ultimate owner (a family, the state, a foreign company, a bank).
consistent with the definition of turnover, The second dummy variable (owner with
the owner-manager dummy variable is set high incentives) identifies the companies in
equal to one when at least half of the top which the controlling shareholder owns a
executives belong to the family of the con- large fraction of the cash-flow rights. More
trolling shareholder. This happens in about specifically, I define as such all companies
30% of the observations. The purpose of in which the controlling shareholder owns
this first dummy variable is to address the more than 50% of the cash-flow rights, a
Entrenchment Hypothesis described in requirement that roughly selects 30% of
Section 2. the observations. The purpose of this
268 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
second dummy variable is to address the The results in Table 10.4 are as follows.
Incentive Hypothesis. Regressions (1) and (2) show that turnover
According to the Outside Monitoring does not always decrease with performance
Hypothesis discussed in Section 2, when the latter is proxied by the change in
turnover should be more sensitive to earnings. Specifically, in regression (1), a
performance when the company has large 10% decrease in earnings over assets
minority shareholders. The intuition is that increases the probability of executive
large minority shareholders can play an turnover by 5.8%. However, in regression
active role in corporate governance. This (2), the increase in the probability of exec-
hypothesis is tested in Table 10.4 by using utive turnover for a similar change is only
two alternative proxies for minority share- 2.8% and is not statistically significant.
holders.The first is an indicator of whether When the controlling shareholder sits as
the second largest shareholder owns a top executive of the firm, the relationship
fraction larger than 5% of the voting between turnover and performance
rights (large minority shareholder). This becomes even weaker, as shown by the pos-
happens in about 40% of the observations. itive (although not significant) coefficient
The second one identifies cases where the on the first interactive term. The level of
firm is controlled by a voting syndicate, as turnover is also significantly lower when a
happens in 15% of the observations member of the family of the controlling
(voting syndicate). shareholder sits among the top executives.
The methodology I use to address the The probability of executive turnover
hypotheses above is to run a probit regression decreases with 6.5% when the controlling
with the dummies just described entering shareholder is a top executive.These results
both additively and in interaction with the are weakly in favor of the Entrenchment
measure of performance. Eq. (1) describes Hypothesis.
the probit model estimated in Table 10.4: The first two regressions also show that
the sensitivity of turnover to performance
Pr(Turoverit 1) Φ(t 0 Performanceit increases significantly when the controlling
4 shareholder owns a large stake in the firm.
(
j1
j Performanceit If the largest shareholder owns more than
50% of the cash-flow rights, a 10%
j
j)Dit
Sizeit decrease in earnings over assets increases
Φ Industury dummyit turnover by 10%. This result is in favor of
it), (1) the Incentive Hypothesis.
Regarding the Outside Monitoring
where Φ(.) is the cumulative gaussian Hypothesis, regression (1) suggests that
distribution, t is the year dummy, and the presence of a large minority share-
firm’s size and an industry dummy (firms holder does not increase the sensitivity of
are classified in nine industries according turnover to performance. However, regres-
to the official classification on the Milan sion (2) shows that a voting syndicate does
stock exchange) are introduced as control indeed increase the sensitivity of turnover
variables.3 to performance. After a 10% decrease in
In the regression, I cannot control for earnings, executive turnover is 7% more
firm effects, as with traditional linear models, likely if the firm is controlled via a voting
because a fixed-effect estimator is not syndicate than otherwise. The intuitive
available in a probit model. Since the obser- explanation for these results is that minority
vations are likely not to be independent, the shareholders have the power to play a
standard errors are corrected for correlation governance role (and they do so) only when
and clustering at the firm level. In an control is not locked in the hands of the
unreported regression, I estimated a linear controlling shareholder. That is, only if the
OLS model with fixed effects at firm level, controlling shareholder needs a voting
finding similar results to the ones reported. syndicate to control the firm.
GOVERNANCE WITH POOR INVESTOR PROTECTION 269
Regressions (3) and (4), in which (1988) and McConnell and Servaes
performance is proxied by the stock return, (1990). This analysis is performed in
do not produce any significant coefficient Table 10.5. The results of four fixed-effect
on the interactive terms. The signs on the regressions are represented in the table:
coefficients are consistent with the Outside regressions (1) and (2) include firm
Monitoring Hypotheses, but against effects; regressions (3) and (4) include
the Entrenchment and the Incentive only industry effects but report standard
Hypothesis.The results in these two regres- errors corrected for clustering at the firm
sions can be explained by the fact that the level. The rationale to present the second
stock return is not likely an ideal measure pair of regression together with the first
of performance in the sample of Italian one is that controlling for firm effects, as in
firms. The stock return is a noisy measure (1) or (2), may reduce the significance of
of performance for many companies in the the coefficients because it eliminates most
sample because many stocks suffer a lack of the variability in the firm ownership
of liquidity and infrequent trades. structure, which is relatively constant
across years. As done in Table 10.4, in
regressions (1) and (3) the power of minor-
4.3. Valuation
ity shareholders is measured by the dummy
An alternative procedure to assess the variable large minority shareholders; in
efficiency of a governance regime is to eval- regressions (2) and (4) minority shareholders’
uate the impact of ownership structure on power is proxied by the dummy variable
firm valuation, as done by Morck et al. voting syndicate.
The results on Q match one-to-one those that turnover is significantly higher for
obtained in the analysis of turnover: firms other managers (18.4%) than for owner-
where turnover is unaffected by perform- managers (6.5%). Similarly, the sensitivity
ance are discounted by the market. Indeed, of turnover to past performance is signifi-
Table 10.5 shows that Q is significantly cantly higher for other managers than for
smaller (between 8% and 11% discount) owner-managers. In Panel B, the turnover
for firms in which relatives of the control- of family and non-family executives is com-
ling shareholder are among the top pared by running a regression on both sets
executives. This is consistent with the of data with a dummy variable (entering
results in Table 10.4 and with the both additively and in interaction with per-
Entrenchment Hypothesis. Table 10.5 also formance) identifying whether the data on
shows that Q is significantly larger when turnover refers to family or other execu-
control is partially contestable as in tives.The findings are that both the level of
the case in which a voting syndicate con- turnover and its sensitivity to performance
trols the firm (between 14% and 40% is significantly lower for family managers.
premium).This result is consistent with the In regression (1), a 10% decrease in earn-
findings in Table 10.4 and with the Outside ings over assets implies a 6% increase in
Monitoring Hypothesis. Finally, the table the probability of turnover for a non-family
shows that Q increases only weakly with executive and only a 2% increase for a
the fraction of cash-flow rights owned family executive. In regression (2), a real-
by the controlling shareholder: if the ization of the stock return equal to 10%
controlling shareholder owns more than causes the probability of turnover to
50% of the cash-flow rights, the firms increase by 0.6% for non-family managers
trades at a premium estimated between and only by 0.3% for family ones. These
0% and 8%. This final result supports the results suggest that, when a company is in
Incentive Hypothesis. Consistent with the trouble, non-family managers are likely to
finding in Table 10.4, the dummy for large be replaced, but family managers are likely
minority shareholder does not affect the Q to stay, that is, family managers are
ratio, as shown by regressions (1) and (3) entrenched.
of Table 10.5.
4.5. International comparison
4.4. More on the Entrenchment
This section compares the findings on Italy
Hypothesis
with evidence available on other countries.
An alternative test of the Entrenchment Few studies have produced results that are
Hypothesis is to measure directly the directly comparable across countries
probability of turnover and its sensitivity to because of differences in the econometric
performance (for executives who belong to procedures and in the performance and
the family of the controlling shareholder turnover measures used. A partially com-
and those who do not), and compare them. parable set of coefficients can be obtained
This is done in Table 10.6. from Kaplan (1994a, b) for Germany,
The executives in the sample are classi- Japan, and the US. In Table 10.7, I repro-
fied into two groups: family-executives duce the results in these two papers and
(that is, firm’s executives who belong to the add comparable results obtained from the
family of the controlling shareholder of the sample of (family-owned) Italian firms.
firm) and other executives (that is, execu- The coefficients on Germany are
tives who do not belong to the family of the extracted from Kaplan (1994b),Table 10.2.
controlling shareholder). For each group, The measure of turnover is the fraction of
year, and company, I compute the fraction members of the management board
of executives replaced in one year. The replaced in one year. Similarly, turnover in
average turnover is reported in the third the Italian sample is measured by the frac-
column of Table 10.6, Panel A, which states tion of top executive replaced in one year.
GOVERNANCE WITH POOR INVESTOR PROTECTION 271
The results on Japan and US are from Italy as much as in the other countries.This
Kaplan (1994a),Table 10.2, where turnover finding holds across all performance
is measured over a two-year period and measures. Also, the magnitude of the coef-
refers to representative directors in Japan ficients is not significantly different across
and executive directors in the US. Due to the countries. However, in the Italian sample
difference in the time horizon of the two performance explains a much smaller frac-
turnover measures, the results are only par- tion of the variability in turnover than in
tially comparable. Indeed, both turnover and Germany, Japan and US. The R2 ranges
performance are computed over a two-year between 2% and 4% in Italy, between
period for Japan and the US, and over a one- 12% and 15% in Japan and the US, and
year period for Germany and Italy. between 5% and 10% in Germany. This
The results in Table 10.7 show that suggests that, with respect of the other
turnover is correlated with performance in countries, executive turnover in Italy is
272 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Table 10.7 International comparison
OLS regressions. The dependent variable is the fraction of top executives replaced in a year for Italy
and the fraction of members of the management board replaced for Germany. For Japan the dependent
variable is the fraction of representative directors replaced in a two-year period and for the US the
fraction of executive directors replaced over the same horizon. Stock returns, sales growth, earnings
growth, and negative net income dummies are used as performance measures. A separate regression
is run for each performance measure where the latter enters contemporaneously and once lagged. For
Italy and Germany, one lag is one year; for Japan and the US, one lag is two years. In Panel D for
Japan and US, the regression does not contain the lagged negative income dummy. Robust standard
errors are in parentheses. Year dummies are included but the coefficients are not reported.
A. Stock return
Contemporaneous 0.034 0.080* 0.056 0.072**
(0.026) (0.046) (0.038) (0.029)
Once lagged 0.043** 0.103** 0.100*** 0.081***
(0.022) (0.046) (0.037) (0.030)
R2 0.020 0.102 0.139 0.148
B. Sales growth
Contemporaneous 0.063*** 0.066 0.177** 0.171***
(0.015) (0.083) (0.076) (0.083)
Once lagged 0.064** 0.057 0.064 0.031
(0.022) (0.086) (0.076) (0.050)
R2 0.040 0.054 0.133 0.150
C. Change in earnings/assets
Contemporaneous 0.160 0.494 0.121 0.208
(0.197) (0.376) (0.460) (0.147)
Once lagged 0.636*** 0.205 0.707* 0.219
(0.173) (0.485) (0.447) (0.152)
R2 0.030 0.058 0.127 0.130
D. Negative net income
Contemporaneous 0.035* 0.095*** 0.148*** 0.074***
(0.020) (0.034) (0.035) (0.028)
Once lagged 0.066*** 0.042
(0.022) (0.028)
R2 0.026 0.059 0.155 0.137
Mean dependent variable 0.152 0.099 0.285 0.234
* **
, , and *** denote significance at the 10%, 5%, and 1% level, respectively. The results on Germany are from
Kaplan (1994b,Table 2, p. 150); the results for Japan and the US are from Kaplan (1994a,Table 2, pp. 526–527).
explained by other factors than performance, two extensions. First, more than half of
such as ownership and control structure, the companies traded on the Milan stock
personal relationships, family ties, and exchange belong to pyramidal groups. It is
nonobservable performance measures. interesting to see whether the pyramidal
structure affects the dynamics of the
executive turnover in specific ways not
5. EXTENSIONS captured by the analysis in Section 4. The
second extension explores whether the
So far, the paper has evaluated the impact sensitivity of turnover to performance is
of the firm’s ownership structure on top due to internal or external governance
executive turnover. This section considers forces.
GOVERNANCE WITH POOR INVESTOR PROTECTION 273
5.1. Turnover in pyramidal groups pyramidal group (that is, firms that are
controlled directly or indirectly by a
This section evaluates the impact on company of type 3). The table presents the
turnover and valuation of the firm organi- mean of a selection of variables conditional
zational structure distinguishing between on the type of ownership structure.
stand-alone firms and companies that About 55% of the observations in the
belong to pyramidal groups. sample come from pyramidal groups. The
Table 10.8 provides detailed information remaining firms in the sample belong to
on the organizational structure of family- horizontal groups (42 observations only) or
controlled firms. The observations are are stand-alone firms (all the rest). The
divided into four categories: (1) first impact of the pyramidal structure is to
stand-alone firms and firms that belong to create a wedge between the fraction of
horizontal groups (that is, firms that are cash-flow and voting rights owned by the
not controlled by any other traded controlling shareholder. As shown in Table
company and do not own a controlling 10.8, the separation of voting and cash-
stake in any other traded company); (2) flow rights (as measured by the ratio of
firms that are at the top of a pyramidal voting to cash-flow rights) increases from
group (that is, firms that are not controlled 1.4 at the top of the pyramid to seven at
by any other traded company and do con- the bottom of it. The difference between
trol at least one other traded company); voting and cash-flow rights in cate-
(3) firms that are at level 2 of a pyramidal gories (1) and (2) is due to nonvoting
group (that is, firms that are directly con- shares (azioni di risparmio) and voting
trolled by a company of type 2); and (4) syndicates.
firms that are at level 3 or higher of a
in firms at the bottom of a pyramid (those the analysis of the determinants of the
in the category “Pyramidal level 3 ”). executive turnover, the source of the
This is consistent with the recent theoreti- agency costs created by the separation of
cal literature on pyramidal groups (see, ownership and control is to be found in
e.g.,Wolfenzon, 1998; Bebchuk et al., 1998). other areas (like, for example, the firm’s
Since this difference does not show up in investment policy).
276 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
(1) (2)
Pyramidal level 1 dummy 0.086 0.013
(0.065) (0.075)
Pyramidal level 2 dummy 0.030 0.010
(0.097) (0.114)
Pyramidal level 3 dummy 0.268* 0.129*
(0.156) (0.070)
Size 0.085*** 0.058***
(0.028) (0.027)
Fixed effects Firm Industry
Adjusted R2 0.656 0.201
No. of observations 1,234 1,234
* ** ***
, , and denote significance at the 10%, 5%, and 1% level, respectively.
out of the 65 cases of changes of control, shareholder. This result suggests that
16 qualify as forced and 49 as voluntary. governance improves when the controlling
Overall, the results in this section suggest shareholder’s objectives are more aligned
that the market for corporate control does with those of minority share-holders. Third,
not play an important role in the corporate turnover is more sensitive to performance
governance regime operating in Italy. The when control is, to some extent, con-
main reason is probably that control is not testable, as in the case of a voting syndicate.
really contestable, since all firms in the Fourth, aside from the role of a voting
sample have a controlling shareholder. syndicate, large minority shareholders do not
seem to improve governance.
These findings are confirmed by an analysis
6. CONCLUSION of the firms’ Q ratio.I find that Q increases with
the fraction of cash-flow rights owned by the
This paper provides direct evidence on the controlling shareholder, is significantly smaller
performance of the corporate governance for firms in which controlling shareholders are
regime operating in countries characterized among the top executives, is significantly larger
by low legal protection for investors, firms when a voting syndicate controls the firm, and
with large controlling shareholders, and is unaffected by the presence of large minority
some separation of ownership and control shareholders. The combined results on execu-
created via pyramidal groups and nonvoting tive turnover and Q suggest a one-to-one
shares. Studying Italy, a country that shares correspondence between bad governance (that
all these features, I find that the probability is, low sensitivity of turnover to performance)
of turnover and its sensitivity to performance and low valuation (that is, low Q).
are significantly lower for top executives who The paper also provides evidence on
belong to the family of the controlling share- pyramidal groups. I find that executive
holder than for other executives.This result is turnover is not significantly affected by the
evidence that controlling shareholders are firm’s organizational structure. Indeed, the
entrenched. Second, turnover is more sensitive sensitivity of turnover to performance is not
to performance the larger the fraction of significantly different in stand-alone firms
cash-flow rights owned by the controlling and in firms that belong to pyramidal groups.
278 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Within pyramidal groups, turnover is very suggest that the market for corporate control
low in the holding company, while it is still plays a very limited corporate governance
relatively high in the subsidiaries. This result role. According to Bebchuk (1999), large
suggests that pyramidal groups help the con- shareholders stay firmly in control wherever
trolling shareholder select the best managers control is extremely valuable. Ultimately, as
whereby well-performing managers are pro- argued by La Porta et al. (1999a), control is
moted to higher level of the pyramid while valuable because low legal protection for
poor-performing ones are replaced. Finally, investors enables the controlling shareholder
there is a significant discount in firms at the to enjoy large private benefits. This suggests
bottom of a pyramid since the Q ratio is that the solution lies at a political level in new
between 13% and 27% lower in those firms legislation to improve the quality and the
if compared to similar standalone companies. enforcement of investor protection. The
This is consistent with the recent theoretical so-called Draghi Reform introduced in Italy
literature on pyramidal groups, suggesting in 1998 effectively increased the protection
that the separation of ownership and control of minority shareholders and may be one step
created by these organizational structures is in this direction.
likely to generate large agency problems The diffusion among firms of codes of
because of the conflict of interests between best practice is another way to improve
controlling and minority shareholders. corporate governance. Their effectiveness
What can be done to improve corporate has been testified by a recent paper on the
governance in a country like Italy? The UK. Employing a methodology similar
finding that voting syndicates increase the to the one in this paper, Dahya et al. (2000)
sensitivity of turnover to performance and show that the Cadbury report in the United
increase the firm’s Q suggests that the Kingdom improved corporate governance
creation of a more competitive and active by increasing executive turnover and its
market for corporate control may be the sensitivity to performance, especially in the
answer. However, an effective market for firms that adopted the code. A similar
corporate control requires control to be con- phenomenon could take place in Italy and
testable, otherwise control can be transferred other countries, given the increasing atten-
only with the consent of the controlling share- tion to corporate governance throughout
holder or if the company defaults. Consistent the world and the growing pressure from
with this view, the results in this paper the international capital markets.
APPENDIX A
APPENDIX B
Size 0.043
Change in 0.135*** 0.002
EBIT/TA
Stock return 0.049* 0.102*** 0.163***
Owner-manager 0.104*** 0.041 0.019 0.015
dummy
High-incentive 0.006 0.202*** 0.001 0.068** 0.145***
dummy
Voting syndicate 0.021 0.073** 0.018 0.004 0.116*** 0.219***
dummy
Large minority 0.030 0.118*** 0.012 0.030 0.043 0.089*** 0.306***
dummy
* ** ***
, , and denote significance at the 10%, 5%, and 1% level, respectively. The number of observations
is 1,611.
Unpublished working paper, Cardiff Kaplan, S., Minton, B., 1994. Appointments of
University, Cardiff, UK. outsiders to Japanese boards: determinants
Denis, D., Denis, D., 1994. Majority owner- and implications for managers. Journal of
manager and organizational efficiency. Financial Economics 36, 225–258.
Journal of Corporate Finance 1, 91–118. La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Denis, D., Kruse,T., 2000. Managerial discipline Vishny, R., 1998. Law and finance. Journal of
and corporate restructuring following Political Economy 101, 678–709.
performance declines. Journal of Financial La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Economics 55, 391–424. 1999a. Corporate ownership around the
Denis, D., Denis, D., Sarin, A., 1997. Ownership world. Journal of Finance 54, 471–517.
structure and top executive turnover. Journal La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
of Financial Economics 45, 193–221. Vishny, R., 1999b. Investor protection
Franks, J., Mayer, C., 1996. Hostile takeovers in and corporate valuation. Unpublished
the UK and the correction of managerial working paper, Harvard University,
failure. Journal of Financial Economics 40, Cambridge, MA.
163–181. La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Franks, J., Mayer, C., 2001. Ownership and Vishny, R., 2000. Investor protection and
control of German corporations. Review of corporate governance. Journal of Financial
Financial Studies 14, 943–977. Economics 58, 3–27.
Franks, J., Mayer, C., Renneboog, L., 2001. Who Martin, J., McConnell, J., 1991. Corporate
disciplines management in poorly performing performance, corporate takeovers and
companies? Journal of Financial management turnover. Journal of Finance 46,
Intermediation 10, 209–248. 671–687.
Galgano, F., 1997. Diritto Commerciale. McConnell, J., Servaes, H., 1990. Additional
Zanichelli, Bologna, Italy. evidence on equity ownership and corporate
Gibson, M., 1999. Is corporate governance value. Journal of Financial Economics 26,
effective in emerging markets? Unpublished 595–612.
working paper, Federal Reserve Board, Morck, R., Shleifer, A., Vishny, R., 1988.
Washington, DC. Management ownership and market valua-
Gilson, S., 1989. Management turnover and tion. Journal of Financial Economics 20,
financial distress. Journal of Financial 293–315.
Economics 25, 241–262. Nicodano, G., 1998. Corporate groups, dual-
Gilson, S., Kose, J., Lang, L., 1990. Troubled class shares and the value of voting rights.
debt restructurings: an empirical study of Journal of Banking and Finance 22,
private reorganizations of firms in default. 1117–1137.
Journal of Financial Economics 27, Pagano, M., Roell, A., 1998.The choice of stock
315–353. ownership structure: agency costs, monitor-
Johnson, S., La Porta, R., Lopez-de-Silanes, F., ing, and the decision to go public. Quarterly
Shleifer, A., 2000. Tunneling. American Journal of Economics 113, 187–275.
Economic Review Papers and Proceedings Renneboog, L., 2000. Ownership, managerial
90, 22–27. control and the governance of companies
Kang, J., Shivdasani, A., 1995. Firm perform- listed on the Brussels stock exchange.
ance, corporate governance, and top executive Journal of Banking and Finance 24,
turnover in Japan. Journal of Financial 1959–1995.
Economics 38, 29–58. Warner, J., Watts, R., Wruck, K., 1988. Stock
Kaplan, S., 1994a. Top executive rewards and prices, event prediction, and event studies: an
firm performance: a comparison of Japan and examination of top management restructur-
the United States. Journal of Political ing. Journal of Financial Economics 20,
Economy 102, 510–546. 461–492.
Kaplan, S., 1994b. Top executives, turnover, and Weisbach, M., 1988. Outside directors and
firm performance in Germany. Journal of Law, CEO turnover. Journal of Financial
Economics and Organization 10, 142–159. Economics 20, 431–460.
282 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Wolfenzon, D., 1998. A theory of pyramidal Journal of Financial Economics 40,
ownership. Unpublished working paper, 185–211.
Harvard University, Cambridge, MA. Zingales, L., 1994.The value of the voting right:
Yermack, D., 1996. Higher market valuation of a study of the Milan stock exchange experi-
companies with a small board of directors. ence. Review of Financial Studies 7, 125–148.
Erratum
Paolo F. Volpin*
Erratum to ‘Governance with poor investor protection:
evidence from top executive turnover in Italy’
[Journal of Financial Economics
64 (2002) 61–90†]
Due to a mistake an error was made The corrected Table 10.5 is on the
in Table 10.5 on p. 269. Two minus- following page.
signs were accidentally put in the The Publisher apologizes for the
row of “Voting syndicate dummy”. inconvenience caused.
The dependent variable is the firm’s Q ratio. Size is the logarithm of total assists (in millions of
Liras). Owner-manager is a dummy variable that identifies the cases when at least half of the top
executives belong to the family of the controlling shareholder. Owner with high incentives is a
dummy variable that identifies the cases when the controlling shareholder owns more than 50% of
the cash-flow rights. Voting syndicate is a dummy variable that takes value 1 when the firm is
controlled by a voting syndicate. Large minority shareholders is a dummy variable that takes value
1 when the second largest shareholder owns a fraction larger than 5% of the firm’s voting rights.
Robust standard errors (in parentheses) control for correlation and heteroskedasticity. Year
dummies are included but the coefficients are not reported.
*, **, and ***, denote significance at the 10%, 5%, and 1% level, respectively.
Chapter 11
Luc Renneboog
OWNERSHIP, MANAGERIAL
CONTROL AND THE GOVERNANCE
OF COMPANIES LISTED ON THE
BRUSSELS STOCK EXCHANGE
Source: Journal of Banking & Finance, 24(12) (2000): 1959–1995.
ABSTRACT
This paper examines how corporate control is exerted in companies listed on the Brussels Stock
Exchange. There are several alternative corporate governance mechanisms which may play a
role in disciplining poorly performing management: blockholders (holding companies, industrial
companies, families and institutions), the market for partial control, debt policy, and board
composition. Even if there is redundancy of substitute forms of discipline, some mechanisms
may dominate. We find that top managerial turnover is strongly related to poor performance
measured by stock returns, accounting earnings in relation to industry peers and dividend cuts
and omissions.Tobit models reveal that there is little relation between ownership and managerial
replacement, although industrial companies resort to disciplinary actions when performance is
poor. When industrial companies increase their share stake or acquire a new stake in a poorly
performing company, there is evidence of an increase in executive board turnover, which
suggests a partial market for control. There is little relation between changes in ownership
concentration held by institutions and holding companies, and disciplining. Still, high leverage
and decreasing solvency and liquidity variables are also followed by increased disciplining, as
are a high proportion of non-executive directors and the separation of the functions of CEO
and chairman.
That different classes of owner have (Hypothesis 4). Hence, the relation between
different abilities to extract control rents is turnover and direct ownership (voting rights)
empirically supported for the US by by category of owner is expected to be less
Demsetz and Lehn (1985), Barclay and statistically significant than the one between
Holderness (1991) and Holderness and turnover and ownership concentration
Sheehan (1988). Holding companies are whereby the direct equity stakes (voting
prevalent in Belgium and their private rights) are reclassified based on the share-
benefits and reasons for control accumulation holder category of the ultimate owner.
are manifold: capturing tax reductions by
facilitating intercompany transfers, reducing
transaction costs by offering economies of 2.5. The disciplining role of the
scale or by supplying internal sources of market for share stakes
funds (Banerjee et al., 1997). Likewise, Burkart et al. (1997) argue that the degree
corporate shareholders may hold substantial of voting right concentration acts as a
share stakes in a target that may be a commitment device to delegate a certain
supplier or customer, in order to influence degree of authority from shareholders to
and/or capitalise on the target’s strategic management. They show that the use of
decisions. In contrast, there is little or no equity implements state-contingent control:
systematic evidence of monitoring actions in states of the world with decreasing
by institutions (investment funds, banks, corporate profitability, close monitoring
insurance companies . . . ). In Belgium, resulting from strong ownership concentra-
many institutions are affiliated with tion is desirable. In other states of the
financial institutions and are legally obliged world, it may not be optimal to have close
to avoid conflicts of interest (Renneboog, monitoring as this may reduce managerial
1997). No such impediments hinder moni- discretion and hence management’s effort
toring by holding companies, industrial and (also in Bolton et al., 1998). Hence, when
commercial companies, individual investors performance is poor, a partial corporate
or families. We therefore expect a positive control market may arise, consisting of
relation between turnover and ownership large (controlling) blocks. Furthermore,
concentration held by holding companies, poor performance may reflect not simply
industrial and commercial firms, individuals poor management but also ineffective
and families and no relation between monitoring and control. If this is the case,
turnover and institutional shareholder share poor performance may lead low quality
concentration (Hypothesis 3b). monitors to sell their stakes and new
(controlling) shareholders could improve
2.4. Ultimate ownership and dilution future corporate performance by substituting
of control incumbent management (Hypothesis 5).
Shleifer and Vishny (1986) show that once
Ownership structures are frequently a block of shares is assembled, the position
complex and pyramidal, and are con- is unlikely to be dissipated. It is in the large
structed for reasons of control leverage shareholder’s interest to wait until
(Wymeersch, 1994). Therefore, decisions someone who values control expresses
about disciplining management may not be interest in this block because if the block is
taken by direct investors but rather by the broken up and sold on the open market,
ultimate shareholders4 who control these part of the firm’s value arising from the
direct shareholders directly or through possibility of value-increasing monitoring
multiple tiers of ownership. Monitoring is is lost.
not performed by intermediate holding
companies which are investment vehicles of 2.6. Leverage as a bonding device
controlling industrial companies or individ-
uals and families, but by these industrial Creditor intervention may be expected
companies and families themselves when the probability of defaulting on debt
290 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
PERF performance variable measured Logit models are used if the dependent
by lagged (1) market adjusted returns, variable is a dummy (in the case of CEO
(2) changes in earnings after tax, (3) earn- turnover). For executive director and
ings losses, (4) ROE, (5) ROE – industry management committee turnover, GLS
median ROE (with earnings after tax), (6) models and OLS models with a logarithmic
ROA, (7) ROA – industry median ROA transformation of the dependent variable
(with earnings from operations before are used and the estimation is conducted
interest and taxes), (8) changes in with heteroscedasticity consistent covariance
dividends, (9) changes in ROE, (10) changes matrix estimator (White, 1980). Tobit
in ROE – industry median of ROE changes, models are also used to address that fact
(11) changes in cash flow on equity, that the dependent variable (executive and
(12) changes in cash flow on equity – industry committee turnover) is censored. Industry
median of changes, (13) changes in cash and time effects are accounted for by
flow margin, (14) changes in cash flow including industry and time dummies,
margin equity – industry median of changes. respectively. Corporate board size and firm
CONC ownership concentration (%) size are included as control variables.9 The
by class of owner: (i) holding companies, relations are also tested including corporate
(ii) banks, (iii) investment companies dummies and taking innovations to remove
(pension funds, investment funds), firm-specific effects. In order to address the
(iv) insurance companies, (v) industrial and endogeneity problems lagged data for
commercial companies, (vi) families and ownership, performance and debt policy
individual investors, (vii) federal or were utilised in the models. Over- or under-
regional authorities, (viii) realty investment performance in relation to industry peers was
companies. Both the percentages of owner- measured by correcting performance
ship by category of owner and the percentage variables for the median industry performance.
held by the largest shareholder are In Section 5, Tobit models are shown, but
included (in separate regressions). Both tables with other estimation methods are
direct shareholdings by category of owner available and the robustness of the results
are included as are the direct shareholdings across estimation techniques is discussed.
reclassified into the categories of owner
based on the category of the ultimate
(reference) shareholder (in separate 4. OWNERSHIP STRUCTURE AND
regressions). Herfindahl indices of the CONTROL OF BELGIAN LISTED
largest 3 shareholders by category of owner COMPANIES: STYLISED FACTS
are also used as concentration measures.
INCCONC purchases of share stakes 4.1. Ownership concentration
(in %) by category of owner. Both direct
shareholdings and reclassified ones based on In a nutshell, the characteristics of Belgian
ultimate shareholder are included, see CONC. corporate ownership can be summarised as
DEBT debt policy and debt structure follows: (i) few–only 165–Belgian compa-
variables: debt/equity ratio, current ratio, nies are listed, (ii) there is a high degree of
quick ratio, interest coverage (EBIT/interest ownership concentration, (iii) holding com-
expenses). In each model, gearing was only panies and families, and to a lesser extent
included along with one of the other industrial companies, are the main investor
variables in order to avoid multicollinearity. categories, (iv) control is levered by pyram-
BOARD board composition (% of idal and complex ownership structures and
non-executive directors), separation of the (v) there is a market for share stakes.
functions of CEO and chairman (1 no Properties (i) to (iv) imply that Belgium
separation), board size, tenure of CEO. can be portrayed as a Continental
European blockholder system rather than a
SIZE logarithm of total assets or of market based system (Bratton and
total employees. McCahery, 1999). However, typical for
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 293
All investors Holding co’s Families Indus. co’s Belgian investors Foreign investors
1994 MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP MIN MAJ SUP
a Percentage of the sample companies with a minority, majority or supermajority shareholdings held by the main shareholder categories. MIN % of companies with a stake
of 25% or larger, MAJ % of companies with a stake of 50% or larger, SUP % of companies with a stake of 75% or larger. Direct stands for the direct shareholdings.
Dir. and indirect refers to the fact that the direct shareholdings are classified according to the shareholder class of the ultimate investor: direct shareholdings belonging to the
same ultimate; investor group were subsequently summed. Ultimate control (direct and indirect) is control based on (i) a majority control (minimal 50% of the voting rights)
on every ownership tier of the ownership pyramid or (ii) shareholdings; of at least 25% on every tier in the absence of other shareholders holding stakes of 25% or more.
A chain of fully owned subsidiaries are considered as one single shareholder.
Source: Own calculations based on BDPart and Ownership Notifications.
294 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 295
Table 11.2 Largest direct and ultimate (direct and indirect) levered shareholdings, and the control
leverage factora
a This table presents the ultimate ownership level, defined as the highest level of ownership in an uninterrupted
control chain (direct shareholdings are level 1). Ultimate control is control based on (i) a majority control (min-
imal 50% of the voting rights) on every ownership tier of the ownership pyramid or (ii) shareholdings of at least
25% on every tier in the absence of other shareholders holding stakes of 25% or more. A chain of fully owned
subsidiaries are considered as one single shareholder.The direct largest shareholding is the average direct largest
share stake of at least 25%. The levered shareholding is calculated by multiplying the share stakes of subse-
quent ownership tiers. The control leverage factor is the ratio of the direct shareholding divided by the ultimate
levered shareholding. For instance, company A, whose shares are widely held, owns 40% of company B which,
in turn, owns 40% of company C. The ultimate shareholder level is 2, the direct largest shareholding (of B in C)
is 40%, the ultimate shareholding is 16% (40% 40%), and the leverage factor is 2.5 (40/16). There was no
direct shareholding of at least 25% in 17 sample companies, which were not included in this table. Standard
deviation in parentheses.
Source: Own calculations based on data from the BDPart database and the Notifications of Ownership.
Table 11.3 The market in share stakes over the period 1989–1994a
a This table gives the size distribution of purchases and sales of large shareholdings by category of owner over
the period 1989–1994. All changes are given excluding changes in government stakes and real estate as these
categories are minor. Purchases and sales are calculated by comparing the share stakes of a shareholder category
of a fiscal year to the shareholdings of previous year. Institutional investors consists of banks, investment and
pension funds and insurance companies. Total number of firm-years over the period is 1024.
Source: Own calculations based on BDPart and Ownership Notifications.
retirement age, death or illness of directors, from disciplinary actions.We also investigate
the turnover data may still contain some when such corporate governance actions
non-conflictual turnover since corpora- are undertaken and whether disciplining
tions do not generally release information takes place at an early stage, i.e. rapidly
regarding management replacement or do after earnings, cash flows or share price
so in euphemistic terms. declines or, rather late when the company is
no longer able to generate profits or has
to cut dividends? Including lagged
5.1. Board restructuring in industrial performance up to three years prior to
and commercial companies turnover allows us to investigate the reac-
tion time of board restructuring.14 Warner
5.1.1. Executive board turnover et al. (1988) and Coughlan and Schmidt
(1985) report that US boards react
quickly to poor performance in their deci-
5.1.1.1. Corporate performance and
sion to replace management because share
disciplining of management. performance lagged up to two calendar
A first question is whether or not turnover, years helps predict current-calendar-year
corrected for natural turnover, is related to management changes. Share price
poor corporate performance and results performance may underestimate the true
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 297
relation between performance and executive models yields weaker correlations with
turnover given that share prices reflect board restructuring.
current profitability as well as expected Levels of performance as well as changes
future opportunities including the potential in performance, corrected by industry
performance improvements under new medians, are analysed as it may well be
management (Weisbach, 1988). As that it is not just low earnings which trig-
accounting earnings depend on discre- ger managerial disciplining but peer group
tionary managerial accounting choices, we (industry) underperformance. Morck et al.
use a combination of accounting, dividend, (1989) find that when a firm significantly
cash flow measures and market adjusted underperforms its industry, the probability
share returns as performance benchmarks of complete turnover of the top manage-
in the Tobit models of Tables 11.4 and ment team rises. Table 11.4 shows that
11.5. Operating earnings before interest both industry adjusted levels and changes
and taxes (standardised by total assets) in ROE and in cash flow are negatively
are used as they are not sensitive to correlated to management changes prior to
financing policy, tax regime, windfall profits turnover, but more so for changes than
or extra-ordinary losses. The use of operat- for levels.
ing income rather than net earnings after All in all, the evidence of Table 11.4
tax reduces the impact of the described fails to reject Hypothesis 1: it shows that
‘earnings management’ (Dennis and the poorer the performance, the higher is
Dennis, 1994). ROE is taken after interest, the turnover of the executive board. These
extraordinary results and taxes. The indus- results are consistent through different
try medians are substracted from both the estimation techniques (Tobit and OLS
levels of and the changes in ROE and cash with and without fixed effects).
flow on equity.15 Companies only resort to substituting
Table 11.4 (lines 2–4) shows that, for executive directors when accounting
listed industrial and commercial compa- returns are very weak: when the company
nies, there is a negative significant relation was not able to generate profits or was
between executive director replacement forced to cut dividends in prior periods.
and market adjusted performance in the Furthermore, disciplinary actions are under-
three years prior to management substitu- taken when the company under-performs
tion. Earnings losses over the fiscal year its industry peers and when market
prior to turnover are followed by increased adjusted returns are negative in the period
levels of executive board turnover. Warner prior to board restructuring.
et al. (1988), amongst others, confirm for
the US that unless performance is
5.1.1.2. Ownership concentration.
extremely good or bad, their management
turnover models have little predictive As a single shareholder(group) controls a
value. Another critical performance bench- voting rights majority in more than half of
mark, substantial cuts in dividends (of at Belgian listed companies, and as a blocking
least 25%) or omissions, also precede minority exists in 85% of firms, the control
board restructuring. Given that deviations percentage of the largest block is included
from expectations about dividend policy as an explanatory variable.16 The free
usually contain signalling information, riding control-hypothesis predicts that
management is generally reluctant to large share blocks facilitate disciplining of
reduce dividends unless such a reduction is management. However, Table 11.4 (lines
unavoidable. Hence, dividend cuts are asso- 5–8) shows that that the presence of large
ciated with unusually poor stock-price and share blocks held by holding companies and
earnings performance (Healey and Palepu, institutions (banks, investment funds or
1988; Ofer and Siegel, 1987; Marsh and insurance companies) is not related to
Merton, 1987). Including changes in board restructuring. In contrast, manage-
earnings or dividends into the monitoring ment replacement is influenced by large
Table 11.4 Tobit model of the determinants of executive board restructuring in listed industrial and commercial companiesa
Market adj.return Operating earnings Dividend cuts ROE-indus. in ROE– in Cash flow on eq. – in CF/Eq– in
(%) losses (1 yes) (1 yes) median (%) industry median industry median (%) industry median
Performance =Par.Estim P(Chi) Par.Estim P(Chi)Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi)
1 Intercept 1.59057* 0.09 1.27277** 0.04 2.59715** 0.03 5.26084*** 0.00 2.24604** 0.02 2.35075** 0.01 4.40507*** 0.00
2 Perf. t1 0.78401*** 0.00 0.31887* 0.09 0.29769** 0.05 0.00470 0.23 0.05700* 0.09 0.01383 0.61 0.08407* 0.10
3 Perf. t2 0.35742* 0.08 0.09706* 0.08 0.00136 0.70
4 Perf. t3 0.56100* 0.07
Share stake held by the largest shareholder by category of owner at t1:
5 Hold. co’s 0.00346 0.60 0.01271*** 0.00 0.00457 0.51 0.01330 0.07 0.00180 0.75 0.00123 0.85 0.00668 0.11
6 Institutions 0.00549 0.44 0.01753*** 0.00 0.00246 0.72 0.00006 0.99 0.00590* 0.09 0.00143 0.85 0.00344* 0.10
7 Indus. co’s 0.02212* 0.06 0.00347 0.26 0.01100** 0.02 0.01164** 0.01 0.01087*** 0.00 0.00987** 0.04 0.02046*** 0.00
8 Fam/Ind. 0.02297* 0.08 0.01192* 0.09 0.02706 0.17 0.01456* 0.07 0.01491** 0.02 0.00087 0.90 0.01169** 0.03
Interaction between share stake held by the largest shareholder by category of owner at t1 and performance at t1:
9 Hold. co’s 0.05232 0.11 0.00001 0.77 0.00016 0.37 0.00015 0.68 0.00019 0.66 0.00007 0.85 0.00037 0.11
10 Institutions 0.00696 0.81 0.00016*** 0.00 0.00000 0.98 0.00097 0.16 0.00058 0.50 0.00011 0.64 0.00018 0.35
11 Indus. co’s 0.06148** 0.05 0.00002 0.32 0.00007 0.44 0.00042* 0.08 0.00037 0.21 0.00034 0.15 0.00035* 0.10
12 Fam/Ind. 0.11505** 0.03 0.00007 0.31 0.00066* 0.10 0.00061 0.36 0.00100* 0.06 0.00047 0.32 0.00043 0.43
298 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
a Perf. Performance, Hold. co’s Holding co’s, Indus. co’s Indus. and comm. co’s, Fam/Ind. Families and Individuals, D/E Debt/Equity, Intcov Interest coverage,
% non-ex. Percentage non-executive directors, Num. dir. Total number of directors. A dummy variable equal to 1 is included if the functions of CEO and chairman are
combined by one person.
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 299
Table 11.5 Tobit model of the determinants of executive board restructuring in listed holding companiesa
Market adj.return Operating earn. Dividend cuts ROEindus. in ROE– in Cash flow on eq. – in CF/Eq– in
(%) losses (1 yes) (1 yes) median (%) industry median industry median (%) industry median
Performance = Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi) Par.Estim P(Chi)
1 Intercept 2.37284* 0.09 3.68603 0.12 0.34073 0.97 1.36375 0.54 0.42521 0.83 1.51379 0.31 1.85787 0.31
2 Perf. t1 1.30870 0.38 0.30066* 0.09 0.74106*** 0.00 0.00229 0.48 0.30803** 0.01 0.10645 0.32 0.01220 0.20
3 Perf. t2 0.02472 0.75 0.14491* 0.10 0.20236* 0.02
4 Perf. t3 1.08528** 0.01
Share stake held by the largest shareholder by category of owner at t 1:
5 Hold. co’s 0.00067 0.93 0.01075 0.24 0.00906** 0.01 0.02882*** 0.00 0.01048 0.14 0.03187** 0.00 0.01475 0.12
6 Institutions 0.00973 0.41 0.01048 0.68 0.00248 0.82 0.02129* 0.10 0.03376 0.17 0.02671* 0.06 0.00557 0.59
7 Indus. co’s 0.02852** 0.01 0.12457 0.24 0.02773 0.50 0.06358** 0.01 0.01492 0.56 0.22205** 0.01 0.02355 0.52
8 Fam/Ind. 0.00109 0.89 0.01489 0.16 0.01479*** 0.00 0.03182*** 0.00 0.00506 0.67 0.03388** 0.01 0.01417 0.47
Interaction between share stake held by the largest shareholder by category of owner at t1 and performance at t1:
9 Hold. co’s 0.03150 0.20 0.00007 0.38 0.00094*** 0.00 0.00395*** 0.00 0.00059 0.38 0.00325*** 0.00 0.00007 0.94
10 Institutions 0.01317 0.19 0.00017 0.41 0.00048 0.45 0.00287* 0.05 0.00035 0.50 0.00329*** 0.00 0.00053 0.54
11 Indus. co’s 0.07345* 0.10 0.00160 0.24 0.00028 0.80 0.00570** 0.03 0.00761*** 0.00 0.01645 0.27 0.03967* 0.06
12 Fam/Ind. 0.03263* 0.07 0.00001 0.71 0.00024** 0.05 0.00464** 0.01 0.00071 0.33 0.00890** 0.00 0.00132 0.72
300 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
a Perf. Performance, Hold. co’s Holding co’s, Indus. co’s Indus. and comm. co’s, Fam/Ind. Families and Individuals, D/E Debt/Equity, Intcov Interest coverage,
% non-ex. Percentage non-executive directors, Num. dir. Total number of directors. A dummy variable equal to I is included if the functions of CEO and chairman are
combined by one person.
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 301
302 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
industrial investor shareholdings (in 6 out exception: when industrial companies and
of 7 models) and by blocks held by families families obtain substantial share stakes,
(5 models). Piecewise regressions – with changes in management are implemented.
dummies indicating whether or not the Such board restructuring takes place (lines
largest owner holds a blocking minority, 19– 20) when prior performance was poor
majority or supermajority (as in Hermalin (negative market adjusted returns, negative
and Weisbach, 1991) – reveal that minor- changes or levels of performance), which
ity stakes held by industrial companies are suggests a partial corporate control
sufficiently large to exert control and to market (Hypothesis 5). It can be observed
restructure the board.17 that disciplining underperforming manage-
Table 11.4 (lines 9–12) also investigates ment happens in the year of turnover or in
whether the ownership structure plays a the subsequent fiscal year.20
performance-induced disciplining role. Concerning the role of ownership
None of the categories of large blockholders concentration and the partial market
seem to be involved in disciplinary actions for control, it is important to realise that
against management when performance is above results were obtained after classifying
poor. The lack of institutional investor all blocks of voting rights into ownership
involvement is in line with Hypothesis 3 categories based on the identity of the ulti-
which states that they abstain from moni- mate or reference owners of each of these
toring to avoid conflicts of interest. In blocks. No significant results were attained
contrast, the fact that the large holding in a first set of regressions where all own-
companies do not seem to monitor is ership variables (levels and increases) were
surprising as these often cite superior included by category of shareholder owner
corporate governance as one of the core on the direct ownership level. For example,
contributions of their stable ownership if a holding company holds 10% of the
stakes as ‘reference shareholders’.18 The voting rights in a listed company, this 10%
lack of significance of the interaction stake is classified as a stake owned by a
terms between large industrial and family holding company.The fact that the interme-
owners, and performance, raises doubt diate holding company may be directly or
about the fact whether board restructuring indirectly controlled by e.g. a family is
is initiated by families or industrial compa- ignored in this first set of regressions.
nies as a result of poor performance.19 All However, when we reclassify all direct
in all, there is little evidence about the shareholdings (voting rights) in ownership
corporate control role of existing large categories based on the identity of the true
shareholders. (i.e. ultimate) owner – in the example
above the 10% stake is a family controlled
stake – we find the conclusions discussed
5.1.1.3. The market in share stakes above: significant results for the presence
When performance is poor, shareholders of industrial co’s and families versus
without a distinct interest in monitoring insignificant ones for institutions and holding
sell stakes, while those with strong moni- companies. This suggests support for
toring abilities increase their stakes in Hypothesis 4 and implies that the ultimate
order to reinforce their position as (major) or ‘reference’ shareholder, who controls the
shareholder. If this were true, we would voting rights of a listed target company by
expect positive signs for the increases in ways of a cascade of intermediate holdings,
shareholdings (Hypothesis 3). In spite of exerts corporate control.21
the fact that institutions and holding com-
panies actively trade in share stakes over 5.1.1.4. Gearing as a bonding mechanism
1989–1994 (Table 11.3), ownership
increases by these categories are not corre- High leverage encourages management to
lated with changes in board structure (lines generate sufficient funds to service the
13–16 of Table 11.4). However, there is one debt commitments. Consequently, a high
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 303
Although larger companies may have a companies and which internal and external
bigger internal managerial labour market control mechanisms trigger such gover-
and have better access to the external nance actions. Support for Hypothesis 1 is
managerial labour market, corporate size presented in Table 11.5: past poor share
is negatively related to executive board price and accounting performance in
replacement (line 30).25 Are these holding companies lead to executive board
disciplinary actions only directed at top restructuring, although this relation is
managers who are on the board or do these weaker than for listed industrial companies.
actions extend to other managers, c.q. the Over the period 1988–1994, dividend cuts,
members of the direction committee? earnings losses, low industry corrected
The results with management committee ROE and cash flows precede executive
turnover (executive directors and other top director replacement (lines 2–4). There is
managers) as dependent variable gives evidence of leverage related monitoring
weaker results than with executive (lines 18–21; Hypothesis 6) as high gearing
turnover. This implies that performance ratios result in increased managerial
related turnover is targeting the very top removal when performance is poor. In
management, namely predominantly those addition, board composition does also seem
three managers appointed to the board. to influence corporate control as high
executive board turnover is positively
related to a high percentage of non-executive
5.1.2. Disciplinary actions against CEO26 directors on board (Hypothesis 2).
Although some holding companies are
CEO replacement in industrial and commercial widely held, most of the others are con-
companies (corrected for natural turnover) trolled by families or other holding compa-
is correlated with the presence of large nies. Contrary to Hypothesis 3, there is no
shareholdings held by families and individu- consistent relation between the presence of
als, institutional investors and industrial large share blocks and executive board
companies. The fact that the interaction restructuring of listed holding companies.
terms of ownership and performance are The evidence in Table 11.5 even hints
significant (10% level) suggests that these (in 3 out of 7 models) that less monitoring
replacements are the result of corporate is expected in the presence of a controlling
governance actions (Hypotheses 3 and 4). holding company (negative correlation in
Furthermore, CEO substitution also follows line 5).28 Furthermore, the negative
increases in share stakes held by holding parameter coefficient in line 13 reveals
companies and industrial firms, but due to that when there is poor performance and
lack of significance in interaction terms it is increases in share stakes controlled by
questionable whether this stake accumula- holding companies, there is a reduced man-
tion is induced by poor performance agerial disciplining.This raises the question
(Hypothesis 5). Internal governance struc- why holding companies with controlling
ture like separation of the functions of CEO stakes or holding companies acquiring
and chairman, and the presence of a high large stakes do not monitor. Like in France
proportion of non-executives facilitates disci- (Banerjee et al., 1997, hence BLV), holding
plining of the CEO when the companies companies in Belgium may enhance wealth
results are bad (Hypothesis 2).27 Finally, creation by offering a internal capital
there is evidence that low interest coverage
market to the companies they control, by
leads to higher CEO turnover (Hypothesis 6).
offering the possibility to invest indirectly in
non-listed companies to individual investors,
5.2. Listed Belgian holding by smoothing tax liabilities of the group or
companies by lowering bankruptcy costs by facilitating
workouts. However, both the BLV study
Table 11.5 investigates whether executive (1997) for France and Daems (1998) for
director replacement takes place in holding Belgium fail to find wealth creation by
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 305
holding companies. The share price of expansion strategy and the call to form one
Belgian holding companies is even estimated large Belgian commercial bank (merger
to be 35–39% lower than the market value with the Generale Bank), GBL has not
of all their equity participations.29 favored such strategies.33 All in all, the
The potential corporate governance system with strong ownership concentration
benefits by holding companies consist of comprises important drawbacks (see also
the supply of strategic advice and a reduc- Daems, 1998; De Wulf et al., 1998).
tion in agency costs due to economies of
scale in monitoring. Still, BLV state that 5.3. Post-disciplining performance
“holding companies can create agency
problems of their own and, with well- The effectiveness of the corporate control
diversified portfolios, it is unclear why they mechanism could be assessed by analysing
would be willing or be able to engage in performance following the installation of
costly monitoring activities for each of the new management (Hypothesis 7).
companies they control.The quality of their Improved performance after executive
monitoring activities has never been board restructuring would confirm that the
ascertained, not has it ever been compared ousted directors and management had
to that provided by other large sharehold- underperformed and that the monitors
ers, the external market for corporate were able to attract a management better
control, . . . ”. The findings in this paper suited to reorganise the company.The post-
corroborate this statement as controlling disciplining earnings evolution is analysed
holding companies or holding companies by industry:34 for the companies in which
which acquire large share stakes in listed more than 25% of the executive board or
Belgian industrial companies (Table 11.4)30 the CEO resigned, the returns on equity two
and in listed Belgian holding companies years prior and two years subsequent
(Table 11.5)31 do not seem to be involved to executive board restructuring are
with performance correcting governance compared. Prior and subsequent ROE are
actions. Furthermore, listed Belgian hold- calculated as deviations from the ROE of
ing companies may suffer from the very companies which did not experience
agency problems they pretend to solve for executive board restructuring of 25% or
the companies they control because the CEO replacement. Even after the executive
controlling shareholders in holding compa- board restructuring, the level of ROE in
nies do not seem to discipline holding most industries subsequent to the board
companies’ underperformance. restructuring is still lagging the ROE of
This conclusion about holding companies companies without board restructuring.
can be extended to the quoted financial There is little evidence of ROE improve-
institutions – in this paper defined as a ments in time windows subsequent to board
rather heterogeneous sample of banks, restructuring with exception of the industry
insurance companies and real estate firms.32 of consumer goods/health/pharma. This
Most of these institutions are part of a finding may not come as a surprise as it is
holding group of which the monitoring a well documented fact that in US and UK
abilities have been questioned. Bank companies, a decrease in earnings often
Brussels Lambert (BBL) is often mentioned follows the departure of the CEO because
as an example of the break down of new CEOs often write off as many expenses
‘reference’ or ultimate shareholder as possible during their first year (Murphy
monitoring. Over whole time period of the and Zimmerman, 1993). Moreover, the
paper (1989–94), BBL is controlled by the success of corporate governance actions,
Group Brussels Lambert who holds a direct may not be visible in subsequent performance
stake of 13% (the largest stake) and controls because, in specific cases, earnings stabili-
indirectly via the Royal Belge insurance sation at a low level or avoidance of further
group another 9.3%. In spite of the need financial distress or even bankruptcy may
for the development of an international be considerable achievements.
306 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Dividend policy (changes in dividends per levered by pyramidal and complex ownership
share) are also compared for periods two structures and there is an important
years prior and two years subsequent to market for share stakes. Little relation was
board restructuring. Similarly, changes in uncovered between ownership structures
dividends are defined as the difference in and the disciplining of top management in
changes of high turnover companies from listed industrial and commercial compa-
the benchmark, namely dividend changes of nies. However, the presence of large indus-
companies with low board restructuring trial shareholders (and to a lesser extent of
(less than 25% of executive board turnover family shareholdings) is related to high
and no CEO turnover). For the sectors of executive board turnover when perform-
electrical equipment and electronics, and ance is poor, whereas no evidence was
services, there is a significantly higher found for a monitoring role by institutions
growth in dividends per share after board or holding companies.
restructuring than prior. For holding A fraction of the market for share stakes
companies, and the sector chemicals and may be considered as a corporate control
materials, this effect is only marginally market because industrial and commercial
significant.This result may indicate that, as companies increase their shareholdings (or
changes in dividends tend to have a perma- purchase large share blocks) in listed
nent character, increases in dividends poorly performing industrial companies in
reflect some more confidence in future which there is, subsequently, increased con-
profitability after managerial disciplining. flictual management replacement. In the
light of disciplining actions undertaken by
industrial companies (who are often in the
6. CONCLUSION same industry as the target company), the
lack of active corporate control in the wake
In this paper the importance of several of poor performance by holding companies
internal and external mechanisms of is striking. High ownership concentration
corporate governance was analysed in held by a holding company group can lead
terms of disciplining management of poorly to strategic deadlocks for quoted compa-
performing companies. There may be nies, as illustrated by the BBL case with
substitute forms of discipline, and even if the Group Brussels Lambert as reference
there is redundancy, one form may domi- shareholder. Daems (1998, p. 65) illus-
nate the other consistently. Disciplinary trates this deadlock with another example:
actions against management are taken ‘A reference shareholder [belonging to a
when market adjusted share returns are holding group] will tend to concentrate on
negative and when the company generates the interests of the group as a whole. He
operating earnings’ losses or resorts to sub- might be tempted to divide its markets over
stantial cuts in dividends in the years prior its subsidiaries such that they do not com-
to the restructuring. There is also evidence pete too intensively with each other. Hence,
that companies with levels of and changes [the French holding group] Suez, could
in ROE and cash flows below those of have an interest in dividing the interna-
industry peers are subjected to increased tional [utility] markets over its sub-
monitoring. These performance–turnover sidiaries, [the French] Lyonnaise des Eaux
relations are much stronger for listed and [the Belgian] Tractebel. This limitation
industrial and commercial companies than of strategic freedom of the subsidiaries is
for listed holding companies and financial not in the interest of minority shareholders
institutions. and investors who are participating in the
Belgian equity markets are characterised group via the stock exchange.’
by few listings, a high degree of ownership Although high leverage also seems
concentration held by holding companies instrumental to replace poorly performing
and families, and to a lesser extent indus- management, there is no direct evidence
trial companies. Furthermore, control is that it is bondholders or banks who
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 307
force underperforming management out. Theo Vermaelen, Michel Habib, Ian Cooper,
Interviews with non-executive directors Lorenzo Caprio, Marc Goergen, Rafel
disclosed that high leverage or low interest Crespi, Rez Kabir, Marco Becht, Piet
coverage stimulate actions by the (non- Moerland and Joe McCahery. Helpful
executive) board (and as such the share- advice was given by two anonymous
holders) rather than creditor intervention. referees, by the participants to the finance
We find that the role of the non-executive seminars at the universities of Tilburg,
directors is important in the disciplining Antwerp (UFSIA), Barcelona (Autonoma),
process: a high proportion of non-executive Venice, Oxford and by discussants at
directors leads to increased executive several conferences (EFA, EEA, Financial
board turnover. Given that companies usually Markets Symposium of the CEPR in
do not reveal the representation of the share- Gerzensee). I am grateful to Mr. Maertens
holders of the board of the directors, the of the Department of Information and
percentage of non-executive directors was used Statistics of the Brussels Stock Exchange
as a proxy for independence of the board for allowing access to some of the data
from management, but the board structure used in this study. Financial support was
may very well be influenced by ownership provided by the European Commission and
concentration. Furthermore, a higher proba- the Netherlands Organisation for Scientific
bility of CEO replacement was found when Research (NWO).
the tasks of CEO (‘delegated’ director) and
(non-executive) chairman are separated.
Corporate governance relations in holding NOTES
companies and especially financial
institutions are much weaker than in listed * Present address: Department of Finance
industrial companies. Board composition and CentER, Tilburg University, Warandelaan
and leverage have a substantial impact on 2, 5000 Tilburg, Netherlands. Tel.: 31-13-
board restructuring, but neither ownership 466-8210; fax: 31-13-466-2875. E-mail
concentration nor a partial market for address: Luc.Renneboog@kub.nl (L. Renneboog).
share stakes leads to increased disciplining 1 The recent changes in legislation on disclosure
of voting rights now allow detailed corporate
in holding companies. Belgian listed governance studies in Europe. Description of
holding companies may to a large extent ownership and voting rights in Europe can be
suffer from a lack of corporate control and found in Barca and Becht (2000, forthcoming.
seem to have discharged themselves from Who Controls Corporate Europe?, Oxford
efficient monitoring of the companies they University Press). The countries covered are
control, as seems also be the case for Austria (Gugler, Kalss, Stomper and Zechner),
French holding companies (Banerjee et al., Belgium (Becht, Chapelle and Renneboog),
1997). In spite of the presence of a large France (Bloch and Kremp), Germany (Becht
shareholder in a vast majority of Belgian and Bohmer), Italy (Bianchi, Bianco and
listed companies and of the breakdown – Enriques), Netherlands (De Jong, Kabir, Mara
and Roëll), Spain (Crespi and Garcia-Cestona),
both in terms of performance and Sweden (Agnblad, Berglof, Hogfeldt and
governance – of the ownership cascade Svancar), UK (Goergen and Renneboog,
system involving holding companies, it is 2000a,b), US (Becht).
problematic that the recent codes for good 2 Still, a priori, it is not certain whether one
corporate governance do not encompass specific corporate governance mechanism is
any recommendation with regard to large positively related to performance as, even if one
shareholder monitoring.35 mechanism may be used more frequently, the
existence of other corporate governance devices
and their interdependence may result in compa-
rable equilibrium performance (Agrawal and
ACKNOWLEDGEMENTS Knoeber, 1996).
3 Such recommendations have been formu-
I am indebted to Julian Franks for stimulat- lated in the US Bacon report (1993), the UK
ing discussions as well to Colin Mayer, Cadbury Committee report (1992), the French
308 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Viénot report (1995), the Dutch Peeters the kind: “pursuing other interests”, “spending
Commission report (1997), the Belgian corpo- more time with the family” or “retirements” at
rate governance guidelines by the Stock an age of 62 or below.
Exchange Commission, the Association of 9 Including board size controls for the fact
Employers and the Commission for Banking and that different governance mechanisms may
Finance (all in 1998). prevail in large versus small companies. Large
4 An investor is considered to be the ‘ultimate companies may have a larger internal managerial
or reference shareholder’ in an ownership–control labour market and have better access to an
chain if control is maintained through multiple external managerial labour market.
tiers of ownership. Interlocking ownership via a 10 We define a control relation between an
holding company or through a more elaborate ultimate shareholder and a target company if
stock pyramid enables a given investor to own (i) there is a series of uninterrupted majority
different quantities of voting and cash flow shareholdings on every ownership tier through-
rights. For instance, 50.1% of ownership (and out the pyramid or (ii) if there is a large share-
voting rights) held by the ultimate shareholder holding of at least 25% on every ownership
in an intermediary holding company which, in level in the absence of other shareholders with
turn, owns 50.1% of an operating subsidiary stakes of blocking minority size or larger.
could guarantee majority control on the 11 Often, Luxembourgian intermediate invest-
subsidiary’s board with only a 25.1% interest in ment companies are used.
its common stock cash flow. 12 These changes exclude shareholding
5 Following management changes, asset restructuring within investor groups, as these
write-offs (Strong and Meyer, 1987), changes to changes do not have any impact on control.
income reducing accounting methods (Moore, 13 We find a negative correlation (significant
1973) or income reducing accounting accruals at the 1% level) between past corporate
(Pourciau, 1993) frequently occur. performance and increases in ownership; the
6 The sample size was reduced by 9 companies lower the performance, the larger the increases
in 1989 and by 10 in 1994 as these listed firms, in ownership. Note that all increases, regardless
all in coal mining and steel production, were of their size, are taken into consideration
involved in a long liquidation process but were because some shareholders only need a small
still listed. increase in the percentage of their voting rights
7 The results do not change when we exclude to reach a blocking minority or a majority.
from the sample recent IPOs or companies that 14 If the fiscal year end is e.g. March 1994,
went bankrupt. Sector codes, dates of introduc- the data of this fiscal year are included in the
tion and of delisting are provided by the regressions as 1993 as most of the fiscal year is
Documentation and Statistics Department of in 1993. If the fiscal year end is 30 June 1994
the Brussels Stock Exchange. Companies dis- or later in 1994, the data of the year are
appearing as a separate entity following included in the regressions as 1994. The yearly
absorption by another company as a result of a market adjusted returns are calculated such
merger are included until the year prior to that they coincide with the fiscal years of the
the merger. corporations. Only lagged performance variables
8 Warner et al. (1988) and Weisbach (1988) are included because a performance variable of
also mention that reasons for turnover are often the year coinciding with the year of turnover
lacking. Weisbach also only excludes retire- may be a (partial) lead variable especially if the
ments if they are age related (63 years or older) turnover takes place early in the fiscal year.
which eliminates most of the non-linearity in the 15 Apart from the performance measures
turnover–age relationship: “ . . . companies do given in Tables 11.4 and 11.5, models with lev-
not announce the true reason behind their els and changes of return (after interest, taxes
CEOs’ resignations. Therefore, I ignore the and extraordinary) on assets (both with and
stated reasons for resignation in constructing without industry median correction) and cash
my sample. I do, however, eliminate the resigna- flow margin were estimated.The results of these
tions for which I am able to corroborate the models are in line with the ones discussed.
cause independently. Changes in CEOs caused by 16 Including the total share concentration by
death and preceding a takeover are excluded class of owner or Herfindahl indices, yields –
because these ‘resignations’ are totally verifi- expectedly – similar results. Including squared
able.” (p. 438). This bias is also mentioned by, ownership does not yield robust results across
among others, Dennis and Dennis (1995) and models.
Hermalin and Weisbach (1991). Non-informative 17 Piecewise regressions are not shown, but
reasons found for leaving the company are of tables are available.
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 309
18 In the years following the take over battle along with gearing and interest coverage in
between the French Suez group and the Italian order to avoid multicollinearity.
group of de Benedetti in 1989, the Generale 24 Several directors (among others, M. Davignon
Maatschappÿ van België or the Société (Generale Maatschappij van België/Société
Générale de Belgique, was restructured using a Générale de Belgique), M. Bodson (Tractebel),
focus strategy on 8 industrial and financial M. Samyn (NPM/SNP) ) were interviewed and
sectors. The Group Brussels Lambert, another asked to comment on the results of this study.
large holding company, has often been criticized They confirmed the limited role of creditor
for failing to establish a strategic plan for the monitoring unless a large refinancing takes
companies it controlled and is often given as an place and emphasized the role of the large
example of a stalemate situation brought about shareholders (for which we only found limited
by the reference shareholder model. The fact support) and of the board of directors (for
that some of these large holding companies, which we found strong support).
which control several listed (and many unlisted) 25 Using the logarithm of the total number of
companies, may fail in their monitoring role has employees as size variable, yields less significance
an important impact on our conclusions regarding (only 3 regressions out of 7 in Table 11.4).
the governance ability of holding companies. 26 Tables with logit models on CEO turnover
For a discussion, see Daems (1998) and Dewulf are available upon request.
et al. (1998). 27 CEOs with long tenure are less easily
19 The findings described are robust across removed when performance is poor as they may
estimation methods. OLS with fixed effects yield have a good track record. However, this conclu-
somewhat stronger significance for the presence sion is only based on one fourth of the sample
of large shareholdings held by industrial and companies as these data were only disclosed for
commercial companies and by families: in three a limited number of companies.
regressions, industrial and performance effects 28 The probability that the CEO of listed
are significant. holding companies is replaced increases with
20 Unlike the models with executive and CEO the degree of family control. This is especially
turnover, regressions with management the case when the market adjusted return is
committee turnover only yielded weakly signifi- low and when the company faces losses and
cant results. This implies that only those top has to reduce dividends. In contrast to the
managers who hold a board seat are held capital structure, board composition is again a
responsible and subject to disciplinary corpo- determining factor. Both a large percentage of
rate control. non-executive directors and the separation of
21 Although control is exerted by the ultimate the functions of CEO and non-executive
shareholder, there is evidence that when control- chairman increases the probability of CEO
ling stakes are held through multiple tiers of substitution.
ownership and when intermediate shareholdings 29 De Standaard of 7, 8 August 1999.
deviate from full ownership, ultimate investor 30 See lines 5, 9, 13, and 17.
control is diluted. For instance, sequences of 31 See lines 5, 9, 13, and 16.
majority control in the form of e.g. stakes of 32 Note that the reason why the proposed
50.1% throughout the pyramid might not guar- governance model does not fit the financial
antee the same degree of control a first tier companies is not due to lack of turnover. The
majority holding would give, unless there is dependent variable executive turnover is
strong board representation on each ownership between 22% and 37% depending on the year,
tier. Consequently, the larger the number of which is similar to turnover in industrial and
ownership tiers and the larger the deviation commercial companies.
from full ownership, the weaker the relation 33 A similar example of a reference share-
between turnover and ownership concentration. holder hampering a fair M&A competition
Including levered shareholdings (see Section 4) between Fortis and ABN Amro for control of
by category of owner gives similar results but the Generale Bank in 1997 has been detailed
reduces significance. extensively in the financial press.
22 Including changes in gearing did not give 34 Due to sample size limitations, several
any significant results. Changes in capital struc- sectors are added (based on NACE industry
ture in the form of new equity issues which took classifications) to form industry subsamples:
place in 40 firm-years was also included in the holdings companies, energy and utilities, chemical
models but are not correlated to board turnover. and materials (metal and non-ferro), electrical
23 Current and acid ratios were significant and electronics, consumer products and health
but were not included in the model of Table 11.4 care/pharma. services (transport, leasing,
310 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
hotel, . . . ). The utility sector was subsequently Burkart, M., Gromb, D., Panunzi, F., 1997.
deleted due to small sample size. Large shareholders, monitoring and the value
35 See also Meeus (1998) and Wymeersch of the firm. Quarterly Journal of Economics,
(1998) for a discussion of the content of the 693–728.
codes of good corporate governance. Cadbury, A., 1992. Report of the committee on
the financial aspects of corporate governance,
Gee & Co Ltd, London, December.
REFERENCES Coughlan, A.T., Schmidt, R.M., 1985. Executive
compensation, managerial turnover, and firm
Aghion, Bolton, 1992. An incomplete contracts performance: An empirical investigation.
approach to financial contracting. Review of Journal of Accounting and Economics 7,
Economic Studies 59, 473–494. 42–66.
Agrawal, A., Knoeber, C., 1996. Firm performance Daems, H., 1998. De paradox van het Belgisch
and mechanisms to control agency problems kapitalisme, Lannoo, Brussels.
between managers and shareholders. Journal Demsetz, H., 1983. The structure of ownership
of Financial and Quantitative Analysis 31, and the theory of the firm. Journal of Law
377–397. and Economics 26, 375–390.
Bacon, J., 1993. Corporate Boards and Demsetz, H., Lehn, K., 1985. The structure of
Corporate Governance, published by The corporate ownership: Causes and conse-
Conference Board in co-operation with Booz quences. Journal of Political Economy 93,
Allen & Hamilton Inc. 1155–1177.
Banerjee, S., Leleux, B., Vermaelen, T., 1997. Denis, D.J., Denis, D.K., 1993. Managerial
Large shareholdings and corporate control: discretion, managerial structure, and
An anlysis of stake purchases by French corporate performance: A study of leveraged
holding companies. European Financial recapitalizations. Journal of Accounting and
Management 3, 23–43. Economics 16, 209–236.
Barca, F., Becht, M., 2000. Who Controls Denis, D.J., Denis, D.K., 1995. Performance
Corporate Europe. Oxford University Press, changes following top management dis-
Oxford. missals. Journal of Finance 50, 1029–1057.
Barclay, M.J., Holderness, C.G., 1989. Private DeWulf, H. et al., 1998. Corporate Governance:
benefits from control of public corporations. Het Belgische Perspectief, Intersentia
Journal of Financial Economics 25, Rechtswetenschappen, Antwerpen-Groningen,
371–395. 245 p.
Barclay, M.J., Holderness, C.G., 1991. Elliott, J. and Shaw, W. 1988. Write-offs as
Negotiated block trades and corporate accounting procedures to manage percep-
control. Journal of Finance 46, 861–878. tions. Journal of Accounting Research
Baysinger, B.D., Butler, H.N., 1985. Corporate Supplement to vol. 26, 32–40.
governance and the board of directors: Fama, E., Jensen, M., 1983. Separation of own-
Performance effects of changes in board ership and control. Journal of Law and
composition. Journal of Law, Economics and Economics 26, 301–325.
Organization 1, 101–124. Franks, J., Mayer, C., 1998. Ownership and
Berkovitch, W., Israel, R., Spiegel, Y., 1997. control of German corporations. Working
Managerial compensation and capital paper, London Business School.
structure, Working paper. Franks, J., Mayer, C., Renneboog, L., 1998. Who
Bolton, P., von Thadden, E., 1998. Blocks, disciplines bad management? Discussion
liquidity and corporate control. Journal of paper, CentER, Tilburg University.
Finance 53(1), 1–25. Goergen, M., Renneboog, L., 2000a. Insider
Bratton, W., McCahery, J., 1999. Comparative control by large investor groups and managerial
corporate governance and the theory of the disciplining in listed Belgian companies.
firm: The case against global cross reference, Managerial Finance, forthcoming.
Working paper, Law faculty, Tilburg Goergen, M., Renneboog, L., 2000b. Prediction
University. of ownership and control concentration
OWNERSHIP, MANAGERIAL CONTROL AND GOVERNANCE 311
by the extent to which different interventions these results stand in marked contrast to
contribute to high board turnover in poorly what was described above for the United
performing companies. A governance Kingdom.
mechanism can be focused on poorly We believe that these differences are at
performing companies but have an insignif- least in part a consequence of regulation.
icant effect on overall board turnover in At first sight, this is surprising since the
these firms, i.e., have a high level of Type 1 United Kingdom and United States are
errors. A governance mechanism can be countries usually characterized as having
significant in dismissing a large number of similar “common law” regulatory systems
managers in poorly performing companies (see La Porta et al., 1998). Four examples
but be unfocused in also dismissing a large will illustrate significant differences in
number of other managers in well perform- regulation and how these lead to different
ing firms, i.e., have a high level of Type 2 governance outcomes in the two countries.
errors. First, the United Kingdom has a Takeover
The evidence reported in Franks and Code that makes accumulation of control-
Mayer (1996) illustrates this distinction. ling blocks expensive; the United States
They report that, on average, target firms does not. Second, the United Kingdom has
are not poorly performing companies. stronger minority protection laws making
Bidders do not therefore provide a focused the acquisition of partial controlling blocks
form of corporate control. Nevertheless, as well as takeovers expensive. For exam-
they could still perform a significant ple, in the United Kingdom a transaction
disciplinary function if they give rise to a between a large shareholder and its con-
high level of managerial replacement in nected company must be undertaken at
poorly performing firms. arms length and requires the consent of
The horse race evaluates both the focus noncontrolling shareholders; in the United
and significance of disciplining by different States, shareholder protection is limited to
parties.The results are quite striking.We find seeking redress in the courts for unfairly
that at least two parties—nonexecutive priced transactions. Third, in the United
directors and directors with large share Kingdom all seasoned equity issues, above
stakes—tend to entrench management by 5% of share capital, have to be in the form
reducing board turnover in poorly performing of rights issues and rights requirements can
firms. Neither existing holders nor new only be waived with considerable difficulty.
purchasers of large share blocks exert much These requirements provide dispersed
disciplining and, as noted above, bidders shareholders with significant control when
impose high board turnover after takeovers firms need to raise new equity financing. In
but in an unfocused way. It is only when there the United States, shareholders can and do
is financial distress, requiring equity issues waive rights issue requirements for almost
and capital restructuring, that disciplining is all seasoned offerings. Finally, in the United
both significant and focused on the manage- States there are significant fiduciary
ment of poorly performing firms. obligations on directors, breaches of which
Some of the results reported for the create high management turnover
United Kingdom are similar to those (Romano, 1991). There are few such
recorded for the United States, most obligations in the United Kingdom where
notably entrenchment by insiders. “actions to enforce the duties of directors
However, others are quite different. In the of quoted companies have been almost
United States, nonexecutives perform a nonexistent” (see Stapledon, 1996, pp.
disciplinary function (see Hermalin and 13–14). Where the role of nonexecutives is
Weisbach, 1991); active outside shareholders strengthened in the United Kingdom,
discipline management when share blocks through, for example, the adoption of the
change hands (see Bethel et al., 1998); and Cadbury Code (see Cadbury, 1992), it has
there is no reported role for new share been shown that the disciplinary function of
offerings in disciplining management. All boards increases (see Dahya et al., 2000).
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 315
In summary, we find that there are randomly selected from all companies
differences in regulation, even within two quoted on the London Stock Exchange in
common law countries, and that these 1988. We collected data on their perform-
differences are associated with signifi- ance and board turnover over the period
cantly different governance outcomes. The 1988 to 1993 and combined these with
results of this paper therefore enrich the information on ownership, sales of share
La Porta et al., (1998) view of regulation blocks, takeovers, board structure, and
as an important influence on the operation capital structure.
of capital markets. To be included in the sample, companies
Section 2 of the paper describes the were required to have data on boards and
methodology employed in this paper. ownership for at least 3 of the first 6 years
Section 3 examines the relation between of the sample period to allow panel data
board turnover and performance and the analyses to be performed. Companies
role of each of the five parties in disciplin- delisted through takeovers or insolvencies
ing poor management in the United for the first 3 years, between 1988 through
Kingdom. Section 4 reports regressions of to 1990, were therefore excluded. In
board turnover on performance: Section 4.1 addition, 7 of the original 250 companies
reports the results for the total sample and were dropped because of lack of perform-
Section 4.2 for poorly performing firms. ance data.
Section 5 contrasts the UK results in this Data on the composition of the board of
paper with those reported for the United directors were compiled for each year from
States and shows how regulation may 1988 to 1993 from annual reports,
account for these differences. Section 6 Datastream, the Financial Times, and
summarizes the paper. Nexus databases. They include the names,
tenure, and age of the CEO, Chairmen, and
all directors, both executive and nonexecutive.
2. METHODOLOGY We measured annual executive turnover
of the board from 1988 to 1993. Board
This paper is concerned with identifying turnover is calculated by dividing the total
who precipitates board restructuring in number of directors who leave the
poorly performing firms. The literature on company by total board size. Executive and
managerial disciplining points to five nonexecutive turnover is calculated in the
parties: (i) Shleifer and Vishny (1986) same way, except that the denominator is
show that large shareholdings mitigate free the number of executive and nonexecutive
rider problems of corporate control; directors, respectively. CEO and Chairman
(ii) Scharfstein (1988) models the way in turnover represents the proportion of
which takeovers perform a disciplinary sample companies where the CEO and
function; (iii) Burkart et al. (1997, 1998) Chairman, respectively, leave the company.
argue that trades in share blocks may be All turnover figures are corrected for
more cost effective than full takeovers; natural turnover. We distinguish between
(iv) Fama (1980) and Fama and Jensen natural and forced turnover, classifying a
(1983) describe how managerial labor resignation as “natural” if the director was
markets and nonexecutive directors assist described as having left the board for
in the governance of firms; and (v) Jensen reasons of retirement, death, or illness.
(1986) and Aghion and Bolton (1992) Otherwise the resignation was classified as
discuss the role that capital structure plays being forced. The normal retirement age is
in reducing agency costs. between 62 and 65 but some voluntary
This paper examines the role of all 5 retirement does occur before that; we took
forms of interventions in disciplining 62 as the minimum retirement age and
management. A sample of 250 companies, viewed any earlier retirement as forced.3
excluding financial institutions, real estate This reflects the difficulty of establishing
companies, and insurance companies, was whether public announcements of resignations
316 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
In addition we include interactive terms its effect on board turnover of the worst
between performance and the above five performing companies, in other words in
categories.The results reported below refer having a high level of Type-1 errors. A
to interactive terms with performance governance mechanism can be significant
lagged one year; regressions using interac- in dismissing a large number of managers
tive terms with contemporaneous perform- in poorly performing companies but
ance were also performed. unfocused in dismissing a large number of
We report executive board turnover other managers as well; i.e., it has a high
panel regressions estimated using a Tobit level of Type-2 errors.
regression to take account of fact that To rank the contribution of different
there are frequently no changes to boards. governance mechanisms to board turnover,
We also undertook OLS regressions with we report their economic as well as
logistic transformations of the dependent statistical significance. Economic significance
variable. Within (fixed effect) regressions is measured as the effect on board turnover
were also performed and time dummies for of moving from the mean to the extreme
individual years were used. Since high (upper or lowest) decile value of the
board turnover in one year might lead to relevant independent variable. The eco-
low turnover in a subsequent year, we inves- nomic significance of events, such as
tigated the robustness of the results using a takeovers and new equity issues, are meas-
cross-sectional OLS regression where the ured by their marginal impact on board
dependent variable is accumulated board turnover, i.e., the coefficient on the relevant
turnover for the year of poor performance dummy variable.
and two subsequent years. The results
reported below refer to the panel regres-
sions that include time as well as cross- 3. DATA ON BOARD TURNOVER
sectional effects. AND PERFORMANCE AND PARTIES
In Section 4.1 we report the results for
INITIATING INTERVENTIONS
the complete randomly selected sample of
firms. Although the potential size of our
sample is 1458 firm years (number of com- This section investigates the relation
panies 6 years), it is reduced to 1193 between board turnover, performance,
firm years as a result of takeovers (180), and the five different parties initiating
bankruptcies (10), and missing data (75), interventions.
and if the independent variable is lagged
then there is a further loss of 243 firm 3.1. Board turnover and
years. In Section 4.2 we report regressions performance
for a sample of poorly performing firms
drawn from the lowest decile of abnormal Table 12.1 provides a snapshot of the
returns in any one of the years 1988 to relation between board turnover and
1990. performance. Panel A of the table parti-
The results in Section 4.1 provide a tions the sample into deciles of perform-
measure of focus—To what extent are ance using abnormal returns. We choose
different interventions focused exclusively the sample of companies from 1990 and
on high executive board turnover in poorly aggregate data over three years from 1990
performing firms? The results in Section 4.2 to 1992. The period 1990 to 1992 is
provide an indication of significance—To chosen because 1990 is the first year when
what extent do different interventions con- the threshold for disclosing share stakes
tribute significantly to high board turnover held by outsiders was reduced from 5 to
in badly performing companies? A gover- 3%, and subsequent regressions demon-
nance mechanism can be focused, in the strate that performance has an impact on
sense of being well targeted on just poorly executive board turnover in the current and
performing companies but insignificant in following two years.
Table 12.1 Annual Board Turnover in 1990–1992 Partitioned by Decile of Abnormal Share Price Performance Calculated in 1990 for a Sample of 243
UK Quoted Companies
Panel A: Annual Board turnover over 1990–1992 partitioned by decile of abnormal share price performance in 1990 t-test on decile
difference
Decile of abnormal returns Worst 1 2 3 5 9 Best 10
Annual board turnover of: 45% 45% to 27% 27% to 19% 13% to 5% 12% to 23% 23% Average 1 vs 5 1 vs 10
Total board 15.5% 7.0% 7.7% 6.8% 6.0% 6.4% 7.3% 2.122 2.085
Executive directors 21.1% 10.0% 9.9% 8.1% 8.0% 6.9% 9.2% 2.300 2.388
Nonexecutive directors 7.4% 1.7% 4.9% 4.2% 3.6% 4.8% 4.2% 0.758 0.511
CEOs 28.8% 19.2% 11.4% 11.6% 14.5% 10.4% 13.6% 1.815 1.953
Chairmen 15.8% 6.9% 5.8% 7.2% 2.9% 5.9% 7.0% 0.862 1.007
Sample size 24 24 24 23 23 23 23
Panel B: Board turnover over 1990–1992 partitioned by dividends changes and by earnings, respectively during 1989–1990
t-test on t-test
Annual board Dividend cuts Constant or differences in Earnings Positive on differences
turnover of: and omissions increased dividends means losses earnings in means
Total board turnover 11.5% 6.2% 2.640 14.7% 6.4% 3.303
Executive directors 14.9% 7.0% 2.776 17.5% 8.0% 2.754
Nonexecutive directors 6.0% 5.8% 0.077 9.8% 3.7% 1.965
CEOs 25.9% 7.0% 2.795 26.0% 11.2% 1.846
Chairmen 15.9% 8.4% 1.464 19.1% 5.1% 2.302
Sample size (average) 36 181 18 190
Note. This table reports the turnover of the total board, the executive and nonexecutive directors, the CEO and chairmen for the three-year period 1990–1992. Board turnover
is calculated by dividing the total number of directors who leave the company (excluding those leaving as a result of retirement, death, or illness) by total board size. Executive
and nonexecutive director turnover is calculated in the same way, except that the denominator is the number of executive and nonexecutive members, respectively. CEO and
chairman turnover represents the proportion of sample companies where the CEO and the chairman, respectively, leave the company (corrected for natural turnover). Board
turnover data for the sample of 243 companies are categorized by performance: panel A categorizes companies by abnormal share price performance in 1990 for six deciles;
318 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
panel B categorizes companies by dividend cuts and omissions, and earnings losses in 1990.
Source: Annual reports, Datastream and London Share Price Database.
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 319
Table 12.1 records that there is a high (vi) executive directors and their immediate
level of board turnover in poorly performing family and trusts, and (vii) nonexecutive
companies. It also shows that the relation directors and their immediate family and
between board turnover and performance is trusts.8 We will refer to directors and their
highly nonlinear. Annual board turnover is families as “insiders” and financial
substantially higher in decile one than in institutions, industrial and commercial
any of the other deciles, for example, companies, and other major shareholders
15.5% compared with 6.8 and 6.4% for as “outsiders.”9 The sizes of share stakes
deciles five and ten, respectively. Executive held by government and real estate compa-
board turnover is much higher than nonex- nies were collected but not reported
ecutive turnover because nonexecutives because they are so small.
perform both a monitoring and advisory The distinction between different outside
function. CEO turnover is also much higher holdings is important because some may be
in decile 1 (at almost 28.8%) than in other passive in the face of poor management
deciles (for example, 11.6% in decile 5). performance while others are active. For
Turnover of chairmen is relatively high in example, institutional shareholders are
decile one, although the level is lower than often regarded as passive, and industrial
for CEOs, reflecting the fact that some companies and individuals/families as
chairmen are nonexecutive and perform a active. Corporate investors may have more
monitoring role. When the sample was knowledge about the industry than other
partitioned using abnormal returns accu- investors, and individuals and families may
mulated over two years, 1989–1990, the have more incentive to intervene as
relation between performance and turnover principals rather than agents.
was very similar to that in panel A.6 This section reports the pattern of
Panel B reports that companies with ownership for the sample companies. Panel A
losses and companies with dividend cuts or of Table 12.2 records the largest individual
omissions have more than twice the execu- share holding for all companies in each
tive board turnover of better performing year from 1988 to 1993 with the average
companies. Companies with dividend cuts for all years being 15.3%. The largest five
have 3.7 times the CEO turnover of those shareholders accounted for between 29.7
with increasing or stable dividends, and and 36.7% of the company’s shares
companies with earnings losses have 2.3 depending on the year. There is a large
times those without. We therefore find that increase in reported blocks from 31.4 to
there is a strong, nonlinear relation 41.0% between 1989 and 1990, which we
between board turnover and performance. attribute to the change in the disclosure
rule on block ownership, referred to earlier.
Panel B of Table 12.2 reports that the
3.2. Ownership concentrations median size of the largest stake lies in the
We collected data on the size of shareholdings range 5–15% for all individual years, but
over the period 1988–1993. All directors’ there are a significant number of blocking
holdings greater than 0.1% are included as minorities, defined as a stake of at least
well as outside share stakes greater than 25%. For example, in 1988 almost 24% of
5% until 1989. From 1990, the statutory stakes are in excess of 25%.10 Panel C dis-
disclosure threshold for outside sharehold- aggregates large shareholders by their type
ers was reduced to 3%.7 and size of holding in 1991. Institutional
Shareholdings were classified according investors hold the highest proportion
to 7 categories: (i) banks, (ii) insurance (52.6%). Insiders, directors, and their
companies, (iii) institutional shareholders families are the next most significant hold-
including investment trusts, unit trusts, and ers. The difference between outsider and
pension funds, (iv) industrial and commer- insider blocks is important because the
cial companies, (v) families and individuals, latter may entrench poor management.
not directly related to any director, Although not shown in the table, insider
Table 12.2 Size and Category of the Largest Shareholder in the Sample of UK Quoted Companies, 1988 to 1993
Panel A: Size of largest shareholding, sum of ownership of largest 5 share stakes and all reported large shareholdings respectivelya
Year Largest shareholding Largest 5 shareholdingsb All reported large shareholdingsb
Note. Panel A shows the average largest shareholding, the sum of the largest 5 shareholdings, and the sum of all reported large shareholdings over the period 1988 to 1993.
Panel B shows the percentage of companies with a large shareholding in five size ranges over the period 1988 to 1993. Panel C reports the size distribution of the largest
shareholding by type of shareholder in 240 companies in 1991. All holdings by directors greater than 0.1% are included; shareholdings by outsiders in excess of 5% are
reported in 1988 and 1989 and in excess of 3% thereafter. Source: Annual reports.
320 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
a Sample sizes were 239, 241, 241, 240, 207, and 205 in each of the six years 1988 to 1993, respectively.
b The increase in 1990 reflects the reduction in the disclosure threshold from 5 to 3% in 1989.
c The sum is less than 100% because small shareholdings held by the government and real estate companies are not included.
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 321
Note. The table reports average ownership concentration over the period 1990 to 1992 for firms with below median and above median market capitalization in 1990.
Ownership concentration is reported for the largest shareholding and the sum of all disclosed shares held by investment institutions, industrial companies, families and
individuals, and insiders. Board turnover is calculated by dividing the total number of directors who leave the company (excluding those leaving as a result of retirement,
death, or illness), by total board size. Executive and nonexecutive director turnover is calculated in the same way, except that the denominator is the number of executive and
nonexecutive members, respectively. CEO and chairman turnover represents the proportion of sample companies where the CEO and the chairman, respectively, leave the com-
pany (corrected for natural turnover).
Source: Annual reports, London Share Price Database.
322 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 323
post takeover they are 88 and 94%, In Panel A of Table 12.4 we show the
respectively.15 Moreover, there is little total number of purchases of share blocks
difference in turnover between poorly per- by new shareholders in excess of 5% for
forming companies and better performing the three-year period, 1991 to 1993.16 We
companies, suggesting that takeovers pro- choose this subperiod because the disclo-
vide relatively unfocused disciplining. For sure threshold changed to 3% in 1990.
example, if we define poorly performing There are a total of 303 purchases of
companies as those in the bottom decile of stakes greater than 5% and 82 greater
abnormal share price performance in one than 10%; the latter represents an annual
of the two years prior to takeover, total rate of 9% per year compared with 6.7%
board turnover is identical at 88% post for the United States cited in Bethel et al.17
takeover in both samples of poorly Almost one-half of block sales greater than
performing and better-performing firms. 10% are made by companies, families, and
In the United States, Bethel et al. insiders, suggesting greater scope for more
(1998) examine the relation between block active investing. In Panel B of the table we
purchases of 5% or more and firm per- examine changes in the level of concentra-
formance.They find that activist blockhold- tion by both existing and new shareholders.
ers acquired stakes in highly diversified We accumulate share blocks by adding
firms with poor profitability. They also find together individual purchases in each com-
that the target firm’s profitability pany greater than 5% in any one year. The
increased after the block purchase, and as panel reports that in 89 companies more
a result, they conclude that this market than 10% of the equity was purchased by
works as a market for corporate control. existing and new shareholders over the
We investigate the size of this market in 3-year period. Panel B also shows that a
share blocks in the United Kingdom and the single investor or a coalition of blockholders
extent to which it is motivated by poor per- purchases an equity stake of 25% or more
formance leading to the disciplining of in 19 companies.The acquisition of blocking
management. We also distinguish between minority stakes takes place on average in
active and passive blockholders. 3.1% of listed firms per year, similar to the
Table 12.4 Purchases of Share Blocks for 243 Companies for the Period 1991 to 1993
Panel A: Number of share blocks purchased by new shareholders exceeding 5% for the period
1991 to 1993
[5–10%] [10–25%] [25–50%] 50% Total
Institutions 168 38 5 0 211
Companies 24 15 5 0 44
Families 19 6 2 0 27
Directors 10 11 0 0 21
Total 221 70 12 0 303
Panel B: Number of companies with increases in concentration greater than 5% over the period
1991 to 1993
Year [5–10%] [10–25%] [25–50%] 50% Total
1991 37 36 4 2 79
1992 16 25 5 2 48
1993 14 9 4 2 29
Total 67 70 13 6 156
Note. Panel A reports purchases of share blocks of between 5 and 10%, 10 and 25%, 25 and 50%, and more
than 50% over the period 1991 to 1993 by type of purchaser. Panel B reports the number of companies with
increases in concentration of more than 5% by new and existing shareholders over the period 1991 to 1993.
Source: Annual reports.
324 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
annual rate of takeover activity in the in both sets of firms over the period, reflecting
United Kingdom of 3–4%, where control the increasing emphasis on nonexecutives
passes when a majority of shares are in corporate governance in the United
acquired by the bidder, usually via a tender Kingdom.18 More strikingly, in those
offer. companies where there is a change in CEO,
Although not reported in the table, we there is a significant reduction in the
examined the relation between purchases of proportion of companies with combined
share blocks and performance, measured roles of CEO and chairman from 42.9%,
over two years prior to, and the year of, the before 1990, to 14.3% after. Subsequent
purchase. The number of share blocks is regressions suggest that separation of CEO
virtually identical in the worst and best and chairman plays an important role in
deciles of performance and similar to the disciplining of management of poorly
average for the complete sample. performing firms.
In summary, there is a high level of both In summary, there is little relation
takeovers and trades in share blocks in the between the proportion of nonexecutives on
United Kingdom. There is some evidence of the boards of firms and corporate perform-
a relation between the incidence of ance suggesting little disciplinary role asso-
takeovers and poor performance but not ciated with nonexecutive directors.
between trades in blocks and performance. However, there is evidence that the role of
This suggests that takeovers may be per- CEO and chairman is more frequently
forming a disciplinary function but trades separated after changes in CEOs.We examine
in blocks do not. We investigate this further these observations further in the regressions
in Section 4. in Section 4.
Total number of directors 8.7 8.7 8.4 9.6 10.9 8.4 9.3
Proportion of nonexec. dir. (%) 37.5% 37.3% 38.2% 42.4% 43.5% 40.7% 39.8%
Combined CEO/Chairman 27.1% 43.1% 31.0% 42.0% 34.8% 28.4% 32.9%
CEO age (years) 49.7 53.2 51.5 55.9 54.4 54.5 52.6
CEO tenure (years) 4.1 5.1 5.1 8.2 5.2 5.7 5.3
Chairman age (years) 55.0 60.3 59.8 59.6 60.3 58.5 59.0
Chairman tenure (years) 5.5 5.1 5.9 6.5 7.1 6.1 6.1
Panel B: Board characteristics of companies with and without CEO turnover in 1990 two years prior and two years post 1990
Companies Sample Two years prior Two years after t-test on difference in
with CEO turnover size (Average 1988–1989) (Average 1991–1992) means
Proportion nonexec. dir. (%) 28 36.4% 41.5% 1.464
Combined CEO/Chairman 28 42.9% 14.3% 2.449
CEO age (years) 19 51.9 52.6 0.329
Chairman age (years) 19 59.4 59.4 0.032
Companies Sample Two years prior Two years after t-test on difference in
without CEO turnover size (Average 1988–1989) (Average 1991–1992) means
Proportion nonexec. dir. (%) 180 36.2% 40.8% 2.483
Combined CEO/Chairman 180 42.2% 38.9% 0.643
CEO age (years) 110 51.6 53.1 1.801
Chairman age (years) 121 57.8 59.1 1.419
t–statistics of differences between companies with and without CEO turnover in 1990
Proportion nonexecutive directors 0.068 0.305
Combined CEO/Chairman 0.062 3.213
CEO age (years) 0.130 0.332
Chairman age (years) 1.144 0.221
Note. Panel A shows board size, the proportion of nonexecutives on the board, the proportion of companies with combined CEO and Chairman, the age and tenure of the CEO
and Chairman averaged annually over the period 1990 to 1992 for 6 deciles of abnormal share price performance in 1988. Panel B reports the proportion of nonexecutives,
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 325
combined CEO and chairman, and age of CEO and chairman averaged over two years either side of 1990 for companies in which there was a change in CEO in 1990 and for
companies in which there was no change in CEO in 1990.
326 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
in decile 5: a median of 1.8 compared with issue and another five of a capital
4.0 in the worst performing decile.19 reconstruction or public recontracting of
Table 12.6 shows that high leverage, existing debt. In one case the bank stated
combined with poor performance, is related at an Extraordinary General Meeting
to increased executive board turnover. that a renewal of loan facilities was
Companies in the lowest decile of share conditional on a resolution to approve
price performance and the lowest quartile the sale of assets. Since much of UK debt
of interest coverage had significantly higher is in the form of private bank debt, the
executive board and CEO turnover in the actual level of bank restructuring is
year of poor performance and the two years much greater than that which is publicly
subsequently than those in the highest quar- revealed.
tile of interest coverage, 69.6 and 24.2%, It is clear from the descriptions in The
respectively. Companies in the lowest decile Financial Times (FT) that the party initiat-
of share price performance and the highest ing the boardroom changes is not necessar-
quartile of capital leverage had higher exec- ily creditors. For example, in the departure
utive board and CEO turnover than those in of the CEO of Burton the FT reported
the lowest quartile of capital leverage “that he had not performed with sufficient
(though only the latter was significant). vigor to impress [the board’s] non-executive
The annual rate of new equity issues in directors” (November 30, 1990). In
the sample of poorly performing firms used another company, Cookson’s, the FT stated
in the subsequent regressions at 11.5% is that “the CEO/Chairman resigned after it
almost identical to that of the total sample became clear that he had lost the confi-
at 11.6%. This suggests an important role dence of the company’s own senior
for new equity issues in distressed compa- executives.” (November 30, 1990). For
nies. We investigated this further by under- Platon, “a series of boardroom changes
taking press searches on 34 firms that were was foreshadowed when the company
both below average performers and had detailed plans for a sterling open [equity]
high levels of debt.The criteria for selection offer.” The chairman of Era resigned a
were that firms both had interest coverage week after “a long and angry shareholders’
less than two and were in the bottom three meeting.” Finally, in the case of Caledonian
deciles of performance in at least one year Newspapers, a large shareholder when
during the period 1990–1993.20 In 28 approached about subscribing for new
firms the CEO or chairman resigned, or equity responded that “they would put
both resigned. Eighteen firms or about more money up, but if so, it was good-bye
54% of the sample raised new equity management.”
finance. Of these, 15 were rights issues or An important question is why large insti-
open offers, while the remaining three were tutional shareholders are active only when
offered to new shareholders in the form of poorly performing firms make distressed
placings.21 In three cases the offer took the rights issues. Senior management at the
form of convertible preference shares, largest fund managers in the United
otherwise it was for straight equity. Kingdom informed us that although they
There was a substantial number of other might intervene where there was very poor
ownership changes. In 24 companies, or performance, in the face of management
72% of the sample, there was at least one opposition, they were likely to avoid con-
of the following: a new issue, a takeover, or frontation because they disliked the conse-
the emergence of a large shareholder. In quent publicity and the costs of organizing
some cases board changes coincided with other shareholders. However, it was a dif-
one of these events, but in many cases cap- ferent story when the poorly performing
ital or ownership changes preceded board company required new financing: “it comes
changes by a matter of several months. to a crunch when companies raise addi-
Debt restructuring is also important. tional finance” or “it all unpicks when a
There were five cases of a public debt company needs money.”
Table 12.6 The Relation between Board Turnover and Leverage for Companies in the Lowest Decile of Performance
Cumalative executive Cumalative CEO Cumalative executive Cumalative CEO
turnover: Interest cover turnover: Interest cover turnover: Capital leverage turnover: Capital leverage
Leverage Sample Mean Stdev. Sample Mean Stdev. Sample Mean Stdev. Sample Mean Stdev.
Performance
4. REGRESSION RESULTS FOR
Lines 3 to 5 of Table 12.7 show a strong
BOARD TURNOVER ON GOVERNANCE
negative relation between board turnover
AND PERFORMANCE for four out of five measures of perform-
ance either concurrently or with lags (the
This section reports the results of regres- exception being industry corrected return
sions of executive board turnover on per- on equity). The economic effect of abnor-
formance and the five sets of governance mal returns in the lowest decile is to raise
variables described above over the period board turnover by 7.0% two years later.
1988 to 1993. Section 4.1 discusses the The economic effect of earnings losses is
results for the total sample and Section 4.2 much larger in raising board turnover by
for the worst performing companies. 24.6% over a three-year period. This sug-
gests that board turnover is more sensitive
4.1. The total random sample to earnings losses than contemporaneous
abnormal share price returns.24 It may be
Table 12.7 records the results of a Tobit that management and shareholders regard
panel regression on executive board earnings losses as a more serious sign of
turnover; although not formally reported in managerial failure than abnormal returns.
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 329
We report below the regression results This is consistently observed across different
describing the influence of existing share- estimation techniques. Purchases of share
holders, changes in shareholdings, stakes are for the most part new holdings
takeovers, leverage, and board structure on rather than increases in existing holdings.
board turnover. The economic effect of increases in
holdings by families and individuals (on
average 0.7%) is to increase board
Existing shareholders turnover by between 5.4 and 7.7%,
depending on the performance measure
If concentration of ownership overcomes a used. The economic effect of increase in
free rider problem of corporate control, we holdings by executives (on average 0.7%)
would expect there to be higher board is between 10.5 and 12.5% increases in
turnover in poorly performing firms where board turnover. However, the interaction
concentration of ownership is high. terms do not support the view that changes
The signs of the coefficients on the interactive in family and executive holdings are
terms between ownership and performance performing a disciplinary function (lines 23
should therefore be negative in lines 11 to and 24). We do not observe a relation of
15. In fact, there are few significant terms executive board turnover to industrial
in lines 6–15, indicating that ownership companies shareholdings either on their
concentration on its own or with interaction own (line 17) or interactively with per-
terms does not play a significant role in disci- formance (line 22). There is therefore little
plining management. An exception is holdings difference in impact of active and passive
of executive directors, which are consistently shareholders on corporate governance.25
negatively related to board turnover for four
performance measures. The economic effect
of executive ownership (which has a mean Takeovers
value of 7.6%) is to lower board turnover by
between 12.2 and 15.1%, depending on the We included a dummy variable in the
performance measure, but it is generally regression for whether the firm was taken
unrelated to performance (line 14 in Table over and an interactive dummy variable
12.7). This suggests that large ownership with performance. The takeover variable
gives executives modest protection against was significant in every regression but the
outside intervention, irrespective of company interactive term with performance was sig-
performance. This is robust to different nificant in none. The economic effect of
estimation techniques. takeovers on board turnover is in the range
89.5 to 113.2% depending on the measure
of performance used. This confirms the
Increases in shareholdings result in Franks and Mayer (1996) that
there is a very high level of board turnover
Significant changes in shareholdings are associated with takeovers but it is unre-
likely to give rise to changes in management lated to poor performance. In Section 3.3,
irrespective of corporate performance. we observed that over an extended period
However, if these changes are performing a of time, the incidence of takeovers is higher
disciplinary function then there should be in poorly performing than other firms. This
higher board turnover with poor perform- suggests that the disciplinary effect of
ance and with changes in shareholdings. We takeovers may be delayed beyond the one-
would therefore expect to observe positive or two-year lags in these regressions.
coefficients on lines 16 to 20 and negative
coefficients in lines 21 to 25.
Table 12.7 shows a strong positive rela- Board structure
tion between increases in share holdings by
both families and executive directors, and If nonexecutive directors perform a corpo-
executive board turnover (lines 18 and 19). rate governance function then we would
Table 12.7 Regressions of Executive Board Turnover on Governance and Performance for the Total Sample, 1988–1993
Independent variable
1. Sample size 905 905 938 948 948
Noncensored 337 337 349 352 352
Left censored 568 568 589 596 596
2. Intercept 6.4890 0.16 4.1165 0.36 5.3199 0.23 1.0762 0.81 4.6022 0.33
Performance
3. Performance T-2 0.1041 0.00 0.1057 0.00 10.2785 0.04 0.0085 0.53 10.4561 0.00
4. Performance T-1 0.0928 0.45 0.0536 0.68 28.3968 0.16 0.0761 0.31 5.8946 0.59
5. Performance T 0.0367 0.18 0.0278 0.34 14.3359 0.00 0.0124 0.20
Existing shareholders
6. institutions 0.0019 0.98 0.0038 0.97 0.0206 0.82 0.0141 0.87 0.0532 0.59
7. industrial cos 0.0713 0.51 0.1010 0.33 0.0801 0.45 0.1179 0.27 0.1395 0.21
8. families and indiv. 0.2040 0.25 0.2512 0.16 0.3520 0.06 0.4198 0.04 0.2151 0.26
9. executive directors 0.3576 0.00 0.3397 0.00 0.3433 0.00 0.3671 0.00 0.4200 0.00
10. non-executive dir. 0.1022 0.51 0.0655 0.66 0.0321 0.82 0.0475 0.74 0.0852 0.57
Interaction terms : cum ownership per category* performance T-1
11. institutions 0.0027 0.37 0.0043 0.19 0.8986 0.03 0.0006 0.80 0.0470 0.82
330 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
12. industrial cos 0.0016 0.60 0.0015 0.67 0.2851 0.42 0.0026 0.40 0.1131 0.68
13. families and indiv. 0.0004 0.94 0.0010 0.85 0.1298 0.84 0.0191 0.02 0.6004 0.21
14. executive directors 0.0018 0.55 0.0003 0.92 0.9214 0.02 0.0007 0.75 0.0603 0.79
15. nonexecutive dir. 0.0007 0.88 0.0006 0.89 0.2432 0.65 0.0017 0.64 0.6443 0.06
Increases in shareholding
16. institutions 0.1607 0.28 0.1774 0.22 0.2059 0.16 0.1029 0.48 0.2616 0.09
17. industrial cos 0.1111 0.71 0.0650 0.82 0.0535 0.85 0.3865 0.14 0.0041 0.99
18. families and indiv. 0.9933 0.01 1.1462 0.00 1.2302 0.00 1.4283 0.00 1.1242 0.00
19. executive directors 1.3076 0.00 1.1639 0.00 1.1922 0.00 1.2396 0.00 1.3886 0.00
20. nonexecutive dir. 0.5908 0.28 0.6350 0.21 0.6241 0.13 0.4658 0.28 0.5673 0.21
Interaction terms : increases in shareholdings per category * performance T-1
21. institutions 0.0042 0.29 0.0051 0.23 0.4138 0.44 0.0016 0.67 0.6734 0.07
22. industrial cos 0.0072 0.35 0.0126 0.15 1.4877 0.05 0.0048 0.49 0.9204 0.13
23. families and indiv. 0.0105 0.29 0.0086 0.45 0.7496 0.66 0.0220 0.14 0.9359 0.29
24. executive directors 0.0171 0.14 0.0064 0.59 0.9735 0.33 0.0031 0.84 0.6782 0.41
25. non-executive dir 0.0113 0.56 0.0090 0.64 1.0021 0.66 0.0030 0.69 0.5732 0.58
Board structure, leverage and takeovers
26.% nonexec dir. 0.3185 0.00 0.3022 0.00 0.3263 0.00 0.3006 0.00 0.3855 0.00
27. Separ. CEO-chair 1.9722 0.41 1.7487 0.45 0.4776 0.84 1.5695 0.50 0.8427 0.73
28. Cap. Lev. (book) 0.0481 0.37 0.0440 0.40 0.0567 0.25 0.1656 0.00 0.0977 0.08
29. New equity issue 9.4970 0.00 9.9686 0.00 4.9607 0.12 6.6196 0.04 7.1537 0.03
30. Merger/Acquis 113.2149 0.00 96.2547 0.00 98.4948 0.00 89.4849 0.00 96.7921 0.00
Interaction terms: board structure, leverage, takeovers, and performance at T-1:
31.% nonexec dir. 0.3300 0.16 0.2614 0.29 2.8602 0.93 0.0152 0.94 0.3294 0.09
32. Separ. CEO-chair 0.0024 0.97 0.0420 0.59 9.9548 0.35 0.0285 0.62 0.1324 0.98
33. Leverage 0.0026 0.05 0.0026 0.06 0.0574 0.50 0.0011 0.11 0.0151 0.85
34. Equity issues 0.0299 0.75 0.0320 0.75 36.2614 0.00 0.0890 0.19 2.9039 0.77
(Table 12.7 continued)
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 331
Table 12.7 Continued
35. Merger/Acquis 0.9832 0.17 0.6005 0.17 6.8505 0.79 0.1227 0.53 27.4024 0.16
36. Disclosure dum 6.8984 0.02 3.6863 0.17 5.7543 0.03 4.5981 0.09 5.1204 0.05
37.Total sales (log) 1.56E-07 5.79E-01 1.42E-07 6.14E-01 4.47E-08 8.72E-01 6.69E-08 8.20E-01 5.76E-09 9.84E-01
p-value of F-test 0.001 0.0001 0.001 0.0001 0.001
Rsq adj 0.1957 0.1962 0.2119 0.1512 0.1913
Note. The table records a Tobit regression of annual board turnover on performance, ownership, increases in shareholdings, board composition, and takeovers for the period
1988 to 1993. The measures of performance include annual abnormal returns, both with and without a correction for industry abnormal returns, earnings losses (dummy,
loss –1), return on equity corrected for average industry performance, dividend decreases and omissions (dummy, dividend cuts or omissions –1). Nine blocks of
independent variables are reported.The first relates to performance with up to two-year lags.The second relates to concentration of ownership by five classes of investors.The
third relates to the interaction of concentration of ownership with prior year performance for the different classes of investors. The fourth relates to increases in concentrations
of share holdings. The fifth relates to the interaction between these increases and performance in the prior year. The sixth relates to the structure of the board (separation of
CEO-Chairman 1 if there is separation, %nonexec. directors is the proportion of nonexecutives on the board), leverage, new equity issues 1 if there is a new issue and
takeovers 1 if the firm is acquired. The seventh relates to the interaction of the board, financial structure, and takeover variables with prior year performance. The eighth
relates to a dummy variable which equals zero in the years 1988–1989 and 1 in the other years (1990–1993) reflecting the reduction in ownership disclosure from 5 to 3%
over the period 1989–1990.The ninth relates to size of firm measured by log of total sales. Ownership variables, capital leverage, and % of nonexecutives are all in percentages.
*
Significant at better than the 10% level.
**
Significant at better than 5% level.
***
Significant at better than 1% level.
Source: own calculations based on annual reports, Datastream, London Business School’s Risk Measurement Service.
332 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 333
both significant and focused and are thus 4.4 and 6.8% increases in board turnover,
far ahead in the horse race. respectively. Increases in institutional
ownership are associated with significant
4.2. Poor performance decreases in board turnover, suggesting that
institutions sell to other investors to accom-
The results in the previous section were plish board replacements in poorly performing
based upon all firms in the sample. In this companies. The economic effect of such
section, we examine only poorly performing institutional sales is between 3.5 and 7.2%
firms. We look at how board turnover is increases in board turnover for a reduction in
related to the five governance mechanisms holdings from upper decile to average levels.
in the presence of five definitions of poor
performance. The first is annual abnormal Board structure
returns of less than —50%. The second is
incidence of earnings losses. The third is The proportion of nonexecutives on the
dividend cuts and omissions. The other two board of poorly performing companies is
definitions take even more extreme measures not significantly related to board turnover.
of poor performance: earnings losses com- The number of nonexecutives does not
bined with dividend cuts and omissions, and therefore affect managerial disciplining.
abnormal returns of less than 50% However, separation of the roles of chair-
combined with earnings losses and dividends man and chief executive is associated with
cuts and omissions. Clearly, the sample sizes significantly higher board turnover—
for the fourth and fifth definitions are between 8.5 and 12.0% more turnover.
appreciably smaller than for the other three.
Table 12.8 reports regression results for
Financial structure
our sample of poorly performing firms with
the same variables as in Table 12.7, exclud- Capital leverage is associated with signifi-
ing the interactive terms with performance. cantly higher board turnover. An increase
from average to upper decile levels of lever-
Existing shareholders age raises board turnover by between 2.5
and 3.9%. Firms with earnings losses that
The results for the worst performing firms make rights issues have a 13.8% higher level
are almost identical to those of the com- of board turnover than those that do not.
plete sample. There is no evidence of share
concentrations affecting turnover, except Takeovers
for holdings by executive directors. Again
there is strong evidence of entrenchment. Takeovers are a highly significant influence
The economic effect of moving from on board turnover in the worst performing
average to upper decile executive ownership companies and are associated with a 64.1
is a decrease of between 10.4 and 17.9% to 78.7% increase in executive board
in executive board turnover in the worst turnover.This is, however, appreciably lower
performing firms depending on what measure than the board turnover of firms in the
of poor performance is used. complete sample that are targets of
takeovers.
Increases in shareholdings
Summary
As in the complete sample, increases in
shareholdings by families and executive The results for the worst-performing
directors are associated with significant companies are very similar to those for the
increases in board turnover. The economic complete sample.There is evidence of man-
effects of moving from average to upper agerial entrenchment through executive
decile ownership by families and executive share ownership while block holdings in the
directors are between 5.8 and 8.2%, and hands of other investors are not associated
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 335
Families and indiv. 1.0760 0.08 1.1109 0.06 1.5230 0.01 1.2999 0.06 1.0070 0.33
Executive directors 0.4983 0.06 0.4874 0.11 0.7542 0.01 0.7155 0.02 0.4494 0.16
Nonexecutive dir. 0.1054 0.87 0.1597 0.82 0.3288 0.69 1.2174 0.34 2.6717 0.06
Board structure, leverage, and takeovers
% nonexec. dir. 0.0870 0.54 0.0625 0.70 0.1074 0.45 0.0903 0.61 0.2975 0.22
Separ. CEO-chair 8.4895 0.09 9.8290 0.09 2.3962 0.63 12.0251 0.06 5.4033 0.51
Cap. Leverage (book) 0.0751 0.03 0.0707 0.04 0.0982 0.00 0.0835 0.02 0.0635 0.07
New equity issue 11.0971 0.12 13.6903 0.08 5.3056 0.49 11.7129 0.20 15.7946 0.18
Mergers/acquisitions 78.7200 0.00 76.0784 0.00 76.5755 0.00 69.1113 0.00 64.0700 0.00
Disclosure dummy 5.4827 0.33 5.3182 0.51 4.3605 0.48 6.5226 0.45 13.9756 0.16
Total sales (log) 1.19E-07 0.85 3.12E-08 0.96 3.08E-07 0.55 1.80E-08 0.98 2.60E-06 0.65
variables are exogenous. To investigate this United States. In Section 5.3, we describe
question we lagged all independent variables how regulatory differences between the
by one year in a panel OLS regression with United Kingdom and United States might
fixed effects.We find that most of the results explain differences in outcomes between
in the previous regressions are supported: two seemingly similar capital markets.
board turnover increases with past poor
performance and it decreases with high 5.1. Influence of UK regulation
concentration of ownership when held by
executive directors (entrenchment). Book
leverage is positively correlated with execu- 5.1.1. Protection of minorities
tive turnover, but interest coverage is not. A Minority and dispersed shareholders are
Tobit with lags produced some significance protected in the United Kingdom by The
on rights issues with board turnover. City Code on Takeovers and Mergers
Demsetz and Lehn (1985) and (called “The Code”) and by UK company
Himmelberg et al. (1999) argue that own- law.These create obstacles to building con-
ership is endogenous to performance. We trolling stakes. For example, an outside
investigated this by running reversed blockholder who owns 15% or more of the
regressions of performance on ownership equity of a firm must make public their
and changes in ownership, disaggregated by intentions of launching a takeover. Where a
different classes of investor. We ran six stake of 30% or more has been acquired,
regressions using abnormal returns, earn- there is a compulsory tender provision for
ings losses, changes in earnings per share, all remaining shares and the tender price
return on equity, cash flow margins, and must be at least that paid for any shares
changes in dividends per share as depend- acquired over the previous 12 months.
ent variables. No consistent relation was Purchases of share blocks in excess of 3%
found between these measures of perform- together with the identity of the buyer must
ance and either ownership or changes in be disclosed to the market. These rules are
ownership. designed to establish “fair play” in
takeovers and to reduce the potential for
Summary predators to purchase stakes cheaply.
However, they have the effect of raising
We have investigated the robustness of the acquisition costs.
results reported in the previous sections to Other rules affect incentives to acquire
replacing board turnover with CEO less than 100% of the shares of a target.
turnover and to tests of serial correlation The Stock Exchange lays down specific
and exogeneity of the independent variables. rules concerning transactions between
We have found that the conclusions of the controlling blockholders, who own more
previous section are robust to these than 50% of shares, and related parties.
changes and that there is some evidence These state that the firm “must be capable
that board structure has more of an at all times of operating and making
influence on CEO replacement than on decisions independently of any controlling
executive board turnover. shareholder and all transactions and
relationships in the future between the
applicant and any controlling shareholder
5. INFLUENCE OF REGULATION must be at arm’s length and on a normal
ON GOVERNANCE commercial basis.”29 A majority of the
directors of the board of the subsidiary
In Section 5.1 we discuss how UK regulation must be independent of the parent firm and
may have affected the above results on minority shareholders have the right to be
governance. In Section 5.2, we compare consulted about, and approve, transactions
the results that we have found in the United with the parent firm (Sections 11.4 and
Kingdom with those reported in the 11.5 of The Stock Exchange rules). The
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 339
effect of these rules is to increase the costs Olivier Company, shareholders controlling
of partial stakes. They explain why almost 30% of the shares prevented a waiver of a
all bids are made conditional on acceptance rights issue.33 Even where shareholders
by 90% or more of target shareholders.The vote to drop pre-emption rights the dis-
remainder can be purchased compulsorily at count of any new issue must not exceed
the original bid price using a squeeze out 10% of the market price at the time of the
rule under the 1948 Companies Act. In a issue’s announcement (paragraph 4.26,
recent bid by Capital Shopping Centres Stock Exchange Rules, 1999). These rules
(CSC) for 25% of shares of Liberty ensure that old shareholders cannot be
International that they did not already own, diluted by new shareholders.34 They are
the independent directors of Liberty advised reinforced by guidelines, authorized by the
minority shareholders not to accept the National Association of Pension Funds and
offer. They argued that “it does not give any the Association of British Insurers, limiting
premium for full control of the company” companies to raising 5% of their share
(The Financial Times, October 19, 2000). capital each year by any method apart
from rights issues—and 7.5% in any
5.1.2. Fiduciary duties of directors rolling three-year period.
However, in other respects UK and U.S. where nonexecutive directors dominate the
results differ significantly. There is more board. Also, Gilson (1990) and Kaplan and
evidence in the United States than in the Reishus (1990) find that nonexecutive
United Kingdom of disciplining associated directors of poorly performing companies
with sales of share blocks. In the United lose reputation and are frequently unable
States, Bethel et al. (1998) report that to find replacement positions. In the United
purchases of share blocks by active Kingdom, we found no evidence of
investors are targeted on poorly performing disciplining by nonexecutive directors;
companies. Holderness and Sheehan indeed, the relation is negative between the
(1988) find that when their majority proportion of nonexecutives and board
blocks trade, there is substantial manage- turnover. Second, we find that capital
ment turnover and stock prices increase. In structure and new equity financing are
the United Kingdom, we have found little particularly significant influences on board
evidence that changes in share blocks by turnover in the United Kingdom.We are not
potentially active investors perform a disci- aware of any U.S. study reporting this
plinary function. relation. The Table below summarizes
There are two still more pronounced dif- evidence on the parties performing
ferences between the United Kingdom and disciplining in the United Kingdom and
United States. First, in the United States, United States. It shows the effect on
Weisbach (1988) reports a closer relation disciplining of different parties’ interventions
of CEO turnover to performance in firms in the two countries.
Effect on
disciplining United Kingdom United States
Negative Executive block holders Nonexecutive Executive block holders
directors
Neutral Other block holders Purchasers of Other block holders
blocks Providers of new finance
Positive Bidders Providers of new finance Some purchasers of blocks
Bidders Nonexecutive directors
5.3. Regulatory differences between Code requiring full bids for companies to
the United Kingdom and United States be made but there are much more extensive
takeover defences in state legislation and
5.3.1. Protection of minorities company charters than in the United
Kingdom (Miller, 1998).39, 40
We have already reported that protection The impediments to the exercise of control
of minorities is extensive in the United by dominant shareholders in the United
Kingdom. In terms of Goshen’s characteri- Kingdom and the more liberal view of
zation (1998) of systems of minority takeovers encourage full acquisitions of
protection, the United Kingdom has a companies rather than control through
“property rule” which prevents any trans- partial share blocks. This may explain why,
action from proceeding without the minor- in contrast to the United States, trades in
ity owner’s consent. In contrast, the United share blocks in the United Kingdom do not
States has a “liability rule” which allows involve active shareholders and are not
transactions to be imposed on an unwilling disciplinary in nature.
minority but ensures that the minority is
adequately protected in objective market 5.3.2. Fiduciary duties of directors
value terms.37, 38 Protection of investors,
especially minorities, is primarily the While powers to enforce fiduciary responsi-
concern of the courts. Until this year, there bilities on directors in the United Kingdom
was no U.S. equivalent of the UK Takeover are weak, in the United States, directors
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 341
and low interest coverage are associated restricted to the very worst performing
with high levels of board turnover in poorly firms. The greater reliance on “unfocused”
performing companies. At first sight, this takeovers may to some extent have com-
suggests that creditor intervention is the pensated for this but, if regulation makes
main source of corporate reorganization. this an expensive form of intervention, then
However, evidence from 34 case studies there may still be inadequate restructuring
and from the regression analyses revealed in the United Kingdom prior to the
an important role for new equity issues in emergence of financial distress.
board restructurings.
Regulation appears to be a significant
influence on this pattern of governance in NOTES
the United Kingdom. Strong minority pro-
tection has discouraged partial accumula- 1 We are grateful for comments from partici-
tion of share blocks in favor of full pants at conferences at the London School of
acquisitions in takeovers. Weak fiduciary Economics, Fondazione ENI Enrico Mattei in
obligations on directors have resulted in Milan, Wissenschaftszentrum Berlin fur
nonexecutives playing more of an advisory Sozialforschung, the University of Amsterdam,
the University of Venice, the Autonoma
than a disciplinary role. Rights issue University of Barcelona, the conference of the
requirements protect existing shareholders ESRC Network for Industrial Economics,
against wealth transfers initiated by the the European Corporate Governance Network,
management of poorly performing firms the European Finance Association Meeting in
and allow outside shareholders to impose Milan, the CEPR European Summer
board changes as a condition for the provi- Symposium in Financial Markets in Gerzensee,
sion of new equity finance. the CEPR workshop on corporate finance in
Thus, despite the frequent categorization Toulouse, the SUERF meeting in Budapest, the
of UK regulation under the same “common TMR Meeting in Florence, the EFMA Meeting
law” classification as the United States, in Lisbon, and the Annual Meetings of the
American Finance Association. We are particu-
there are significant differences—more larly grateful to our discussants Fabrizio Barca,
minority investor protection in the United Alan Hughes, Alexander Ljungqvist, Marco
Kingdom than in the United States, less Pagano, Sigurt Vitols, and Jozef Zechner. We
fiduciary obligations on directors in the received useful suggestions from Margaret
United Kingdom but stricter rights issue Bray, Luis Correia da Silva, Rafel Crespi, Jay
requirements. These differences appear to Dahya, Carles Gispert, Marc Goergen, Michel
be associated with more governance Habib, Uli Hege, Joe McCahery, Piet Moerland,
through partial acquisitions of share blocks Anthony Neuberger, Kjell Nyborg, Enrico
and by nonexecutive directors in the United Perotti, Christian Rydqvist, Henri Servaes,
States, but less governance through the James Tomkins, David Webb, and Michael
Weisbach. We thank Matthew Beardmore-Gray
provision of new financing than in the and Huw Jones of Prudential for very helpful
United Kingdom. Subtle differences in discussions. We are grateful to the referees of
regulatory systems may therefore be our paper and to the editor of the Journal of
associated with pronounced difference in Financial Intermediation for very helpful com-
governance outcomes. ments. Julian Franks and Colin Mayer acknowl-
What is the significance of these differ- edge financial support from the European
ences for corporate performance? We have Union’s Training and Mobility of Researchers
not attempted to answer this question Network (Contract FRMXCT960054) and Luc
directly in this paper; however, it might be Renneboog is grateful for financial support
argued that greater reliance on financial from the Netherlands Organization of Scientific
Research.
constraints and less reliance on boards in 2 To whom correspondence should be
the United Kingdom leads to a greater addressed.
concentration of disciplining on the worst- 3 Weisbach (1988) also assumes that any res-
performing firms.This is consistent with the ignation over 62 is natural turnover, unless there
observation that higher board turnover is is evidence of conflict.
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 343
4 This uses a Capital Asset Pricing Model confined to the Fortune 500 companies,
with Bayesian updating and a thin trading whereas our sample is drawn from all quoted
correction. Further details can be found in companies.
LSPD, London Business School, 1997. 18 Hermalin and Weisbach (1998) predict
5 This is after adjusting for bankruptcies, that the “probability that independent directors
acquisitions, and double counting of those are added to the board increases following poor
companies appearing in the bottom decile in corporate performance.”
more than one year. 19 Similar relations between interest coverage
6 Lai and Sudarsanam (1998) also report and other measures of performance are found. For
declines in board turnover after performance example, median interest cover for companies with
declines. dividend cuts is 0.6 compared with 4.5 for those
7 The disclosure threshold in the United with stable or increasing dividends.
States is 5%. 20 A level of two is chosen because invest-
8 As well as direct (or beneficial) holdings, we ment grade companies “typically have coverage
included all nonbeneficial holdings held by ratios exceeding two times interest expense”
directors on behalf of families and charitable (Copeland et al., 1995, p. 178).
trusts. Directors do not obtain cash flow bene- 21 Rights issues are offered to existing share-
fits from these holdings but they have control holders and any rights not taken up may be sold
rights. We also investigated nominee holdings for the benefit of the shareholder; in open offers,
and found that in 95% of the cases, institu- rights not taken up may not be sold by the
tional investors used the nominee registration to original holder and can be sold by the company
reduce administrative costs. The nominee share- to other shareholders.
holdings were classified according to category 22 We controlled for the change in the disclo-
of shareholder using nominee accounts. sure threshold from 5 to 3% in 1989 by
9 Recent IPOs may particularly affect the including a dummy variable which equals 0 for
pattern of ownership; however, the large majority 1988–1989 and 1 afterwards. We also verified
of our companies, 71%, have been listed for at the robustness of our results, by including
least eight years. dummy variables for the years and by running
10 Individual stakes of 30% or greater were the regression on subsamples which excluded
almost always built up prior to the company’s the years 1989 and 1990. This did not signifi-
IPO. In other cases they resulted from acquisi- cantly alter the results reported below.
tions in which target shareholders did not tender 23 In the case of zero-one (or minus one)
all their shares. dummy variables (earnings losses, dividend cuts
11 Bristow’s data (1995) show that for 3963 and omissions, takeovers, and new equity
firms median insider ownership is 12.5% and issues), economic effects relate to the effect of
for other much smaller samples the median rises switching from zero to one (or minus one). The
to about 16%. tables with the means and percentiles of inde-
12 The quotation of 3 was suspended or pendent variables are available on request.
cancelled, 1 company was taken private and the 24 We examined the effect of interactions
equity of 3 other firms was converted into a between dividend cuts/omissions and negative
different security. share price performance on board turnover by
13 A total of 14% of companies had quota- restricting the regression to the sample of
tions suspended or cancelled. companies that incurred abnormal losses. The
14 The bankruptcy rate is low. For the original regression results were little affected.
sample of 243 companies it is zero. For the 25 The results were rerun assuming coalitions
additional sample of poorly performing compa- of shareholders, for example, all outside share-
nies it is 4% for the period 1988–1993. It holders were grouped together, with executive
increases to 6% if we follow the sample to 1997. and nonexecutives forming two other groups.
15 Martin and McConnell (1991) also report The results are similar to those in Tables 12.7
high board turnover post takeover, between 58 and 12.8.These results are available on request.
and 64%, depending upon whether the takeover 26 Results remained unaltered when leverage
was hostile or friendly. and interest cover were included with a lag of a
16 If we included increases in existing share- year, rejecting a reversed causation explanation
holdings, the totals would increase by 22 for [5, for their significance.
10%], 15 for [10, 25%], and 3 for [25, 50%]. 27 Including a variable for the length of
17 The lower rate of block sales in the United tenure of the CEO eliminated the significance
States may be due to their sample being of the separation variable. However, length of
344 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
tenure was only available for a subsample of of parents. Eisenberg (1976) states that “the
firms (130 observations were lost) and has not checks on unfair dealing by the parent are few.
therefore been shown in the table.There is likely In theory, of course, the fairness of the parent’s
to be less CEO–chairman separation in compa- behaviour is subject to the check of judicial
nies where CEOs have been in place for long review; but in practice such review is difficult
periods of time and a nonexecutive chairman even where the courts have the will to engage in
may be able to exert less influence where CEOs it, and they often lack the will.”
are firmly entrenched. As a consequence, tenure 40 La Porta et al. (1997, 1999) measure anti-
extinguishes the significance of separation. director rights in the United Kingdom and United
28 Interest coverage was substituted for States and find greater protection for minorities
leverage in separate regressions. in the United States. However, they focus on the
29 See Sections 3.12 and 3.13 of Chapter 11 rights under commercial law and do not consider
of the Stock Exchange rules. the influence of the nonstatutory, but highly effec-
30 See also Parkinson (1993). tive,Takeover Code and Stock Exchange rules.
31 In Germany there are particular lawsuits by 41 Clark (1986) describes the circumstances
shareholders that must be funded by the company, under which shareholders can be sued in the
for example, where shareholders object to being United States. For example, in Smith v. Van
“squeezed out” when a blockholder has at least Gorkum, shareholders successfully sued direc-
95% of the target’s shares. The minority may tors for a breach of duty of care with respect to
demand a court hearing at the company’s expense. a merger. However, Clark also notes the paucity
32 If shareholders fail to take up their rights, of such successful cases.
the rights may be sold for the shareholder’s 42 An example in differing interpretations is
benefit. These pre-emption rights are recognised Preemptive Right to New Issues. The evidence is
in European Community law. clear that in the United Kingdom waiver of
33 Financial Times, May 29, 1998. rights issues is very difficult unless the company
34 The Exchange may relax these rules if the is in distress, whereas in the United States
company is in severe financial difficulties. waivers are the norm. Nevertheless, La Porta et al.
35 Morck et al. (1988), and McConnell and (1997) give the two countries the same score.
Servaes (1990) find that corporate performance
as measured by Tobin’s Q initially rises with low
levels of insider ownership (for example, up to
5% in the Morck et al. study) and then declines. REFERENCES
36 Our results are also consistent with the use
of anti-takeover amendments, recorded by Aghion, P., and Bolton, P. (1992). An
Borokhovich et al. (1997), to entrench manage- incomplete contracts approach to financial
ment, and Stultz’s argument (1988) that anti- contracting, Rev. Econ. Stud. 59, 473–494.
takeover amendments substitute for insider Atiyah, P., and Summers, R. (1987). “Form and
ownership as an entrenchment mechanism. Substance in Anglo-American Law,”
37 The difference that is drawn here between Clarendon, Oxford.
UK and U.S. minority protection conforms with
Ball, R., Kothari, S. P., and Robin, A. (1997).
the more general distinction which Atiyah and
Summers (1987) and Posner (1996) draw “The Effect of Institutional Factors on
between reliance on substantive reasoning under Properties of Accounting Earnings:
U.S. law and formal reasoning in UK law. International Evidence,” working paper,
38 The United States does have the Williams University of Rochester, May.
Act of 1968 which introduced rules on block Bethel, J. E., Liebeskind, J. P., and Opler, T.
disclosure (10% later amended to 5%), a min- (1998). Block share purchases and corporate
imum period for which a tender offer must be performance, J. Finance 53, 605–634.
left open, and a provision explicitly allowing Bianco, M., Gola C., and Signorini, L. F. (1996).
targets to sue bidding firms. Dealing with separation between ownership
39 Protection may even fall short of that
and control: State, family, coalitions and
provided by the application of a fair price rule.
Gilson (1995) argues that in Sinclair Oil Corp v. pyramidal groups in Italian corporate gover-
Levien business judgment rather than intrinsic nance, Fondazione Eni Enrico Mattei Note di
fairness tests should have been applied. It is Lavoro: 05/96, February.
debatable whether minority protection in the Borokhovich, K. A., Brunarski, K. R., and
United States interferes with business judgments Parrino, R. (1997). CEO contracting and
MANAGEMENT DISCIPLINE IN POORLY PERFORMING COMPANIES 345
anti-takeover amendments, J. Finance 52, Franks J., and Mayer, C. (2000). The ownership
1495–1517. and control of German corporations, Rev.
Brealey, R., and Myers, S. (1996). “Principles Finan. Stud, in press.
of Corporate Finance,” McGraw–Hill, Gilson, R. J. (1995). “The Law and Finance of
New York. Corporate Acquisitions,” Foundation Press,
Bristow, D. K. (1995). The time series and cross Westbury, New York.
sectional properties of directors and officers’ Gilson, S. C. (1990). Bankruptcy, boards, banks
stock holdings, mimeo, UCLA. and blockholders, J. Finan. Econ. 27,
Burkart, M., Gromb, D., and Panunzi, F. (1997). 355–387.
Large shareholders monitoring, and the Goshen, Z. (1998). Liability rule or property
value of the firm, Quart. J. Econ. 112, rule: Voting in conflict of interests in
693–728. corporate law and minority’s protection,
Burkart, M., Gromb, D., and Panunzi, F. (1998). mimeo.
Why higher takeover premia protect minority Hermalin, B., and Weisbach, M. (1991). The
shareholders, J. Polit. Econ. 196, 176–204. effects of board composition and direct
Cadbury, A. (1992). “Report of the Committee incentives on firm performance, Finan.
on the Financial Aspects of Corporate Manage. 20, 101–112.
Governance,” Gee & Co, London. Hermalin, B., and Weisbach, M. (1998).
Clark, R. C. (1986). “Corporate Law,” Little, Endogenously chosen boards of directors and
Brown & Co., Boston. their monitoring of the CEO, Amer. Econ.
Copeland, T., Koller, T., and Murrin, J. (1995). Rev. 88, 96–118.
“Valuation: Measuring and Managing the Himmelberg, C. P., Hubbard, R. G., and Palia, D.
Value of Companies,” Wiley, New York. (1999). Understanding the determinants of
Coughlan, A. T., and Schmidt, R. M. (1985). managerial ownership and ownership,
Executive compensation, managerial turnover J. Finan. Econ. 53, 353–384.
and firm performance: An empirical investi- Holderness, C. G., and Sheehan, D. P. (1988).
gation, J. Acc. Econ. 7, 42–46. The role of majority shareholders in publicly
Dahya, J., McConnell, J., and Travlos, N. (2000). held corporations: An exploratory analysis,
“The Cadbury Committee, Corporate J. Finan. Econ. 20, 317–346.
Performance and Top Management Turnover,” Jensen, M. (1986). Agency costs of free cash
working paper, Cardiff Business School. flow, corporate finance and takeovers, Amer.
Demsetz, H., and Lehn, K. (1985).The structure Econ. Rev. 76, 323–339.
of corporate ownership: Causes and conse- Jensen, M. (1989). The eclipse of the public
quences, J. Polit. Econ. 93, 1155–1177. corporation, Harvard Bus. Rev. 61–74.
Denis, D., and Denis, D. K. (1995). Performance Kaplan, S., and Reishus, D. (1990). Outside
changes following top management turnover, directorships and corporate performance,
J. Finance 50, 1029–1057. J. Finan. Econ. 27, 389–410.
Denis, D., and Kruse, T. (2000). Managerial dis- Kini, O., Kracaw, W., and Mian, S. (2000). The
cipline and corporate restructuring following nature of discipline by corporate takeovers in
performance declines, J. Finan. Econ. 55, an active takeover market, mimeo, Georgia
391–424. State University.
Eisenberg, M. (1976). “The Structure of the La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Corporation,” Little, Brown & Co, Boston. and Vishny, R. (1997). Legal determinants of
Fama, E. (1980). Agency problems and the the- external finance, J. Finance 52, 1131–1150.
ory of the firm, J. Polit. Econ. 88, 288–307. La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Fama, E., and Jensen, M. (1983). Separation of and Vishny, R. (1998). Law and finance,
ownership and control, J. Law Econ. 26, J. Polit. Econ. 106, 1113–1155.
301–325. La Porta, R., Lopez-de-Silanes, F. Shleifer, A.,
Franks, J., and Mayer, C. (1996). Hostile and Vishny, R. (1999). Ownership around the
takeovers in the UK and the correction of world, J. Finance 54, 471–517.
managerial failure, J. Finan. Econ. 40, Lai, J., and Sudarsanam, S. (1998). Corporate
163–181. restructuring in response to performance
346 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
decline: Impact of ownership, governance and Parkinson, J. (1993). “Corporate Power and
lenders, Europ. Finance Rev. 1, 197–233. Responsibility: Issues in the Theory of
Loderer, C., Sheehan, D. P., and Kadlec, G. Company Law,” Oxford Univ. Press, Oxford.
(1991). The pricing of equity offerings, Posner, R. (1996). “Law and Legal Theory in
J. Finan. Econ. 29, 35–58. the UK and USA,” Clarendon, Oxford.
Marsh, P. (1992). Dividend announcements and Renneboog, L. (2000). Ownership, managerial
stock price performance, unpublished manu- control and the governance of poorly per-
script, London Business School. forming companies listed on the Brussels
Martin, K., and McConnell, J. J. (1991). Stock Exchange, J. Banking Finance 24,
Corporate performance, corporate takeovers, 1959–1995.
and management turnover, J. Finance 46, Romano, R. (1991). The shareholder suit and
671–687. litigation without foundation, J. Law Econ.,
McConnell, J. J., and Servaes, H. (1990). Organ. 7, 55–87.
Additional evidence on equity ownership and Scharfstein, D. (1988). The disciplinary role of
corporate value, J. Finan. Econ. 27, takeovers, Rev. Econ. Stud. 55, 185–199.
595–612. Shleifer, A., and Vishny, R. (1986). Large share-
Miller, G. (1998). Political structure and corpo- holders and corporate control, J. Polit. Econ.
rate governance: Some points of contrast 94, 461–488.
between the United States and England, in Stapledon, G. (1996). “Institutional
“The Sloan Project on Corporate Shareholders and Corporate Governance,”
Governance,” pp. 629–648. Corporate Clarendon, Oxford.
Governance Today, Columbia Law School. Stulz, R. (1988). Managerial control of voting
Millstein, I., and MacAvoy, P. (1998).The active rights: Financing policies and the market for
board of directors and performance of the corporate control, J. Finan. Econ. 20, 25–54.
large publicly traded corporation, Columbia Warner, J. B., Watts, R. L., and Wruck, K. H.
Law Rev. 98, 1283–1321. (1988). Stock prices and top management
Morck, R., Shleifer, A., and Vishny, R. (1988). changes, J. Finan. Econ. 20, 431–460.
Management ownership and market valua- Weisbach, M. S. (1988). Outside directors
tion: An empirical analysis, J. Finan. Econ. and CEO turnover, J. Finan. Econ. 20,
20, 293–316. 431–446.
Chapter 13
ABSTRACT
In 1992, the Cadbury Committee issued the Code of Best Practice which recommends that
boards of U.K. corporations include at least three outside directors and that the positions of
chairman and CEO be held by different individuals. The underlying presumption was that these
recommendations would lead to improved board oversight. We empirically analyze the
relationship between CEO turnover and corporate performance. CEO turnover increased
following issuance of the Code; the negative relationship between CEO turnover and performance
became stronger following the Code’s issuance; and the increase in sensitivity of turnover to
performance was concentrated among firms that adopted the Code.
THE CADBURY COMMITTEE WAS APPOINTED The apparent reasoning underlying the
by the Conservative Government of the Committee’s recommendations is that
United Kingdom in May 1991 with a broad greater independence of a corporate board
mandate to “ . . . address the financial improves the quality of board oversight.
aspects of corporate governance” (Report As of 2001, the Code has not been
of the Committee on the Financial Aspects enshrined into U.K. law and compliance
of Corporate Governance, 1992, Section with its key provisions is entirely voluntary.
1.8). The Committee was chaired by Sir Nevertheless, the Code is not without
Adrian Cadbury, CEO of the Cadbury con- “teeth.” First, the Cadbury Committee’s
fectionery empire, and included other sen- report explicitly recognizes that legislation
ior industry executives, finance specialists, would very likely follow if companies did
and academics. In December 1992, the not comply with the guidelines of the Code
Committee issued its report, the corner- (Report of the Committee on the Financial
stone of which was The Code of Best Aspects of Corporate Governance, 1992,
Practice, which presents the Committee’s Section 1.1). Second, the report has been
recommendations on the structure and given further bite by the London Stock
responsibilities of corporate boards of Exchange (LSE), which, since June 1993,
directors. The two key recommendations of has required a statement from each listed
the Code are that boards of publicly traded company that spells out whether the com-
companies include at least three nonexecutive pany is in compliance with the Code and, if
(i.e., outside) directors and that the positions not, requires an explanation as to why the
of chief executive officer (CEO) and chair- company is not in compliance.
man of the board (COB) of these companies To appreciate the significance of the
be held by two different individuals.1 Cadbury Committee and its recommendations,
348 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
it is important to appreciate the environment who felt that it did not go far enough. The
surrounding the establishment of the general unease of those who felt it went too
Committee. First, the Committee was far can be summarized as a concern that
appointed in the aftermath of the the delicate balance between shareholders
“scandalous” collapse of several prominent and managers is better left to the forces of
U.K. companies during the later 1980s and competition. A less generous interpretation
early 1990s, including Ferranti, Colorol of this perspective, which was most
Group, Pollypeck, Bank of Credit and frequently espoused by corporate managers,
Commerce International, and Maxwell might be characterized as “leave us
Communication. The broadsheet press alone—we know best.”
popularly attributed these failures and
others to weak governance systems, lax There is danger in an over emphasis on
board oversight, and the vesting of control monitoring; on non-executive directors
in the hands of a single top executive. independence . . . [and] on controls over
decision making activities of companies.
The Cadbury Committee was set up in (Green, 1994)
response to a number of corporate
scandals that cast doubt on the systems The general concern of those who thought
for controlling the ways companies are that the report did not go far enough
run.The downfall of powerful figures such centered on the “voluntary” nature of the
as Asil Nadir or the late Robert Maxwell, Report’s recommendations.
whose personal control over their compa-
nies was complete, raised fears about the The Committee’s recommendations
concentration of power. (Self-regulation are steps in the right direction.
seen as the way forward, 1992) But . . . [s]hareholders, investors and
creditors will have been disappointed
Second, historically, executive (i.e., inside) that just when the corporate failures of
directors have heavily dominated U.K. recent years cried out for bold and imag-
boards. For example, during 1988, for only inative legal reform, the body from which
21 companies of the Financial Times (FT) so much had been expected came up with
500 did outside directors comprise a a little tinkering and a voluntary code
majority of the board and, when boards are (Cadbury Committee Draft Orders
ranked according to the fraction of outside Mixed News for Shareholders, 1992).
board members, outsiders comprised only
27 percent of the median board’s member- Against this background, this study empiri-
ship (The Corporate Register, 1989). In cally investigates the impact of the key
comparison, outsiders comprised a majority Cadbury recommendations on the quality
of the board for 387 of the Fortune 500 of board oversight in U.K. firms over the
companies. Furthermore, for the median period 1989 through 1996.
board of the Fortune 500 companies, We begin our investigation with the
outside directors comprised 81 percent of presumption that an important oversight
the membership (Annual Corporate Proxy role of boards of directors is the hiring and
Statements). With respect to the joint posi- firing of top corporate management. We
tion of CEO and COB, the United Kingdom further presume that one indicator of effec-
and United States historically are similar. tive board oversight is that the board
For example, during 1988, a single individ- replaces ineffective or poorly performing
ual jointly held the positions of CEO and top management. Finally, we presume that
COB for 349 of the Fortune 500 and for corporate performance is a reliable proxy
328 of the FT 500. for the effectiveness of top management.
At its issuance, the Cadbury Report was With those presumptions in place, we
greeted with skepticism both by those who empirically investigate the relationship
felt that it went too far and by those between top management turnover and
CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 349
listed firms traded elsewhere prior to In our tests, we employ both accounting
entering the LSE Official List. For these earnings and stock returns to measure
firms, stock price data are collected for up corporate performance. Specifically, as our
to three years preceding their listing dates. measure of accounting earnings, we use
For other firms, we use price data three-year average industry-adjusted return
beginning with their entry onto the on assets (IAROA). For each firm in the
Official List. sample and for each year, we calculate return
To determine top management turnover, on assets (ROA) as earnings before depreci-
we compare the names of top management ation, interest, and taxes (EBDIT) divided by
from year to year over the time period beginning-of-the-year total assets. Then, for
December 1988 through December 1996. each firm with the same FTIC as the sample
For each company, we identify the top exec- firm, we calculate ROA in the same way.
utive as the individual with the title of CEO Next, for each FTIC group for each year, we
or Executive Chairman. In addition, we determine the median ROA. IAROA is calcu-
identify other board members as members lated by subtracting the industry median
of the top management team if the board ROA from the sample firm’s ROA for each of
member is an employee of the firm and the three years prior to a turnover event.The
holds the title of Chief of Operations or average of these three IAROAs is in our
Managing Director. If the name of the top measure of accounting performance.
executive changes between successive For measuring stock price performance,
years, we classify that as turnover in the we use industry- and size-adjusted stock
top executive. For other members of the top returns (ISARs), where ISARs are calcu-
management team, if a name disappears lated by subtracting the daily stock returns
from the top management list, that event of an industry- and size-matched portfolio
is deemed to be a turnover in the top from the return of the sample firm
management team excluding the top executive. beginning 36 calendar months prior to, and
If the top executive exits the list of top ending 2 days prior to, the announcement
management and is replaced by another of the management change. To construct
member of the top team, that event is the industry- and size-matched portfolio,
considered turnover in the top executive for each sample firm, all other firms with
position, but not turnover in the top man- the same FTIC code are ranked from
agement team. We do not count as turnover largest to smallest according to their
the event in which the position of Executive equity market values. The firms are then
Chairman is split into the positions of CEO divided into four size portfolios. The differ-
and COB. (Henceforth, we refer to the top ences between the return on the stock in
executive position as the CEO.) our sample and the equal-weighted average
We further identify turnover as “forced” return of the industry- and size-matched
by examining articles in the Extel Weekly portfolio are calculated. The sum of these
News Summaries, the Financial Times, and differences is the ISAR for that firm.
McCarthy’s News Information Service.
Turnover is labeled forced when (a) a news
article states that the executive was II. CHARACTERISTICS OF THE
“fired”; (b) an article states that the exec- SAMPLE
utive “resigned”; or (c) an article indicates
that the company was experiencing poor To conduct our analysis, we split manage-
performance. In addition, for criteria ment turnover along two dimensions. First,
(b) and (c), the executive must be less than we split turnover events into a pre-Cadbury
60 years old and no other article can time period (1989 through 1992) and a
indicate that the executive took a position post-Cadbury time period (1993 through
elsewhere or cite health or death as the 1996). Descriptive data for these two
reason for the executive’s departure. All samples are presented in the first set of
other turnover is labeled “normal.” columns in Table 13.1.
Table 13.1 Financial, Ownership, and Board Characteristics for 460 U.K. Industrial Firms over the Period 1989 through 1996
Descriptive statistics for a random sample of 460 publicly traded U.K. industrial firms over the period 1989 through 1996. The sample firms are classified
into three sets based on whether they were (a) always in compliance with the Cadbury recommendations, (b) never in compliance with the Cadbury
recommendations, or (c) adopted Cadbury recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods, pre- and postpublication
of the Cadbury Report (1989 through 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, pre- and postadoption of
the Cadbury recommendations (y 4 through y 1 and y 1 through y 4). The sample is taken from industrial companies included in the Stock
Exchange Yearbook, the Corporate Register, and listed on the LSE. Firms that leave the sample between 1989 and 1996 are replaced by firms entering
the LSE on the date closest to departure. Management and board characteristics are from the Corporate Register. Accounting information and share prices
are from Datastream.
Always in Never in
Full Sample Compliance Compliance Adopted Cadbury
N 460 N 150 N 22 N 288
Years Mean (Median) Mean (Median) Mean (Median) Years Mean (Median)
Book value of assets 1989–1992 149.2* (50.8) 148.8* (48.8) 146.8 (45.3) y 4 to y 1 150.1* (50.7)
(£ million) 1993–1996 156.8 (56.7) 155.6 (53.2) 150.7 (49.2) y 1 to y 4 158.6 (58.7)
Top executive ownership 1989–1992 2.23 (2.39) 2.22 (2.23) 9.93 (10.11) y 4 to y 1 2.23 (2.44)
1993–1996 2.24 (2.34) 2.09 (1.96) 8.34 (9.27) y 1 to y 4 2.25 (2.43)
Board ownership 1989–1992 10.70 (10.96) 12.83 (12.60) 16.45 (15.93) y 4 to y 1 9.79 (9.99)
1993–1996 11.35 (10.83) 13.32 (13.01) 16.09 (15.99) y 1 to y 4 11.09 (10.44)
Institutional ownership 1989–1992 21.39 (20.84) 22.55 (20.60) 16.37 (15.93) y 4 to y 1 19.88 (19.82)
1993–1996 20.04 (19.05) 22.09 (19.51) 16.29 (16.22) y 1 to y 4 20.09 (20.44)
Number of block holders 1989–1992 2 (2) 3 (2.04) 1 (1) y 4 to y 1 2* (2)*
1993–1996 3 (3) 3 (3.05) 1 (1) y 1 to y 4 3 (3)
Board size 1989–1992 5.71 (5.00)* 6.69 (6.00) 4.53 (4.00) y 4 to y 1 5.49* (5.00)*
1993–1996 7.29 (7.00) 7.41 (7.00) 5.02 (5.00) y 1 to y 4 7.13 (7.00)
Percent outside directors 1989–1992 35.3** (36.9)** 48.6 (43.4) 17.9 (15.4) y 4 to y 1 26.1** (25.7)**
1993–1996 46.0 (43.1) 48.5 (45.8) 21.5 (20.9) y 1 to y 4 46.6 (40.6)
** and * denote significance at the one and five percent levels, respectively, for both the t-statistic and Wilcoxon statistic. Tests are comparisons of before and after Cadbury
values.
CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 351
352 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Always in Never in
Full Sample Compliance Compliance Adopted Cadbury
N 460 N 150 N 22 N 288
Years Incidence Rate Incidence Rate Incidence Rate Years Incidence Rate
* * **
All CEO Turnover 1989–1992 119 6.48 35 5.44 4 4.55 y 4 to y 1 80 7.24**
1993–1996 138 7.71 37 5.75 3 3.41 y 1 to y 4 98 8.87
Forced CEO Turnover 1989–1992 57* 3.10* 24 3.76 3 3.26 y 4 to y 1 30** 2.71**
1993–1996 79 4.30 20 3.14 1 1.09 y 1 to y 4 58 4.98
** and * denote significance at the one and five percent levels, respectively, for both the t-statistic and Wilcoxon statistic. Tests are comparisons of before and after Cadbury
values.
354 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Table 13.3 Forced CEO Turnover in 460 U.K. Industrial Firms Grouped by Quartiles of Performance, 1989 through 1996
Forced CEO turnover for a random sample of 460 publicly traded nonfinancial U.K. firms grouped into quartiles based on IAROA in the two four-year
periods during the interval 1989 through 1996. IAROA is calculated as earnings before interest, taxes and depreciation divided by the total book value of
assets less the median performance of firms in the same FTIC grouping. Three years of IAROA are averaged. Turnover is classified as forced by examining
news articles in the Extel Weekly News Summaries, the Financial Times, and McCarthy’s News Information Service. The sample firms are classified into
three sets based on whether they were (a) always in compliance with the Cadbury recommendations, (b) never in compliance with the Cadbury recom-
mendations, or (c) adopted Cadbury recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods, pre- and postpublication of
the Cadbury Report (1989 through 1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, pre- and postadoption of
the Cadbury recommendations (y 4 through y 1 and y 1 through y 4).
Interval
Quartile 1 Quartile 4
(Lowest IAROA) Quartile 2 Quartile 3 (Highest IAROA)
Rate Rate Rate Rate
Years Incidence (%) Incidence (%) Incidence (%) Incidence (%)
quartile four containing the 115 firms with for forced turnover. We estimate separate
the highest IAROA. regressions using three-year prior IAROAs
For the full sample, both before and and three-year ISARs as our performance
after Cadbury, the incidence and rate of measures. We include yearly observations
forced CEO turnover increases as we move of four control variables: fraction of
from the best to the poorest performing shares owned by directors, fraction of shares
firms. Additionally, the data indicate that owned by institutions, number of block
the increase in CEO turnover from before to shareholders, and log of total assets.
after Cadbury that we document in Table 13.3 The results of our regressions are
is due to an increase in turnover in the presented in Tables 13.4 and 13.5. In
lowest two performance quartiles in the Table 13.4, the performance variable is
adopted-Cadbury set of firms. For example, logIAROA. In Table 13.5, performance is
for this set of firms, the rate of turnover in logISAR.2 Panel A of each table presents
quartiles one and two increased by nearly regressions with all CEO turnover as the
100 percent, from 5.5 percent to 10.8 percent dependent variable and Panel B presents
(p-value 0.01) and by almost 300 regressions with forced turnover as the
percent, from 2.3 percent to 6.9 percent dependent variable. In total, we have 20
(p-value 0.04), respectively, from before regressions that have either logIAROA or
to after adoption of Cadbury. In comparison, logISAR as an independent performance
for the always-in-compliance set, in the variable. In each regression, the coefficient
same bottom two quartiles, the rate of of the performance variable is negative
turnover is essentially unchanged from and, with two exceptions, each has a
before to after Cadbury. p-value of less than 0.05. Thus, CEO
The data in Table 13.3 are representative turnover is significantly negatively corre-
of the pattern of turnover (not shown) that lated with corporate performance: the
emerges when we consider all CEO poorer the firm’s performance, the greater
turnover and when we evaluate perform- the likelihood that the CEO will depart his
ance based on ISARs. That is, turnover is position. (We also estimate regressions
concentrated in the poorest performing separately for the pre- and post-Cadbury
quartiles of firms, and the increase in time periods [not shown]. In every regres-
turnover is concentrated in the adopted- sion, the coefficient of the performance
Cadbury set of firms. variable is negative with a p-value less than
0.05. Thus, turnover is significantly
C. Multivariate analysis of the negatively correlated with performance
relationship between top both before and after Cadbury.)
management turnover and Of the four control variables, only the
corporate performance fraction of shares owned by directors
regularly has a p-value less than 0.10. The
The final questions, to which we now turn, coefficient of this variable is always
are whether the relationship between negative, which indicates that, after con-
turnover and performance is statistically trolling for performance, increased share
significant and whether the sensitivity of ownership by the board reduces the likeli-
turnover to performance is greater follow- hood that the CEO will depart his position.
ing Cadbury.To answer those questions and We now turn to the effect of Cadbury on
to control for other factors that may influ- CEO turnover and the effect of Cadbury
ence managerial turnover, we estimate logit on the relationship between CEO turnover
regressions with pooled time series, cross and corporate performance. The five
section data. Initially, we estimate regres- regressions in each panel explore that ques-
sions in which the dependent variable is 1 if tion from different perspectives. The first
a firm experiences CEO turnover during a regression in each panel is estimated for
calendar year and 0 otherwise. We estimate the full sample of firms and includes an
separate regressions for all turnover and indicator variable (Dum for 1993–1996)
CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 357
which takes a value of 0 for all observations compliance set. Thus, publication of the
before January 1993 (the pre-Cadbury Code of Best Practice did not have an
period) and a value of 1 for all observations impact, per se, on the rate of turnover
after that date (the post-Cadbury period) among top U.K. executives; rather, the
along with a performance variable, either effect was concentrated among those firms
logIAROA or logISAR, and the four that altered their board structures to com-
control variables. In each panel, in the first ply with the Code. This is not to say that the
regression, the coefficient of the indicator rate of turnover among top executives in
variable Dum for 1993–1996 is positive firms that were always-in-compliance was
with p-values ranging from 0.04 to 0.11. “too low” either before or after Cadbury.
Thus, even after controlling for corporate The data only show that the rate of
performance, turnover is higher in the turnover for these firms did not change
post-Cadbury period. However, as we between the pre- and post-Cadbury periods.
observed in Table 13.4, increased turnover In comparison, the rate of turnover
appears to be attributable to the set of firms increased significantly among firms that
that came into compliance with the came into compliance with the Cadbury
Cadbury Committees’ recommendations (the recommendations during the period of
adopted-Cadbury set) as opposed to those this study.
firms that were always in compliance. To determine whether the increase in
To determine whether the Cadbury/ turnover is correlated with performance, we
turnover relationship is due to a general estimate a regression with only the
phenomenon affecting all firms or whether adopted-Cadbury set of firms that
it is due specifically to a change in board includes the adopted Cadbury dummy
structures traceable to the Cadbury (Dum-for-Adopt) and the adopted Cadbury
recommendations, we next estimate the dummy interacted with our measures
regressions separately for the always- of performance (either Dum-for-Adopt
in-compliance set of firms and for the logIAROA or Dum-for-Adopt logISAR)
adopted-Cadbury set.The only difference in along with our measures of performance
the regressions is that for the adopted- (either logIAROA or logISAR) and our
Cadbury set, the indicator variable four control variables. These are the key
(Dum-for-Adopt) takes on a value of 0 in regressions of our analysis and are given as
all years prior to the year in which the firm the fourth regression in each panel.
came into compliance with the Code and a The coefficient of the interaction
value of 1 for all subsequent years. These variable indicates whether the increase in
are the second and third regressions in turnover among firms that adopted
each panel. Cadbury is randomly distributed across
For the always-in-compliance set, the those firms or is concentrated among the
coefficient of the Cadbury dummy variable poorest performing firms. In each regres-
(Dum for 1993–1996) is always positive, sion, the coefficient of the interaction vari-
but the p-values range from 0.79 to 0.92. able is negative with p-values ranging from
Thus, publication of the Cadbury Report 0.02 to 0.07. Additionally, the coefficient
had a trivial impact, if any, on the rate of of the adopted Cadbury dummy (Dum-for-
turnover among CEOs in firms that were Adopt) is reduced by 60 percent and now
already in compliance with the key provi- has p-values ranging from 0.60 to 0.77.
sions of the Code. For the adopted-Cadbury These results indicate that the increase in
set, the coefficient of the indicator variable CEO turnover is not random; rather it is
Dum-for-Adopt is always positive with (inversely) correlated with performance:
p-values ranging from 0.06 to 0.08. After controlling for performance, the like-
Additionally, the magnitude of the coeffi- lihood that the CEO will depart his position
cient is at least four times the magnitude of is greater once a poorly performing
the coefficient of the Cadbury dummy firm comes into compliance with the key
(Dum for 1993–1996) for the always-in- provisions of the Code. The answer to the
Table 13.4 Logit Regressions of CEO Turnover on IAROA and Status of Cadbury Compliance, 1989 through 1996
CEO turnover for a random sample of 460 publicly traded U.K. industrial firms in two four-year periods during the interval 1989 through 1996. IAROA
is calculated as earnings before interest, taxes, and depreciation divided by the total book value of assets less the median performance of firms in the same
FTIC grouping. Three years of IAROA are averaged. CEO turnover is classified as normal or forced by examining news articles in the Extel Weekly News
Summaries, the Financial Times, and McCarthy’s News Information Service. The sample firms are classified into three sets based on whether they were
(a) always in compliance with the Cadbury recommendations, (b) never in compliance with the Cadbury recommendations, or (c) adopted Cadbury
recommendations. Sample firms in (a) and (b) are analyzed over two four-year periods, pre- and postpublication of the Cadbury Report (1989 through
1992 and 1993 through 1996). Sample firms in (c) are analyzed over two four-year periods, pre- and postadoption of the Cadbury recommendations (y 4
through y 1 and y 1 through y 4). Accounting information and share prices are from Datastream.The dependent variable equals one when turnover
occurs. Dum for 1993–1996 equals one for the period 1993 through 1996. Dum-for-adopt equals one for the period following the adoption of the key
recommendations of the Cadbury Report. The interactive dummy is Dum-for-adopt multiplied by logIAROA. P-values are in parentheses.
Panel A: Logit Regressions of All CEO Turnover on Log IAROA and Cadbury Status
Intercept 1.866 (0.08) 1.849 (0.09) 2.570 (0.00) 2.799 (0.00) 2.583 (0.00)
Performance variable
Log IAROA 2.034 (0.02) 1.859 (0.10) 3.180 (0.00) 3.228 (0.00) 3.019 (0.00)
Cadbury variable
Dum for 1993–1996 0.457 (0.11) 0.055 (0.92)
Dum-for-adopt 0.593 (0.06) 0.148 (0.66) 0.112 (0.72)
Dum-for-adopt logIAROA 0.739 (0.02) 0.038 (0.96)
Board variables
Prop outsiders 0.331 (0.30)
Prop outsiders logIAROA 0.566 (0.08)
358 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Panel A: Logit Regressions of All CEO Turnover on Log ISAR and Cadbury Status
Intercept 3.194 (0.00) 2.925 (0.00) 2.639 (0.00) 2.612 (0.00) 2.495 (0.00)
Performance variable
LogISAR 0.019 (0.00) 0.012 (0.03) 0.022 (0.00) 0.023 (0.00) 0.019 (0.00)
Cadbury variables
Dum for 1993–1996 0.519 (0.09) 0.066 (0.87)
Dum-for-adopt 0.572 (0.08) 0.119 (0.77) 0.090 (0.85)
Dum-for-adopt logISAR 0.680 (0.05) 0.144 (0.60)
Board variables
Prop outsiders 0.262 (0.43)
Prop outsiders logISAR 0.573 (0.08)
360 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
Intercept
Performance variable 4.892 (0.00) 3.023 (0.00) 4.538 (0.00) 4.624 (0.00) 4.291 (0.00)
LogISAR 0.030 (0.00) 0.009 (0.07) 0.049 (0.00) 0.040 (0.00) 0.039 (0.00)
Cadbury variables
Dum for 1993–1996 0.598 (0.04) 0.050 (0.86)
Dum-for-adopt 0.538 (0.08) 0.227 (0.60) 0.030 (0.92)
Dum-for-adopt logISAR 0.590 (0.07) 0.134 (0.65)
Board variables
Prop outsiders 0.272 (0.39)
Prop outsiders logISAR 0.564 (0.08)
Dum for single CEO/COB 0.060 (0.83)
Dum for single CEO/COB logISAR 0.039 (0.93)
Board size 0.042 (0.34)
Board size logISAR 0.045 (0.10)
Control variables
Board share ownership 0.921 (0.05) 0.925 (0.05) 0.807 (0.09) 0.840 (0.07) 0.763 (0.13)
Institutional share ownership 1.040 (0.21) 1.100 (0.19) 0.638 (0.58) 0.635 (0.58) 0.567 (0.67)
Block holders 0.078 (0.25) 0.101 (0.07) 0.044 (0.48) 0.048 (0.47) 0.043 (0.47)
Log assets 0.119 (0.07) 0.142 (0.03) 0.066 (0.26) 0.061 (0.27) 0.049 (0.31)
Observations 3,680 1,200 2,304 2,304 2,304
Log-likelihood 629.65 555.36 581.41 584.07 588.82
Chi-square 117.41 (0.00) 48.37 (0.00) 50.21 (0.00) 50.46 (0.00) 50.61 (0.00)
CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 361
362 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
question of whether the “right” managers p-values that range from 0.07 to 0.08. In
are leaving the firms appears to be yes, contrast, in none of the regressions does
assuming, of course, that our measures of the coefficient of the dummy for the
performance properly identify the right CEO/COB or the coefficient of the interac-
managers. tion of this variable with our measures of
Thus far, we have employed an indicator performance begin to approach statistical
variable to capture the key provisions of the significance (p-values range from 0.69
Code of Best Practice. A further question to 0.93).
is: Which of the key provisions is responsi- Apparently, the increased sensitivity of
ble for the increased sensitivity of turnover turnover to corporate performance for the
to corporate performance? To address that adopted-Cadbury set of firms (and the
question, we estimate a final regression contemporaneous loss in significance of
with the adopted-Cadbury set of firms in the interaction of Dum-for-Adopt with
which we include annual observations on performance) is attributable to the increase
the fraction of outside directors (Prop in the fraction of outside directors.
Outsiders), an interaction between the Splitting the responsibilities of the CEO
fraction of outsiders and our measures of and COB between two individuals appears
corporate performance (either Prop to have had no effect on the rate of CEO
Outsiders logIAROA or Prop Outsiders turnover.
logISAR), an indicator variable to
identify observations in which the positions
of CEO and COB are held by a single indi- D. Spurious correlation?
vidual (equal to 1) or by two individuals A question that may arise is whether the
(equal to 0) and an interaction between correlation between management turnover
this indicator variable and our measures of and corporate performance interacted with
corporate performance (either Dum for Cadbury compliance is spurious. More
Single CEO/COB logIAROA or Dum for specifically, is it possible that both
Single CEO/COB logISAR).These variables turnover and Cadbury compliance are
are designed to capture the changes caused by poor performance, perhaps
brought about by the Code of Best because poorly performing firms adopt
Practice. Because adoption of the Code led Cadbury to placate shareholders and,
to a general increase in board size, we also concurrently, dismiss top managers—a
include the number of directors and an change in management that would have
interaction between the number of direc- occurred even in the absence of Cadbury?
tors and our measure of performance. Several analyses seem to indicate that this
These regressions, which also include a is not the case.
performance measure, Dum-for-Adopt, and First, by construction, for the adopted-
the four control variables, are shown as the Cadbury set of firms, all post-adoption
fifth regression in each panel. CEO turnover follows compliance with the
According to the regressions, when the Code. This occurs because we use year-end
board composition and CEO/COB variables data to determine whether a firm is in com-
are included, the coefficients of the interaction pliance. Only after the year-end in which
of the Dum-for-Adopt and our measures of the firm becomes compliant with the Code
performance are not significant (p-values do we consider turnover to be postadoption.
range from 0.60 to 0.96). Additionally, the Thus, all post-Cadbury CEO turnover is
coefficient of the fraction of outsiders on postadoption. Related to this point, most
the board is positive, albeit not significant, postadoption turnover does not follow
in each regression (p-values range from closely after Cadbury compliance. For
0.30 to 0.43). More interestingly, the example, for the 58 instances of postadoption
coefficients of the interaction between the forced CEO turnover, 18 occur within
fraction of outsiders and our measures 12 months after the year-end of adoption,
of performance are always negative, with 23 occur in months 13 through 24, and
CORPORATE PERFORMANCE AND MANAGEMENT TURNOVER 363
the FMA (Orlando). McConnell acknowledges The Corporate Register, 1989 to 1996, 1998
financial support from the Center for (Hemmington Scott Publications, London).
International Business Education and Research Cotter, James F., Anil Shivdasani, and Marc
(CIBER) at Purdue University. Dahya acknowl- Zenner, 1997, Do independent directors
edges financial support received from the enhance target shareholder wealth during
Nuffield Foundation. Travlos acknowledges
tender offers? Journal of Financial
financial support received from the Kitty
Kyriacopoulos Chair in Finance. Economics 43, 195–218.
1 The report also recommended: (a) full disclo- Coughlan, Anne T., and Ronald M. Schmidt,
sure of the pay of the chairman and the highest 1985, Executive compensation, management
paid director; (b) shareholders’ approval on turnover, and firm performance, Journal of
executive directors contracts exceeding three Accounting and Economics 7, 43–66.
years; (c) executive directors pay be set by a Denis, David J., and Diane K. Denis, 1995,
board subcommittee composed primarily of Performance changes following top manage-
outsiders; and (d) directors establish a subcom- ment dismissals, Journal of Finance 50,
mittee of the board, comprised mainly of outside 1029–1057.
directors, to report on the effectiveness of the
Denis, David J., and Atulya Sarin, 1999,
company’s system of internal control.
2 We also estimated the regressions with Ownership and board structures in publicly
market model excess returns and CAPM excess traded corporations, Journal of Financial
returns as our measure of performance. The Economics 52, 187–224.
p-values of the coefficients are essentially Franks, Julian, and Colin Mayer, 1996, Hostile
unchanged. takeovers and the correction of managerial
3 As a benchmark, we calculated the implied failure, Journal of Financial Economics 40,
instances of total CEO turnover during years 163–181.
y 1 through y 4 to be 96 versus actual Franks, Julian, Colin Mayer, and Luc
turnover of 98 and forced turnover to be 56 Renneboog, 2000, Who disciplines managers
versus actual forced turnover of 58.
in poorly performing companies? Working
4 The results of our analyses of the top team
excluding the CEO are available from the authors. paper, London Business School.
Gilson, Stuart, 1989, Management turnover and
financial distress, Journal of Financial
Economics 25, 241–262.
REFERENCES Green, Sir Owen, 1994, Pall Mall Lecture on
U.K. Corporate Governance, February 24.
Agrawal, Anup, and Charles R. Knoeber, 1996, Hermalin, Benjamin E., and Michael S.
Firm performance and mechanisms to control Weisbach, 1998, Endogenously chosen
agency problems between managers and boards of directors and their monitoring of
shareholders, Journal of Financial and the CEO, American Economic Review 88,
Quantitative Analysis 31, 377–397. 96–118.
Bhagat, Sanjai, and Bernard S. Black, 1999, Huson, Mark R., Robert Parrino, and Laura T.
The uncertain relationship between board Starks, 1998, Internal monitoring
composition and firm performance, Business mechanisms and CEO turnover: A long term
Lawyer 54, 921–964. perspective, Unpublished manuscript,
Bhagat, Sanjai, and Bernard S. Black, 2000, University of Alberta.
Board independence and long-term perform- Kang, Jun-Koo, and Anil Shivdasani, 1995, Firm
ance, Working paper, University of Colorado performance, corporate governance, and top
and Columbia Law School. executive turnover in Japan, Journal of
Byrd, John W., and Kent A. Hickman, 1992, Do Financial Economics 38, 29–58.
outside directors monitor managers? Kaplan, Steven N., 1994, Top executives,
Evidence from tender offer bids, Journal of turnover and firm performance in Germany,
Financial Economics 32, 195–221. Journal of Law, Economics and Organization
Cadbury committee draft orders mixed news 10, 142–159.
for shareholders, 1992, Financial Times, Kaplan, Steven N., and David Reishus, 1990,
June 2, p. 16. Outside directorships and corporate
366 CORPORATE GOVERNANCE, UNDERPERFORMANCE AND MANAGEMENT TURNOVER
performance, Journal of Financial Economics Report on the Financial Aspects of Corporate
27, 389–410. Governance (With the Code of Best
Kini, Omesh, William A. Kracaw, and Shehzad Practice), 1992 (Gee and Co. Ltd., London).
Mian, 1995, Corporate takeovers, firm Rosenstein, Stuart, and Jeffrey G. Wyatt, 1990,
performance, and board composition, Journal Outside directors, board independence, and
of Corporate Finance 1, 383–412. shareholder wealth, Journal of Financial
Klein, April, and James Rosenfeld, 1988, Economics 26, 175–191.
Targeted share repurchases and top Self-regulation seen as the way forward, 1992,
management changes, Journal of Financial The Financial Times, May 28, p. 17.
Economics 20, 493–506. Shivdasani, Anil, 1993, Board composition,
Martin, Kenneth J., and John J. McConnell, ownership structure, and hostile takeovers,
1991, Corporate performance, corporate Journal of Accounting and Economics 16,
takeovers, and top management turnover, 167–198.
Journal of Finance 46, 671–687. Stock Exchange Yearbook, 1989–1996
Mikkelson, Wayne, and Megan Partch, 1997, (London Stock Exchange, London).
The decline of takeovers and disciplinary Warner, Jerold B., Ross L. Watts, and Karen H.
management turnover, Journal of Financial Wruck, 1988, Stock prices and top manage-
Economics 44, 205–228. ment changes, Journal of Financial
Murphy, Kevin J., and Jerold L. Zimmerman, Economics 20, 461–492.
1993, Financial performance surrounding Weisbach, Michael S., 1988, Outside directors
CEO turnover, Journal of Accounting and and CEO turnover, Journal of Financial
Economics 16, 273–315. Economics 20, 431–460.
Part 4
Directors’ remuneration
INTRODUCTION
accounting returns and stock returns are higher after the plan is adopted. Thus, for the
sample of firms, the required increases in the level of managerial equity ownership result
in improvements in firm performance. This study highlights the importance of executive
stock ownership as an effective corporate governance tool.
REFERENCE
Bebchuk L. and Fried J. (2004), Pay without performance, Cambridge, Mass: Harvard
University Press.
Chapter 14
ABSTRACT
We document differences in CEO pay and incentives in the United States and the United
Kingdom for 1997. After controlling for size, sector and other firm and executive characteristics,
CEOs in the US earn 45% higher cash compensation and 190% higher total compensation.
The calculated effective ownership percentage in the US implies that the median CEO receives
1.48% of any increase in shareholder wealth compared to 0.25% in the UK. The differences,
can be largely attributed to greater share option awards in the US arising from institutional
and cultural differences between the two countries.
CORPORATE GOVERNANCE PRACTICES IN THE Although CEO pay levels in the United
United Kingdom received increased attention Kingdom have grown in recent years, they
in the 1990s, culminating in influential remain far behind pay levels enjoyed by
reports issued by the Cadbury (1992), CEOs in the United States. The interna-
Greenbury (1995) and Hampel (1998) com- tional pay gap is especially pronounced
mittees. Among other recommendations, the after including gains realised from exercis-
reports outlined a best-practice framework ing share options. Chief executives in the
for setting executive pay, and significantly 500 largest UK companies in aggregate
expanded disclosure rules for UK executive made £330 million (or £660,000 each) in
compensation. The Greenbury and Hampel 1997, including £74 million from exercis-
reports were, in part, a response to a ing options. In contrast, the top 500 US
growing controversy over chief executive CEOs made in aggregate £3.2 billion (or
officer (CEO) pay levels triggered when £6.3 million each), including £2.0 billion
executives in several recently privatised from option exercises.1 Indeed, Disney’s
electric utilities exercised share options Michael Eisner, dubbed by pay-critic Graef
worth millions of pounds. However, in spite Crystal (1991) as the ‘Prince of Pay’,
of these reports, CEO pay levels rose more exercised options worth £348 million in
than 18% in 1997 alone, even as public- December 1997, thus single-handedly out-
sector workers were being asked to accept earning the aggregate paycheques of the
raises of less than 3% (Buckingham and top 500 CEOs in the United Kingdom.
Cowe, 1998 a, b).The continuing controversy, British Sky Broadcasting’s Sam Chisolm, the
coupled with enhanced data availability highest-paid UK executive, is a mere pauper
through the new disclosure requirements, by American standards: his £6.8 million
has sparked considerable academic interest pay package would only rank as the 97th
in UK executive pay practices. highest among US chief executives.
372 DIRECTORS’ REMUNERATION
These anecdotal comparisons, while driven compensation from 1989–97, and on the
by option gains in the robust US stock prevalence of stock option plans from
market, hint at important differences in 1979–97. We show that, although cash
CEO pay practices in the United Kingdom compensation has been growing at about
and the United States. The purpose of this the same rate in both countries since the
article is to provide a comprehensive mid-1990s, the prevalence of option plans
comparison of pay practices in the two has been growing in the United States while
countries, and to generate stylised facts to declining in the United Kingdom.
stimulate future research. Existing inter- Section 3 explores stock-based financial
national comparisons of pay practices have incentives for CEOs in the two countries.
typically relied on non-comparable survey CEO wealth is linked directly to company
data, or have focused on narrow definitions share-price performance through their
of compensation that usually exclude the share holdings, their option holdings, and
grant-date value of share options.2 In through shares awarded through long-term
contrast, our results are based on compa- incentive plans (LTIPs). We show that
rable and complete measures of CEO pay these various holdings can be aggregated to
and stock-based incentives, utilising detailed form an ‘effective’ ownership percentage (or,
data made available through enhanced dis- following Jensen and Murphy (1990 a,b),
closure requirements in the United States the ‘Pay-Performance Sensitivity’). Similar
and (more recently) the United Kingdom. to our analysis of CEO pay levels, we
Our use of micro data allows us to analyse compute the effective ownership percentage
differences in compensation and incentives for each CEO and analyse how it varies
while controlling for factors such as com- across countries and with different firm
pany size, industry, human capital, growth and executive characteristics. We find that
opportunities, and performance. American CEOs, on average, own much
We begin in Section 1 by offering an larger fractions of their firms’ stock than
introduction to executive compensation, do British CEOs. For example, the median
aimed at academic economists new to the holding for US CEOs is 0.29%, while the
area. In this section, we identify available median holding for UK CEOs is only 0.05%.
data sources for CEO pay in the two The median effective ownership (including
countries, describe our sample, and discuss options and LTIP grants) is more disparate:
how to measure and value share options and 1.48% for US CEOs vs. only 0.25% for
other pay components.3 Section 2 analyses UK CEOs.
the level and structure of executive compen- Section 3 also explores how CEO cash
sation, based primarily on data from 510 UK compensation varies with share-price
and 1,666 US corporations. We show that, performance. We show that the elasticity
while company size is an important deter- of cash compensation to performance is
minant of pay in both countries, the rewards higher in the United States than in the
for scale are more pronounced in the United Kingdom for all industries, although
United States than in the United Kingdom. the difference is only statistically significant
We document that share option grants, for the financial services industry. In
valued at grant date, comprise a fairly addition, we examine the relation between
small percentage of total pay for the CEO turnover and company performance.
typical British CEO, but are much more We find that CEO turnover is negatively
important for American CEOs. In addition, correlated with shareholder returns in both
we document that American CEOs indeed the United Kingdom and United States,
out-earn their British counterpart, earning indicating that CEOs in both countries are
45% more in cash pay and 190% more in more likely to lose their jobs following poor
total pay, even after controlling for size, performance. The cross-country difference
industry, growth opportunities, CEO human in the turnover-performance relationship is
capital, and other observable characteristics. not statistically significant.
Also, Section 2 offers some cross-country Section 4 considers a variety of explana-
time-series evidence on the levels of cash tions for the observed Anglo-American
CEO PAY IN THE US AND UK 373
differences in compensation and incentives, in the United Kingdom, and the proxy
including agency theory, taxes, and culture. statement in the United States.6
Our objective is not to reconcile completely Although we offer some longitudinal
the differences in pay practices, but rather to comparisons, our analysis is based primarily
identify potential explanations as opportu- on 1997 fiscal-year data, since this was the
nities for future research. We argue that first year covered by the new UK disclosure
traditional agency-theoretic considerations requirements. The UK data analysed in this
offer little insight in explaining the differ- paper are drawn directly from the annual
ences, unless US and UK executives differ reports for the 510 largest companies
systematically in their ability, productivity, (ranked by market capitalisation). The
or risk aversion. We document that, while pay and ownership data are matched to
personal income tax rates and rules in Datastream data on company size, industry,
the United States and United Kingdom are and performance.Together, these companies
generally quite similar, there are differences account for virtually all (98%) of the
in corporate tax rules that may explain at market capitalisation of the entire UK stock
least some of the observed differences in market. The fiscal 1997 US compensation
compensation and incentives. Finally, we and company data are extracted from
consider a variety of economic, political, Standard and Poor’s (S&P’s) Compustat’s
and cultural factors that help explain why ‘ExecuComp’ database, which includes
share option compensation has increased proxy-statement data for 1,666 top execu-
dramatically in the United States, but not in tives in the S&P 500, the S&P Mid-Cap
the United Kingdom. Section 5 summarises 400, the S&P Small-Cap 600, and other
our results, and explores implications supplemental S&P indices. For each country
of our results to broader multi-country and company, we identified the CEO (or
comparisons of international pay practices. most senior executive officer), and collected
information on share ownership, current
and prior option grants, salaries, annual
1. AN INTRODUCTION TO bonuses, benefits, and LTIP cash and
EXECUTIVE COMPENSATION share awards.
The largest US companies are consider-
ably larger than the largest UK companies,
1.1. Data sources and our combination of large, mid, and
The United States, United Kingdom, and small-cap US companies is meant to pro-
Canada are currently the only countries vide a distribution of US firms similar to
that require detailed disclosure on the the UK distribution. Nonetheless, our sam-
compensation practices for individual top ple of US firms remains somewhat larger,
corporate executives.4 Disclosure rules with mean (median) 1997 market capitali-
for US executives were standardised and sation of £3.4 billion (£790 million) in the
expanded in 1992 by the US Securities and United States, compared to £2.2 billion
Exchange Commission (SEC), and require (£480 million) in the United Kingdom. Our
details on share ownership, share options, results below include company size controls
and all components of compensation for to adjust for systematic differences in size.
the top five corporate executives. Disclosure
rules in the United Kingdom were signifi-
cantly expanded in recent years following
1.2. Measuring and valuing the
the Greenbury (1995) and Hampel (1998) components of pay
reports, and require disclosure of data com- Compensation arrangements in both the
parable to those available for US executives United Kingdom and the United States
(including previously unavailable details on contain the same basic components. CEOs
share option grants and holdings).5 Although in both countries receive base salaries and
there are a variety of ‘secondary sources’ are eligible to receive annual bonuses paid
for compensation data in the two countries, based on accounting performance. CEOs in
the primary data source is the annual report both countries also typically receive share
374 DIRECTORS’ REMUNERATION
options, which are rights to purchase shares of the prior 48 monthly observations on
of stock at a pre-specified ‘exercise’ price cash dividend per share.8
for a pre-specified term. CEOs also often In spite of its prevalence in both practice
participate in long-term incentive plans and academia, there are many drawbacks
(LTIPs). In the United Kingdom, LTIPs are to using the Black-Scholes formula for
typically grants of shares of stock that calculating the value of an executive share
become ‘vested’ (i.e., ownership is trans- option. First, the Black-Scholes value is,
ferred to the CEO) only upon attainment of at best, a measure of the company’s
certain performance objectives. LTIPs in opportunity cost of granting the option,
the United States take two primary forms: and will typically overstate the value to
(1) ‘restricted stock’ grants that vest with the executive-recipient (Hall and Murphy,
the passage of time (but not with perform- 2000). Second, executive share options are
ance criteria); and (2) multi-year bonus subject to forfeiture if the executive leaves
plans typically based on rolling-average the firm prior to vesting; this probability
three or five-year cumulative accounting of forfeiture reduces the cost of granting
performance. the option and thus implies that the Black-
We define total compensation as the sum Scholes formula overstates option values.
of base salary, annual bonus, LTIP awards, Third, the Black-Scholes formula assumes
and share options valued at grant date. that options can only be exercised at the
We measure LTIP share grants at the face expiration date, but executive options can
value of the shares on the grant date, and be exercised immediately upon vesting,
impose 20% discounts for performance- which typically occurs relatively early in the
contingent UK grants. LTIP cash awards option’s term.9 Finally, following recom-
are valued as the amount actually paid mendations in the Greenbury (1995) report,
during the fiscal year. In valuing share share options granted in the United
options, we follow the approach used by Kingdom typically vest only upon attain-
both practitioners and academic researchers ment of some performance criteria,
by measuring the grant-date expected value often based on earnings-per-share growth.
using the Black and Scholes (1973) formula, Although the existence of performance
adjusted for continuously paid dividends: criteria will naturally reduce the company’s
cost of granting an option, the expected
Option Value Peln(1d)TN(z) discount is fairly modest, because the crite-
ria are seldom binding.10 Moreover, to the
XeLn(1r)TN(z T), (1) extent that the performance criteria are
where P is the grant-date share price, X is correlated with share prices, the criteria
the exercise price, T is the time remaining will be binding only when the intrinsic value
until expiration, d is the annualised dividend of an unrestricted share option is low.
yield, is the stock-price volatility, r is the Subject to these caveats, we present Black-
risk-free discount rate, N( ) is the cumulative Scholes values of stock options in our
normal distribution function, and analyses below.
We measure the risk-free rates for the two 2.1. Summary statistics
countries as the average yield on 7-year
UK and US Treasury bills.7 Volatilities are Table 14.1 provides summary statistics for
defined as the standard deviation of monthly the level and composition of fiscal 1997
continuously compounded returns over the CEO pay, by company size, and industry,
prior 48 months, multiplied by 12. and country.Total pay is defined as the sum
Dividend yields are computed as the average of salaries, bonuses, benefits, other cash
CEO PAY IN THE US AND UK 375
pay, grant-date values of share options, and with 1997 revenues in excess of £1,500
grant-date value of LTIP shares (dis- million is £811,000, far larger than the
counted, where appropriate, for perform- £287,000 median pay for companies with
ance contingencies). Dollar-denominated revenues below £200 million. Median and
data for each US executive are converted average total pay is somewhat less in
to UK pounds using the average exchange utilities than in other industries. The
rate during the company’s fiscal year; this bottom panel shows that American CEOs
rate varied between 1.61 $/£ and 1.65 $/£ earn substantially more than their British
during our sample period. counterparts, for every size and industry
As reported in the top panel of Table 14.1, group. The average total compensation for
the average total compensation for the 510 the 1,666 US CEOs is £3.6 million, or
UK CEOs is £589,000, while the median 500% more than the average pay for UK
pay is £414,000. Total pay increases with executives. Similarly, the US median pay of
firm size: the median pay for companies £1.5 million is 260% more than the
Table 14.1 Summary statistics for 1997 CEO total compensation, by company size and industry
United Kingdom
All companies 510 589 414 59 18 10 9 5
By firm sales (millions)
Less than £200 152 452 287 64 17 10 4 5
£200 to £500 119 403 335 61 19 8 6 6
£500 to £1,500 116 601 507 54 20 10 12 4
Above £1,500 123 927 811 55 16 10 15 4
By industry
Mining/manufacturing 217 564 436 59 17 9 9 5
Financial services 84 559 411 60 22 6 7 4
Utilities 19 448 382 58 15 6 14 8
Other 190 645 397 58 17 11 8 5
United States
All companies 1,666 3,565 1,508 29 17 42 4 8
By firm sales (millions)
Less than £200 339 1,166 686 38 14 43 1 4
£200 to £500 379 1,833 926 36 18 36 3 7
£500 to £1,500 458 3,038 1,604 28 18 40 5 9
Above £1,500 490 7,056 3,552 20 17 48 5 10
By industry
Mining/manufacturing 842 3,388 1,540 28 17 43 3 8
Financial services 198 6,277 2,787 19 20 47 5 8
Utilities 120 1,333 707 43 15 23 6 13
Other 506 3,326 1,438 32 16 43 3 6
Note: UK data from the largest companies in fiscal 1997, ranked by market capitalisation. US data include
firms in the S&P 500, the S&P MidCap 400, the S&P SmallCap 600, and companies in S&P supplemental
indices. Revenues for financial firms defined as net interest income (banks) and total income (insurance
companies). Total compensation defined as the sum of salaries, bonuses, benefits, share options (valued on date
of grant using the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for
performance-contingent awards), and other compensation. US dollar-denominated data are converted to
UK pounds using the average $/£ exchange rate during the fiscal year.
376 DIRECTORS’ REMUNERATION
Table 14.2 Summary statistics for components of CEO pay, by company size and industry
United Kingdom
All Companies 240 81 91 50 69 32 161
By firm sales (millions)
Less than £200 175 76 59 40 70 14 58
£200 to £500 200 82 69 44 45 25 94
£500 to £1,500 264 87 104 59 73 44 148
Above £1,500 410 80 146 59 108 50 294
By industry
Mining/manufacturing 254 84 90 53 67 34 165
Financial services 222 83 115 45 50 29 132
Utilities 240 84 86 47 43 58 120
Other 221 77 83 48 110 29 177
United States
All companies 317 83 270 72 1,142 19 325
By firm sales (millions)
Less than £200 195 74 99 69 592 6 107
£200 to £500 256 82 180 64 787 15 180
£500 to £1,500 335 85 299 71 1,053 22 315
Above £1,500 487 89 518 80 2,505 30 579
By industry
Mining/manufacturing 318 85 280 74 1,111 18 318
Financial services 395 92 437 81 1,713 31 611
Utilities 291 81 157 53 392 25 177
Other 301 78 239 68 1,232 15 336
Note: Median data for bonuses, options, and LTIP represent the median value of award/grant (in £000s) for
the subsample of CEOs actually receiving awards /grants during the 1997 fiscal year. Revenues for financial
firms defined as net interest income (banks) and total income (insurance companies). Share options (valued on
date of grant using the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for
performance-contingent awards), and other compensation. US dollar-denominated data are converted to
UK pounds using the average $/£ exchange rate during the fiscal year.
CEO PAY IN THE US AND UK 377
Kingdom salary of £240,000; median the value of the grant (for those receiving
salaries are higher in the United States for grants) is higher in the US.
each size and industry group.The percentage
of CEOs receiving bonuses is roughly the
2.2. Time-series comparisons
same in the two countries (81% in the UK
compared to 83% in the US). However, Although longitudinal comparisons of total
conditional on receiving a bonus, US compensation are not possible because of
bonuses are much higher: the median bonus UK disclosure requirements prior to 1997,
paid in the United States of £270,000 is time-series data on cash compensation
triple the median bonus paid in the United (salaries and bonuses) are available for
Kingdom. American CEOs are more likely the sample companies. Figure 14.1 shows the
to receive option grants than their UK median cash compensation received from
counterparts (72% in the United States 1989–97 for UK CEOs and US CEOs in the
versus only 50% in the United Kingdom). S&P 500, and from 1992–97 for US CEOs
In addition, they are likely to receive much in the MidCap 400 and SmallCap 600.The
larger grants: the median option grant in figure includes data from firms in our 1997
the United States (for CEOs receiving sample of 510 UK and 1,666 US firms.11
options) of £1,142,000 is nearly twenty As shown in the figure, the median cash pay
times the median grant value for UK CEOs for UK CEOs has grown from £158,000 in
(who receive only £69,000). Finally, 1989 to £340,000 in 1997 (representing
Table 14.2 shows that 32% of UK CEOs a 10% average annual growth). In the
receive LTIP share grants, while only 19% United States, median pay among S&P
of US CEOs receive similar grants. However, 500 CEOs has grown 6.4% annually, from
although fewer US executives receive LTIP £574,000 in 1989 to £945,000 in 1997.
share grants (primarily restricted stock), Since 1992, UK CEO cash compensation
1,000
US S&P 500
900
800
Cash compensation (£000s)
700
500
300
100
0
1989 1990 1991 1992 1993 1994 1995 1996 1997
has increased on average 11.2% per year, in Figure 14.2.The UK data are from Main
compared to growth rates of 10.7%, (1999), and are based on data provided
11.9%, and 10.2% for CEOs in the S&P by a large compensation consulting firm.
500, MidCap, and SmallCap companies, The US data for 1979–96 are from
respectively. the Conference Board ‘Top Executive
The results in Figure 14.1 suggest that cash Compensation’ reports, which cover pre-
compensation in UK firms is roughly ‘on dominately S&P 500 companies.13 The
par’ with median pay among CEOs in Small- data for ‘US Small and MidCap’ firms are
Cap US firms, but still lies significantly extracted from ExecuComp, and defined
below pay in larger US firms.12 Moreover, as the fraction of MidCap and SmallCap
as documented by Hall and Liebman companies in which the top five executives
(1998) and Murphy (1999), the value of hold any options during the year.
share options granted to US CEOs has As reported by Main (1999) and
grown much faster than their cash compen- replicated in Figure 14.2, option grants in
sation since the early 1990s. Indeed, as the United Kingdom grew dramatically in
indicated in Table 14.1, share options have popularity from the mid-1980s to the
emerged as the single largest component of early-1990s. In particular, in 1978 only
compensation for CEOs in the largest US 10% of UK companies offered options to
companies. Although comparable longitudi- their top executives, by 1983 over 30% of
nal data on share-option grants in the companies offered options, and by 1986
United Kingdom are not systematically nearly 100% offered options. However, the
available, time-series data on the preva- use of share options in the UK fell substan-
lence of option plans among UK and US tially in the mid-1990s; by 1997 only 68%
companies are available and are reported of companies offered options to their top
100
90 United States
80
Percentage of firms with option plans
US Small &
70
MidCap firms
60 United Kingdom
50
40
30
20
10
0
79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97
Table 14.3 Estimated elasticities of CEO compensation with respect to firm revenues
All companies 0.316 [0.24] 0.413 [0.28] 0.197* [0.31] 0.217* [0.26]
Mining & manufacturing 0.294 [0.20] 0.423 [0.35] 0.189* [0.31] 0.157* [0.23]
Financial services 0.342 [0.42] 0.465 [0.34] 0.177* [0.26] 0.205* [0.30]
Utilities 0.334 [0.36] 0.459 [0.30] 0.209 [0.17] 0.314 [0.32]
Other 0.329 [0.28] 0.366 [0.20] 0.270 [0.45] 0.296 [0.33]
Note: Elasticities computed from regressions of ln(Pay) on ln(Sales) for 510 UK and 1,666 US companies.
R2 in brackets. Revenues for financial firms defined as net interest income (banks) and total income (insurance
companies).
* indicates that the estimated elasticity in the United Kingdom is significantly different from the US elasticity
at the 1% level.
market capitalisation) or more narrowly chairman. Combining the CEO and chairman
defined industries.The dependent variable in position is expected to increase pay for
columns (1) and (2) is ln(Salary Bonus), two reasons. First, the combination is a
while the dependent variable in columns proxy for added responsibility and/or
(3) and (4) is ln(Total Pay).The coefficient ability of the incumbent CEO. Second, the
on the US dummy in column (1) of 0.3861 combination is a proxy for the CEO’s influ-
indicates that, after controlling for size and ence over the board and the remuneration
industry, CEOs in the United States earn committee. In the United Kingdom, such
approximately 47% more than their British combinations are the exception: only 18%
counterparts.15 Similarly, the US dummy in of the sample companies have CEOs who
column (3) of 1.142 indicates that total also serve as board chairmen. However, in
expected pay is 215% higher in the United the United States such combinations are
States after controlling for size and industry. the rule: nearly two-thirds (66%) of the
Existing empirical studies of executive sample have combined CEO/chairmen.
compensation have consistently documented Columns (2) and (4) of Table 14.4
that company size and industry are the two report coefficients for cash and total
most important factors determining levels compensation.The control for risk is insignif-
of CEO pay. However, researchers routinely icant in the cash compensation regression,
include additional controls for risk (because but significant for total compensation,
CEOs will demand higher expected pay while the control for investment options is
levels if compensation is risky), investment insignificant in both regressions.16 The
opportunities (because firms with growth human capital controls are significant, and
options need better managers), and human suggest that CEO age-earnings profiles for
capital. The regressions in columns (2) and cash (total) compensation turn downward
(4) of Table 14.4 include as additional after age 55 (age 43).17 The coefficients
explanatory variables the volatility of on the CEO/chairman dummy variables are
shareholder returns (as a proxy for risk), positive and significant, indicating that the
the book-to-market ratio (the inverse of a combination increases cash and total com-
proxy for investment opportunities), and pensation by 11% and 19%, respectively.
CEO age and age-squared (traditional Including the control for CEO/chairman
human capital variables to allow concave combinations reduces the estimated US
age-earnings profiles). In addition, the premium slightly, from 47% to 45% for
regressions include a dummy variable set salary and bonus, and from 215% to
to unity if the CEO and board chairman 190% for total compensation.
position are combined, and zero if the Several stylised facts emerge from
corporation has a separate non-executive the results of Tables 14.1–14.4 and
Table 14.4 Explanatory regressions for 1997 CEO compensation
Dependent variable
Note: t-statistics in parentheses. Total compensation defined as the sum of salaries, bonuses, benefits, share options (valued on date of grant using
the Black-Scholes formula), LTIP-related stock grants (valued at 80% of face value for performance-contingent awards), and other compensation.
Book to market defined as the book value of tangible equity divided by the end-of-year market capitalisation. US dollar-denominated data are
converted to UK pounds using the average $/£ exchange rate during the fiscal year.
CEO PAY IN THE US AND UK 381
382 DIRECTORS’ REMUNERATION
Figs 14.1–14.2. First, American chief shares held by the CEO declines with
executives are paid significantly more company size in the United Kingdom, but
than British chief executives, even after increases with company size in the United
controlling for company size, industry, and States. In the largest firms, US CEOs hold
other firm and executive characteristics. on average shares worth £145 million
Second, although CEO pay increases with (median £5 million), while UK CEOs hold
company size in both countries, the pay- shares worth only £3.4 million (median
size gradient is significantly higher in the £330,000).
United States than in the United Kingdom. Since agency costs arise when agents
Third, the US pay premium can be traced receive less than 100% of the value of
largely, but not entirely, to the prevalence output, the CEO’s share of ownership is a
and magnitude of share option grants, natural measure of the severity of the
which have increased in the United States agency problem. As reported in Table 14.5,
since the mid-1990s, while declining in the the average and median shareholdings for
United Kingdom. UK CEOs (expressed as a percentage of
outstanding shares) are 2.13% and 0.05%,
respectively, significantly smaller than the
3. THE RELATION BETWEEN average and median holdings among US
CEO PAY AND CORPORATE CEOs of 3.10% and 0.29%, respectively.
PERFORMANCE In the largest firms, US CEOs hold on
average 1.61% of their company’s shares
Empirical investigations of executive (median 0.09%), while UK CEOs hold only
incentives have typically focused on the 0.21% (median 0.01%). In fact, whether
relation between the CEO’s wealth and measured in pounds or percentages, or
the wealth of the company’s shareholders. at averages or medians, share ownership
CEO wealth is directly related to shareholder among US CEOs is substantially higher
wealth through the CEO’s holdings of shares, than ownership among UK CEOs for each
share options, and LTIP shares. In addition, size and industry group in Table 14.5.
CEO wealth can be indirectly related to Consequently, share ownership mitigates
shareholder wealth through accounting- more of the agency problem in the United
based bonuses (reflecting the correlation States than in the United Kingdom.
between accounting returns and share-price Holdings of unexercised share options
performance) and through year-to-year also provide a direct link between CEO and
adjustments in salary levels, target bonuses, shareholder wealth, because the value of
and option and LTIP grant sizes. Finally, the options held increases with increases
CEO wealth (and reputation) is affected by in the share price. Table 14.5 shows that
performance if CEOs lose their jobs when the average US executive holds options
their firms are performing poorly. to purchase 1.18% of the company’s
outstanding shares (median 0.72%), while
3.1. The direct relation between the average UK executive holds options to
CEO and shareholder wealth purchase only 0.24% (median 0.11%) of
his company’s shares. Again, pair-wise
The most obvious linkage between CEO and comparisons from the top and bottom
shareholder wealth comes from the CEO’s panels of Table 14.5 reveal that holdings of
holdings of company shares. As reported in previously granted but unexercised share
Table 14.5, the share ownership for UK options are substantially higher in the
CEOs is worth an average of £7 million, sub- United States than in the United Kingdom
stantially smaller than the average holdings for every size and industry group.
among US CEOs of over £60 million. The Finally, holdings of unvested LTIP shares
ownership distribution is significantly provide an additional direct link between
skewed: the median holdings for UK and US CEO and shareholder wealth. First, the value
executives are £460,000 and £3.3 million, of the underlying shares naturally increases
respectively. Interestingly, the value of penny-for-penny with increases in the share
CEO PAY IN THE US AND UK 383
Table 14.5 Summary statistics for stock-based CEO incentives, by company size and industry
United Kingdom
All companies 7.01 0.46 2.13 0.05 0.24 0.11 2.33 0.25
By firm sales (millions)
Less than £200 9.86 1.41 4.38 0.63 0.38 0.21 4.72 1.09
£200 to £500 9.50 0.70 2.55 0.14 0.24 0.14 2.75 0.42
£500 to £1,500 4.55 0.13 0.76 0.02 0.19 0.12 0.91 0.16
Above £1,500 3.40 0.33 0.21 0.01 0.10 0.04 0.31 0.05
By industry
Mining/ 6.01 0.26 1.91 0.04 0.24 0.14 2.11 0.23
manufacturing
Financial services 5.95 0.84 1.98 0.07 0.19 0.05 2.14 0.31
Utilities 0.16 0.11 0.01 0.00 0.03 0.02 0.05 0.02
Other 9.31 0.71 2.64 0.10 0.28 0.11 2.89 0.37
United States
All companies 60.37 3.26 3.10 0.29 1.18 0.72 4.18 1.48
By firm sales (millions)
Less than £200 16.63 2.07 5.32 0.96 1.84 1.37 6.98 3.65
£200 to £500 23.84 2.93 3.94 0.58 1.39 0.94 5.20 2.05
£500 to £1,500 32.25 2.64 2.36 0.25 1.12 0.70 3.43 1.26
Above £1,500 145.26 4.96 1.61 0.09 0.62 0.40 2.17 0.56
By industry
Mining/ 33.77 3.14 2.78 0.31 1.22 0.76 3.87 1.53
manufacturing
Financial services 127.46 7.58 2.25 0.31 0.98 0.52 3.17 1.01
Utilities 6.19 0.58 0.25 0.05 0.35 0.12 0.53 0.16
Other 91.34 4.22 4.63 0.38 1.40 0.95 5.96 2.01
Note: UK data from the largest companies in fiscal 1997, ranked by market capitalisation. US data include firms
in the S&P 500, the S&P MidCap 400, the S&P SmallCap 600, and companies in S&P supplemental indices.
Revenues for financial firms defined as net interest income (banks) and total income (insurance companies).
The Pay-Performance Sensitivity is defined as:
%Shares
of firm shares % of firm shares Delta % of firm shares Delta
held as
Options held as
Options
LTIP shares as
LTIP
where 0 Option Delta 1 is the share-weighted-average slope of the Black-Scholes function at the year-end
stock price, for options outstanding at the fiscal year end, and (LTIP Delta) 1.
Note: t-statistics (for OLS) and asymptotic t-statistics (for median regressions) in parentheses. Sales for financial firms defined as net interest
income (banks) and total income (insurance companies).
See the footnote on the prior table for our definition of the Pay-Performance Sensitivity.
CEO PAY IN THE US AND UK 385
386 DIRECTORS’ REMUNERATION
company size, and industry. The average and sensitivities are substantially higher in the
median pay-performance sensitivities for US United States, even after including our
CEOs are 4.18% and 1.48%, respectively, control variables.
significantly higher that the average (2.33%)
and median (0.25%) sensitivities for UK 3.2. The indirect relation between
CEOs. In the largest firms, the median US CEO and shareholder wealth
CEO has a sensitivity, or effective ownership,
of 0.56%, more than ten times the effective In addition to the direct relation through
ownership of their British counterparts. ownership, LTIP shares, and options,
Table 14.6 presents coefficients from CEO wealth is indirectly related to
OLS and median regressions where the company stock-price performance through
dependent variable is the stock-based pay- performance-based bonuses, raises, and
performance sensitivity from (5). The coef- LTIP and option grant sizes. The CEO pay
ficient on the US dummy in column (1) of literature has yet to reach a consensus on
2.467 indicates that the average effective the appropriate methodologies and metrics
ownership is 2.5 percentage points higher in to use in evaluating the indirect relation
the United States than the United between CEO pay and company stock-price
Kingdom, after controlling for company performance.22 However, the most common
size and industry. The regression in column approach has involved estimating some
(2) includes controls for risk, investment variant of the following first-difference
opportunities, and human capital intro- regression:
duced in Table 14.4: the volatility of
returns, the book-to-market ratio, CEO age ln(Salary Bonus)it
and age-squared, and a dummy variable ln(Shareholder Value)it , (6)
indicating that the CEO and board chair-
man position are combined. The volatility where the change in shareholder value,
and book-to-market variables are insignifi- ln (Shareholder Value)t ignores share
cant, while the age variables are significant issues or repurchases and therefore equals
and indicate that pay-performance sensitivi- the continuously accrued rate of return on
ties are monotonically increasing after age common stock, rt.The estimated coefficient
52. The coefficient of the CEO/chairman is the elasticity of cash compensation with
dummy variable is positive and significant, respect to shareholder value (or, following
indicating that pay-performance sensitivi- Rosen (1992), the ‘semi-elasticity’ of pay
ties are higher for combined CEO/chairmen. with respect to the rate of return).23
So, although the level of pay is higher when Table 14.7 reports coefficients from
the posts are combined (and corporate estimating (6) with US interactions. The
governance reformers typically argue for pay-performance elasticity for UK CEOs in
these posts to be separated for this reason) column (1) is 0.1213, indicating that an
the link between pay and performance additional ten percentage point shareholder
(effective ownership percentage) is also return corresponds to an additional 1.2%
higher. Including the control for CEO/ pay raise. The pay-performance elasticity
chairman combinations reduces the coeffi- for US CEOs in column (1) is 0.27
cient on the US dummy variable to 0.9632, (≈0.1213 0.1488), or more than double
but the coefficient remains significantly the elasticity for UK CEOs. The difference
positive. Columns (3) and (4) present in elasticities is statistically significant at
coefficients from median regressions. The the 10% level. The remaining columns in
book to market and volatility measures are Table 14.7 report estimated elasticities for
now significant. Importantly, the US dummy the industrial groups. The estimated US
variables in both columns (3) and (4) are elasticity is higher than the corresponding
positive and significant (although quantita- UK elasticity for all industries, although the
tively smaller than in the mean regressions), difference is only statistically significant for
indicating that median pay-performance the financial services industry.
Table 14.7 CEO pay-performance elasticities for salary and bonus, by industry
Independent variable All industries Mining & manufacturing Financial services Utilities Other industries
Note: White asymptotic t-statistics in parentheses. UK data from the largest 510 companies in fiscal 1997, ranked by market capitalisation. US data include firms in the
S&P 500, the S&P MidCap 400, and the S&P SmallCap 600. Shareholder return from Datastream (UK) and Compustat (US).
CEO PAY IN THE US AND UK 387
Table 14.8 CEO turnover-performance regressions, by industry
Independent variable All industries Mining & manufacturing Financial services Utilities Other industries
Note: Dependent variable 1 if CEO is in last year of office in 1996. Mean of the dependent variable is 0.119. Shareholder return is measured in 1996. z-statistics in
parentheses. UK data from the largest 510 companies in fiscal 1997, ranked by market capitalisation. US data include firms in the S&P 500, the S&P MidCap 400, the
S&P SmallCap 600, and companies in supplemental S&P indices. Shareholder return from Datastream (UK) and Compustat (US).
CEO PAY IN THE US AND UK 389
3.3. CEO turnover and corporate (iii) we are not controlling for corporate
performance governance variables, including the composi-
tion of the board. We leave a comprehensive
Another important potential source of investigation to future research.
managerial incentives is the threat of being
fired for poor performance. Several
researchers, beginning with Coughlan and
Schmidt (1985), Warner et al. (1988), and 4. DISCUSSION
Weisbach (1988) have documented an
inverse relation between CEO turnover and Consistent with our objective of generating
shareholder returns for US companies; stylised facts to stimulate future research,
Conyon (1998) has shown that the inverse our paper to this point has been primarily
relation holds for UK data as well, and descriptive. That is, although we have
Kaplan (1994 a, b) reports similar findings documented that American CEOs are
for Japan and Germany. higher paid and have better stock-based
Table 14.8 reports the coefficient incentives than British CEOs, we have
estimates from logistic regressions of CEO offered little in the way of theory or
turnover on shareholder returns by industry. conjecture to help explain these interesting
The dependent variable is equal to one if differences. In this section, we will briefly
the CEO is in his last full year of office in discuss a variety of factors that may
fiscal year 1996.The independent variables explain the crosscountry differences. As
include 1996 fiscal-year performance, a noted earlier, our objective is not to
US dummy, and the US dummy interacted reconcile completely the differences in
with the performance variable to identify pay practices, but rather to identify
crosscountry differences in the turnover- potential explanations, leaving formal
performance relation. The coefficient on hypothesis development and testing to
the performance variable is negative in all future research.
regressions (and significant in all but
financial services), indicating that CEOs in 4.1. Agency-theoretic
the United Kingdom are indeed more likely
considerations
to depart following poor performance.
The interaction variable is positive but The traditional principal-agent model
insignificant in all regressions, suggesting highlights the trade-off between risk and
that there are no systematic US-UK dif- incentives. Increasing the pay-performance
ferences in the CEO turnover-performance sensitivity imposes more risk on CEOs, who
relation in our data.24 in turn demand higher levels of expected
Our conclusion that the turnover- compensation to compensate for the
performance relation in the United Kingdom additional risk. Therefore, the facts that
is not significantly different from that in US CEOs are better paid and have stronger
the United States is robust to alternative incentives than UK CEOs may both reflect
definitions of performance (using various equilibrium incentive contracts.
lagged performance measures, and allowing To illustrate, suppose that firm value is
performance-size interactions). In addition, given by x e ‡, where e is executive
our conclusion is consistent with Kaplan effort, and ‡ is (normally distributed)
(1994a,b), who finds no systematic differ- uncontrollable noise, ‡ ≈ N(0, 2). Moreover,
ences in turnover-performance relations in suppose that managerial contracts take the
the United States vs. Japan or the United simple linear form (x) s bx, where
States vs. Germany. However, we view our s is a fixed salary and b is the sharing
results here as preliminary because (i) they rate (or ‘pay-performance sensitivity’).
are based on only a single year of turnover Assuming that the executive has exponential
data; (ii) we are not controlling for utility, U(x) er[Wc(e)], where r is the
turnover related to normal retirement; and executive’s absolute risk aversion and c(e)
390 DIRECTORS’ REMUNERATION
is the convex disutility of effort, the optimal may explain divergences in pay between the
sharing rate is given by:25 United States and United Kingdom.
Similarly, American CEOs may have more
b 1 decision rights and influence over corporate
. (7)
1 r2c results than do British executives.
Unfortunately, traditional agency theory
Equation (7) implies that the optimal pay-
gives little guidance on why the career
performance sensitivity b will equal 1 when
paths, production functions, or hierarchical
output is certain (2 0) or executives
structures should vary across international
are risk-neutral (r 0). Incentives will be
boundaries.
weaker for more risk-averse executives
(∂b / ∂r 0), and will also be weaker the
greater the uncontrollable noise in firm
value (∂b / ∂2 0). Moreover, expected
4.2. Taxes
compensation E (w) s bE (x) will As emphasised by Miller and Scholes
increase monotonically with b to compensate (1982) and Abowd and Bognanno (1995),
for both the increased risk imposed, and corporate and personal tax regimes affect
the increased effort induced, by higher the optimal structure of executive compensa-
pay-performance sensitivities. tion contracts. Overall, the personal income
Equation (7) suggests a finite number of tax rules and rates affecting executive pay
factors that might explain higher pay and are quite similar in the United States and
incentives among US executives. First, United Kingdom. Cash compensation and
American CEOs may be less risk averse or gains from ‘unapproved’ (or ‘nonqualified’)
have steeper marginal costs of effort than share options are taxed as personal income
their British counterparts, but to our in both countries, at comparable rates.27
knowledge there is no theory or empirical ‘Approved’ share options, granted infre-
work suggesting such international differ- quently to top executives because of
ences in risk-aversion coefficients. Second, institutional restrictions, are taxed at
UK performance might be measured with capital gains rates in both countries. In
substantially more noise than in the United recent years, the capital gains rate in the
States, leading to lower pay-performance UK has increased from 30% to 40%, while
sensitivities and lower expected levels of falling from 28% to 20% in the US.
pay. However, we find no evidence that cash Apart from the difference in capital
flows or shareholder returns are systemati- gains tax rates, which is both relatively
cally more variable in the United Kingdom recent and fairly benign (given the paucity
than in the United States. of approved share grants), there are no
Overall, the traditional principal-agent major differences in the personal tax regimes
model encapsulated in (7) does not offer in the United Kingdom and the United States
promising explanations for the difference in related to executive compensation. However,
pay levels and incentives in the two countries. there are two significant differences in
Extensions of the model to incorporate the extent to which compensation is
differences in both ability and in the deductible from corporate profits as an
marginal productivity of CEO effort might ordinary business expense. First, the
help reconcile the data, but only given the exercise-date spread between the market
additional assumptions that executives and exercise price on unapproved options is
are more able and more productive in the treated as a deductible compensation
United States.26 For example, Granick expense in the United States, but not in
(1972) shows that US managers have the United Kingdom. Second, under rules
greater exposure to different functions and passed in 1993, the United States limits
operating divisions and that such diversity the deductibility of ‘non-performance-
is rewarded at the top of the hierarchy. The based’ compensation (including salaries,
complexity of the filtering and training restricted stock, and discretionary bonuses)
processes, and differences in each country, to $1 million.28 Thus, UK rules allow
CEO PAY IN THE US AND UK 391
deductions for cash compensation but and opted against requiring an accounting
not for exercised options, while US rules change, adopting instead enhanced footnote
allow deductions for exercised options but disclosure of granting practices.The October
limit deductions for cash compensation. As 1995 FASB retreat ensured the continued
discussed in the next sub-section, we trend in option awards, since the grants
believe that these differences have played remained invisible from an accounting
at least some role in the relative prevalence perspective, but fully deductible from a
of share options in US pay packages. tax perspective.
The S&P500 Index (a broad measure of
4.3. The rise (US) and US stock-market performance) increased
fall (UK) of share options by 300% in the 1990s; the UK FTSE
Index increased by only 150% during the
The fact that US CEOs are better paid with same period. Proponents of large share
stronger pay-performance sensitivities is option grants have pointed to the US stock-
largely attributable to the relative preva- market performance as evidence that
lence and magnitude of share options in the options provide incentives for executives to
US. As documented by Hall and Liebman increase shareholder wealth. But, the robust
(1998) and Murphy (1999) and suggested stock market has also been a contributing
by Figure 14.2, the importance of share factor to the growing demand for option
options in US compensation packages is a compensation among US executives. The
relatively recent phenomenon, reflecting current cohort of executives has not expe-
an explosion in option granting practices rienced a major market downturn, and the
since the mid-1980s. Over the same period, overwhelming majority of US share options
many UK companies have rejected share issued since 1980 have been exercised well
option plans in favour of performance in-the-money. As a result of these past
share plans such as LTIPs (Main, 1999). successes, both executives and employees
Understanding the factors that have led have embraced share options as the quickest,
to the rise in share options in the United and (by their perception) the surest, route
States, and the fall of share options in the to obtaining substantial wealth.
United Kingdom, is critical in understand- Economic, political and cultural factors
ing the differences in compensation and have also shaped option granting policies in
incentives in the two countries. the United Kingdom. In contrast to the
The increased popularity of share option United States, where these factors have
grants for US executives can be traced, in encouraged the use of share option com-
large part, to a combination of economic, pensation, a variety of recent statutory and
political, and cultural factors. Shareholder non-statutory arrangements in the United
groups and academics in the early 1990s Kingdom have discouraged share option
called for more stock-based compensation in grants. Options in the United Kingdom
CEO pay packages, citing the lack of mean- became controversial in 1995, after execu-
ingful rewards and penalties in the typical tives in several recently privatised electric
package. The deductibility limitations on utilities exercised options worth millions
non-performance-related pay introduced by of pounds. The influential Greenbury report
the Clinton Administration in 1993 further (1995) encouraged companies to replace
fuelled the popularity of stock options. The their option plans with LTIP share plans
trend was nearly reversed in the mid-1990s, which ‘may be as effective, or more so,
however, when the Financial Accounting than improved share option schemes in
Standards Board (FASB) announced plans linking rewards to performance’ Greenbury
to institute a grant-date accounting charge (1995, paragraph 6.32). In response to
for companies granting share options. the Greenbury report and the ongoing
Ultimately, FASB yielded to pressure from controversy, the government tightened the
the business community, accounting firms, restrictions on approved option awards,
and the hi-tech (‘Silicon Valley’) sector, reducing the amount that could be awarded
392 DIRECTORS’ REMUNERATION
(expressed as the aggregate exercise price) perhaps it is natural that the United States
from the greater of £100,000 or four times reacts to claims of excessive CEO pay
cash emoluments to only £30,000. by increasing the link between pay and
An important non-statutory consideration performance, thus exacerbating income
for UK companies is the codes of conduct inequality, while the United Kingdom reacts
issued by institutional investors and their through wage compression and reducing
representatives. Although not legally binding, the pay-performance link.
policy guidelines from the Association of The divergence in top wages in the UK
British Insurers (ABI) and other investor vs. US is not, of course, limited to top
groups are highly influential and closely executives. The best doctors, engineers,
followed.The ABI guidelines (1994, 1995), professors, athletes, lawyers, investment
for example, effectively constrain the bankers, and entertainers all earn substan-
issuing of share options—approved and tially more in the United States than in the
unapproved—to four times cash compensa- United Kingdom. The US wage premiums
tion.29 More recently, the Pension Investment for ‘superstars’ in all occupations persist
Research Consultants (PIRC, 1998) called in spite of the similarities in language,
on companies to toughen performance culture, tax regimes, and institutions. These
targets for executive option schemes and premiums may be evidence that the US
other incentive plans. market for superstars (Rosen, 1981) is
In sum, executive pay has been highly more competitive than the UK market,
controversial in both the United States and with ‘winner-take-all’ rents flowing to the
the United Kingdom, and has generated a producers (Frank and Cook, 1995). Thus,
variety of responses from the government perhaps the CEO-pay differentials should
and tax authorities, the accounting boards, be examined in the context of broader
the media, and institutional investors. competitive and culture factors.
Although the root causes of the controversy
in the two countries are basically similar,
the responses have been markedly different: 5. CONCLUSION
while the net effect of these factors has
been to greatly increase option grants in The recent wave of international corporate
the United States, the net effect has mergers, such as Daimler-Chrysler and
discouraged the use of option grants in the British Petroleum-Amoco, has sparked
United Kingdom in favour of LTIPs. academic and practitioner interest in under-
standing international differences in executive
4.4. Culture pay policies. International comparisons are
inherently difficult, because of fundamental
We have documented that US CEOs have differences in data availability, tax regimes,
higher pay and stock-based incentives than corporate governance and the organisation
CEOs in the United Kingdom, and shown of business. For a variety of reasons, the
that these results are driven, in large part, United States and United Kingdom offer a
by the recent divergence in share option natural laboratory to examine in detail
practices. Although it is difficult for us to differences in compensation and incentives
explain why a similar controversy over CEO practices. First, the United Kingdom and
pay has led to increased option grants in United States (along with Canada) are
the United States but decreased grants currently the only countries that require
in the United Kingdom, it is tempting to detailed disclosure on the compensation
attribute at least part of the divergence practices for individual top corporate
to cultural differences between the two executives. Second, the United Kingdom
countries. The United States, as a society, and United States share a common
has historically been more tolerant of language and have similar capital markets
income inequality, especially if the inequality and underlying economies. Third, the
is driven by differences in effort, talent, or United States and United Kingdom employ
entrepreneurial risk taking. In this light, similar corporate governance structures,
CEO PAY IN THE US AND UK 393
S&P 500 companies in which the top five details there is some risk that the abundance of
executives hold any share options.This definition information will mask rather than highlight
of prevalence closely tracks the Conference the nature and scale of option schemes’, and
Board’s survey responses for the 1992–6 period allowed less-than-complete option information
where both sources of prevalence data are when remuneration committees are ‘satisfied
available. that this will not result in failure to disclose
14 Guy (1999) documents empirical evidence information of material importance’. Greenbury
that the annually estimated CEO pay-size elas- (1995, p. 29).
ticity varies positively with annual median pay. 20 About 80% of UK annual reports provided
This is consistent with our results since median full information (i.e. the exercise price and
US pay is higher than in the United Kingdom. expiration date) on every prior grant of options
15 Calculated as e0.3861 1 ≈ 0.47. still outstanding. In the remaining 20% of cases
16 However, in unreported regressions on only an exercise price is almost always given although
UK data, cash compensation is positively and it is weighted exercise price of all options held
significantly related to volatility, and negatively (see Conyon and Sadler, 1999). The expiration
and significantly related to book-to-market date is given in only 50% of cases and is
ratios. Since our primary objective here is to typically the expiration date of the longest-dated
identify cross-country differences in pay levels, option. US proxy statements provide information
we leave further examination of cross-country on the number and intrinsic value of options
differences in slope coefficients to future held at the end of the fiscal year (based on the
research. fiscal year-end stock price, P). The number (N)
17 For example, 0.0663 Age 0.0006 and intrinsic value (Y) of previously granted
Age2 reaches a maximum at age 0.0663/ options is calculated by subtracting new grants
(2 0.0006) 55.25. from total outstanding options, and adjusting
18 The percentage option holdings multiplied the year-end intrinsic value of the new grants
by the option delta is a measure of the change from the total intrinsic value. We treat the
in CEO option-related wealth corresponding to a previously granted options as a single prior grant
change in shareholder wealth. More formally, with exercise price X, where N(P X) Y, or
suppose that the CEO holds N share options, X P (Y/N).
and suppose that shareholder wealth increases 21 See, for example, Conyon and Peck (1998 a).
by £1. If there are S total shares outstanding, 22 See Murphy (1999) for a discussion and
the share price P will increase by P £1/S, comparison of the various approaches.
and the value of the CEO’s options will increase 23 Most of the UK literature has been
by N P(∂V/∂P), where V is the Black-Scholes concerned with estimating this type of equation
value of each option. Substituting for P, the and ignoring the importance of the direct rela-
CEO’s share of the value increase is given by tion between CEO pay and shareholder wealth
(N/S) (∆V/∂P), or the CEO’s options held as a (as reviewed in Conyon et al., 1995). Moreover,
fraction of total shares outstanding multiplied Gregg et al. (1993) show that the ‘implicit’ link
by the ‘slope’ of the Black-Scholes valuation. between CEO pay and shareholder returns
For examples of this approach which yield a became de-coupled in the 1990s. In fact, this
distribution of CEO incentives, see Jensen and may simply be attributable to the restructuring
Murphy (1990b), Yermack (1995), and Murphy of CEO pay packages and the importance of
(1999). Hall and Murphy (2000) offer a modified non-cash forms of compensation.
approach to measure the pay-for-performance 24 Estimating logit regressions for each
incentives of risk-averse undiversified executives. country separately yielded negative and signifi-
An alternative approach, adopted by Jensen cant correlations between CEO turnover on
and Murphy (1990a) for the United States and shareholder returns in both the US and UK data.
Main, et al. (1996) for the United Kingdom, 25 For similar derivations of the optimal pay-
involves estimating the option pay-performance performance sharing rate, see Lazear and Rosen
sensitivity as the coefficient from a regression of (1981), Holmstrom and Milgrom (1991),
the change in option value on the change in Gibbons and Murphy (1992), and Milgrom and
shareholder wealth. The procedure is similar to Roberts (1992).
how we identify the ‘indirect’ relation between 26 Half of us are unwilling to concede to such
pay and performance below. assumptions.
19 Although the Greenbury report generally 27 The highest marginal tax rate is approxi-
required detailed disclosure of prior grants, it mately 40% in both countries. The top UK rate
recognised that ‘in the disclosure of share option affects incomes above £28,000, while the top US
396 DIRECTORS’ REMUNERATION
bracket is indexed for inflation and is currently Conyon, M. J. and Peck, S. I. (1998b). ‘Board
about £175,000 for married taxpayers. control, remuneration committees and top
28 Research on the effects of the deductibility management compensation.’ Academy
limitation, which apply only to the top five Management Journal, vol. 41, pp. 146–57.
executives in publicly traded corporations, Conyon, M. J. and Sadler, G. V. (1999).
has concluded that the rule has had a modest
‘CEO compensation, option incentives and
influence on both the level and structure of US
pay plans (Perry and Zenner, 1999; Rose and information disclosure.’ Warwick University
Wolfram, 1997). mimeograph.
29 Conyon and Sadler (1999) and Main Conyon, M. J. and Schwalbach, J. (1999).
(1999) consider the implications of this rule. ‘Corporate governance, executive pay and
30 Based on data from Forbes’ Annual Survey performance in Europe.’ In (J. Carpenter and
of Executive Compensation, May 19, 1997. D. Yermack, eds.) Executive Compensation
and Shareholder Value: Theory and
REFERENCES Evidence, pp. 13–34. Dordrecht: Kluwer
Academic Press.
Conyon, M. J., Gregg, P. and Machin, S. (1995).
Abowd, J. and Bognanno, M. (1995).
‘Taking care of business: executive compensa-
‘International differences in executive and
tion in the UK.’ ECONOMIC JOURNAL, vol. 105
managerial compensation.’ In (R. Freeman
(May), pp. 704–15.
and L. Katz, eds.) Differences and Changes in
Cosh, A. and Hughes, A. (1997).‘Executive remu-
Wage Structures, pp. 67–103. Chicago:
neration, executive dismissals and institutional
University of Chicago Press.
shareholdings.’ International Journal of
Association of British Insurers (1995) ‘Share
Industrial Organization, vol. 15, pp. 469–92.
options and profit sharing incentive schemes.’
Coughlan, A. T. and Schmidt, R. M. (1985).
ABI, London.
‘Executive compensation, managerial turnover
Association of British Insurers (1994) ‘Long
and firm performance.’ Journal of Accounting
term remuneration for senior executives.’
and Economics, vol. 7, pp. 43–66.
ABI, London.
Crystal, G. (1991). In Search of Excess: The
Black, F. and Scholes, M. (1973).‘The pricing of
Overcompensation of American Executives.
options and corporate liabilities.’ Journal of
New York: Norton & Company.
Political Economy, vol. 81, pp. 637–59.
Ezzamel, M. and Watson, R. (1998). ‘Market
Buckingham, L. and Cowe, R. (1998a).
comparison earnings and the bidding-up of
‘Boardroom pay 98: Pleas for restraint fail to
executive cash compensation: evidence from
curb executive excess.’ The Guardian, 22 July.
the United Kingdom.’ Academy Management
Buckingham, L. and Cowe, R. (1998b). ‘The
Journal, vol. 41, pp. 221–31.
boardroom bonanza: exclusive survey reveals
Frank, R. H. and Cook, P. J. (1995). Winner-
top directors’ pay soaring by 18%.’ The
Take-All Society. New York: Free Press.
Guardian, 22 July.
Gibbons, R. and Murphy, K. J. (1992). ‘Optimal
Cadbury, A. (1992).‘Code of best practice.’ Report
incentive contracting in the presence of
from the Committee on Financial Aspects of
career concerns: theory and evidence.’
Corporate Governance. London: Gee Publishing.
Journal of Political Economy, vol. 100,
Carpenter, J. (1998).‘The exercise and valuation
pp. 468–505.
of executive stock options.’ Journal of
Granick, D. (1972). ‘Managerial Comparisons
Financial Economics, vol. 48(2), pp. 127–58.
of Four Developed Countries: France, Britain,
Conyon, M. J. (1998). ‘Directors’ pay and
United States and Russia.’ Cambridge MA:
turnover: an application to a sample of large
MIT Press.
UK firms.’ Oxford Bulletin of Economics and
Greenbury, R. (1995). ‘Directors’ Remuneration:
Statistics, vol. 60, pp. 485–507.
Report of a study group chaired by Sir
Conyon, M. J. and Peck, S. I. (1998a). ‘Recent
Richard Greenbury.’ London: Gee Publishing.
developments in UK corporate governance.’
Gregg, P., Machin, S. and Szymanski, S. (1993).
In (T. Buxton, P. Chapman and P. Temple,
‘The disappearing relationship between
eds.) Britain’s Economic Performance.
directors’ pay and corporate performance.’
London: Routledge.
CEO PAY IN THE US AND UK 397
British Journal of Industrial Relations, Main, B. G., O’Reilly, C. and Crystal, G. (1994).
vol. 31, pp. 1–10. ‘Over here and over there: a comparison of
Guy, F. (1999). ‘Earnings distribution, corporate top executive pay in the UK and the USA.’
governance and CEO pay.’ ESRC Centre for Contributions to Labour Studies, vol. 4,
Business Research, University of Cambridge. pp. 115–27.
Working Paper No. 126. Milgrom, P. and Roberts, J. (1992). Economics,
Hall, B. J. and Leibman, J. B. (1998). Organizations, and Management. New
‘Are CEOs really paid like bureaucrats?’ Jersey: Prentice Hall.
Quarterly Journal of Economics, vol. 113 (3), Miller, M. and Scholes, M. (1982). ‘Executive
pp. 653–91. compensation, taxes and incentives.’ In (W.
Hall, B. J. and Murphy, K. J. (2000). ‘Optimal Sharpe and C. Cootner, eds.) Financial
exercise prices for executive stock options.’ Economics: Essays in Honor of Paul
American Economic Review, Volume 20, Cootner, pp. 179–201. Englewood Cliffs NJ:
No 2, May, pp. 209–214. Prentice Hall.
Hampel, R. (1998). ‘Committee on Corporate Murphy, K. J. (1999). ‘Executive compensation.’
Governance: Final Report.’ London: Gee In (O. Ashenfelter and D. Card, eds.)
Publishing. Handbook of Labor Economics, vol. 3.
Holmstrom, B. and Milgrom, P. (1991). Amsterdam: North Holland.
‘Multi-task principal-agent analyses: incentive PIRC (1998). Executive Share Schemes:
contracts, asset ownership and job design.’ Trends in 1998. London, Pension Investment
Journal of Law, Economics and Organization, Research Consultants.
vol. 7, pp. 24–52. Perry, T. and Zenner, M. (1999). ‘Pay for per-
Jensen, M. and Murphy, K J. (1990a). formance? Government regulation and the
‘Performance pay and top-management structure of compensation contracts.’
incentives.’ Journal of Political Economy, University of North Carolina mimeograph.
vol. 98(2), pp. 225–64. PriceWaterhouseCoopers Corporate Register,
Jensen, M. and Murphy, K. J. (1990b). ‘CEO various years, published Hemmington Scott.
incentives: it’s not how much you pay, but Rose, N. and Wolfram, C. (1997). ‘Regulating
how.’ Harvard Business Review, vol. 68 CEO pay: assessing the impact of the
(May/June), pp. 138–49. tax-deductibility cap on executive compensa-
Kaplan, S. (1994a). ‘Top executive rewards and tion.’ MIT.
firm performance: a comparison of Japan and Rosen, S. (1981). ‘The economics of superstars.’
the United States.’ Journal of Political American Economic Review, vol. 71 (5),
Economy, vol. 102(3), pp. 510–46. pp. 845–58.
Kaplan, S. (1994b). ‘Top executives, turnover, Rosen, S. (1992). ‘Contracts and the market for
and firm performance in Germany.’ Journal of executives.’ In (L. Werin and H. Wijkander,
Law, Economics and Organization, vol. 10(1), eds.) Contract Economics, pp. 181–211.
pp. 142–59. Oxford: Blackwell.
Lazear, E. and Rosen, S. (1981). ‘Rank order Warner, J., Watts, R. and Wruck, K. (1988).
tournaments as optimum labour contracts.’ ‘Stock prices and top management changes.’
Journal of Political Economy, vol. 89, Journal of Financial Economics, vol. 20,
pp. 841–64. pp. 461–92.
Main, B. G. (1999). ‘The rise and fall of execu- Weisbach, M. (1988). ‘Outside directors and
tive share options in Britain.’ In (J. Carpenter CEO turnover.’ Journal of Financial
and D.Yermack, eds.) Executive Compensation Economics, vol. 20, pp. 431–60.
and Shareholder Value: Theory and Evidence, Yermack, D. (1995). ‘Do corporations award
pp. 83–113. CEO stock options effectively?’ Journal of
Dordrecht, Kluwer Academic Press. Financial Economics, vol. 39, pp. 237–69.
Main, B. G., Bruce, A. and Buck, T. (1996). Zhou, X. (1999). ‘CEO pay, firm size and
‘Total board remuneration and company corporate performance: evidence from
performance.’ ECONOMIC JOURNAL, vol. 106, Canada.’ Canadian Journal of Economics,
pp. 1627–44. (forthcoming).
Chapter 15
Wayne R. Guay*
THE SENSITIVITY OF CEO
WEALTH TO EQUITY RISK: AN
ANALYSIS OF THE MAGNITUDE
AND DETERMINANTS
Source: Journal of Financial Economics, 53(1) (1999): 43–71.
ABSTRACT
To control risk-related incentive problems, equity holders are expected to manage both the
convexity and slope of the relation between firm performance and managers’ wealth. I find
stock options, but not common stockholdings, significantly increase the sensitivity of CEOs’
wealth to equity risk. Cross-sectionally, this sensitivity is positively related to firms’ investment
opportunities. This result is consistent with managers receiving incentives to invest in risky
projects when the potential loss from underinvestment in valuable risk-increasing projects
is greatest. Firms’ stock-return volatility is positively related to the convexity provided to
managers, suggesting convex incentive schemes influence investing and financing decisions.
managers’ wealth–performance relation. I can also contribute to the wealth effect when
conclude with Section 7. their payoffs are concave or convex functions
of firm performance (see Section 5.2 for
further discussion).
2. MANAGERS’ PREFERENCES The presence of convex, or option-like,
TOWARD FIRM RISK AND RISK- payoffs in the structure of executives’
RELATED AGENCY COSTS incentives is well-recognized by both
academics and practitioners. For example,
The relation between firm risk and managers’ Jensen and Meckling (1976) point out that
incentives is well-developed in the literature. equity holders in a levered firm essentially
When a manager holds common stock and hold a European call option to buy the firm
stock options, a dependence exists between at an exercise price equal to the face value
his wealth and the firm’s stock-price of debt. As such, the value of common
performance (see Jensen and Meckling, stock increases with the volatility of the
1976; Jensen and Murphy, 1990). This firm’s cash flows. Haugen and Senbet
dependence is commonly referred to as the (1981) analyze the incentive effects of
wealth–performance relation. Since stock convex payoffs from employee stock
price varies over time, the payoffs to this options. Several empirical studies, such as
incentive scheme are uncertain, and risk is Agrawal and Mandelker (1987), DeFusco
imposed on the manager. In Pratt (1964), et al. (1990), and Tufano (1996), have also
a risk-averse manager is shown to be examined the convexity implications of
indifferent between a risky payoff and the managers’ stock-based wealth. However,
following payoff with certainty: none of these studies has quantified the
magnitude of this convexity for common
Certainty equivalent E(wealth) risk stock or stock options, or explored its
premium. (1) determinants. The following excerpt from
the Committee Report on Executive
Differentiating this expression with respect Compensation in BellSouth’s 1993 proxy
to firm risk yields the following: statement (p. 11) illustrates that corporate
compensation committees also recognize
∂CE/∂ ∂E(wealth)/∂ the need to structure compensation plans
∂(risk premium)/∂. (2) to give executives appropriate risktaking
incentives:
Smith and Stulz (1985) and Lambert
et al. (1991) illustrate how this expression BellSouth’s long term [incentive] program
partitions the effect of firm risk on man- is intended to focus the executive group
agers’ preferences into two components. on the achievement of corporate goals ....
The first component, ∂E(wealth)/∂, Recognizing the dynamic convergence of
which I will refer to as the wealth effect, industries such as telecommunications,
represents the change in expected wealth entertainment, and cable which will
experienced by a manager when firm risk continue throughout the next decade, the
changes. When the payoffs of an incentive Committee wants to motivate BellSouth
scheme are a linear function of firm executives to take the risks necessary to
performance, the wealth effect is zero, secure a strong foothold for BellSouth
since changes in the distribution of firm in this extremely competitive new
performance do not affect managers’ marketplace.
expected wealth. Expected wealth is an
increasing function of risk when managers The second component in Eq. (2), ∂(risk
own securities with convex payoffs, such premium)/∂, which I will refer to as the
as stock options or common stock, that risk-aversion effect, captures the influence of
increase in value as firm risk increases. risk-aversion on managers’ utility. If man-
Bonus plans and other incentive schemes agers are risk-averse and poorly diversified
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 401
constructed portfolios were exercised on combined with the time series of options
December 31, 1993, and compare this granted, to construct an estimate of the
amount to the cash value of the CEOs’ composition of each CEO’s option portfolio
options if exercised immediately. This cash as of December 31, 1993. Detailed infor-
value disclosure is required for all named mation on option exercises is not readily
executive officers. CEOs are excluded from available from proxy statements prior to
the sample if these two cash value measures 1992. I assume that, unless otherwise
differ by more than 25% and $200,000, noted, options with the shortest remaining
resulting in the removal of 14 CEOs. time-to-maturity are exercised first. Since
Finally, CEOs are excluded if they own this exercise strategy is expected to simu-
more than one-third of their corporation’s late the true exercise strategies with error,
common stock, since it is doubtful that it is important to remove CEOs with
these managers’ compensation schemes are constructed option portfolios that differ
designed for contracting purposes. This significantly from the option portfolio
criteria reduces the sample by three CEOs. characteristics disclosed in the 1993 proxy
The results are not sensitive to using a statements, as described in Section 3.
lower ownership threshold, such as five or Table 15.1 indicates that common stock
ten percent stockholdings. The final sample and options are significant components of
contains 278 CEOs. CEOs’ incentive schemes. Using the Black
and Scholes (1973) model, the median
stock option portfolio has a value of $1.78
3.1. Summary of compensation million. Furthermore, the substantial
and stock-based wealth data standard deviation and interquartile range
For each CEO, I compile data on common (not reported) suggest there is considerable
stock, restricted stock, and outstanding variation in the extent to which firms use
stock options held as of December 31, options to provide managers with incentives.
1993, as well as salary and bonus received The median value of CEOs’ common stock-
during 1993. With the exception of the holdings is $2.74 million, which is about
composition of the CEOs’ option portfolios, 50% larger than the median value of
this data is readily available from the firms’ option holdings. Though restricted stock is
1993 proxy statements.To reduce the influ- held by less than half of the CEOs, nearly
ence of a few CEOs that hold extremely 25% of the CEOs hold restricted stock
large quantities of common stock, the upper- valued at $1 million or more. The median
most percentile of aggregate stockholdings value of total stock-based wealth, including
and sensitivities of these stockholdings to stock options, common stockholdings, and
stock price and equity risk are set equal to restricted stock, is $6.79 million, and
the stockholdings and sensitivities of the several times larger than the median
CEOs at the 99th percentiles. annual cash compensation of $866,700.
I construct the CEO’s option portfolios
using the past time-series of proxy state-
ments. For each CEO, I collect data on the 4. ESTIMATING CONVEXITY
number of options granted in each fiscal IN THE WEALTH–PERFORMANCE
period, the average exercise price per RELATION
option granted, and the actual or maximum
allowable time-to-maturity for the options As discussed above, the focus of this study
granted. In approximately 5% of the fiscal is not on the slope of CEOs’ wealth–
periods, it was necessary to average option performance relation, but instead on the
grants over a multi-year period. Beginning convexity of this relation. I measure the
in 1992, proxy statements report the total convexity contributed by a stock option or
number of stock options held by each share of common stock as the change in
named executive officer. I use the number the security’s value for a 0.01 change in the
of options held at the end of 1993, annualized standard deviation of stock
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 403
Table 15.1 Summary statistics for CEOs’ stock option portfolios, common stockholdings, and cash
compensation
The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat,
ranked by market value of equity on December 31, 1988. The CEOs’ salary and bonus are for the
fiscal year ending December 31, 1993. The number of options held by the CEOs and their stock-
based wealth are as of December 31, 1993. Option values are based on the Black-Scholes formula
for valuing European call options, as modified to account for dividend payouts by Merton (1973).
Option value is calculated as [SedTN(Z) XerTN(Z T(1/2))], where Z is equal to [In(S/X)
T(r d /2)]/T(1/2). The parameters in the Black-Scholes model are set as follows: S price
of the underlying stock at December 31st, 1993, E exercise price of the option, annualized
volatility, estimated as standard deviation of daily logarithmic stock returns over the last 120 trad-
ing days in 1993, multiplied by 252(1/2), r ln(1 risk-free interest rate), where the risk-free
interest rate is the yield, as of December 31, 1993, on a U.S. Treasury strip with the same time to
maturity as the remaining life of the stock option, T remaining time to maturity of the option
in years, as of December 31, 1993, and d ln(1 expected dividend rate), where the expected
dividend rate is set equal to the dividends paid during 1993 divided by the year-end stock price.Total
stock-based wealth is the sum of the value of the CEO’s options, stockholdings, and restricted stock.
techniques, all tests are reproduced using the sum of book debt plus market value of
methods suggested in Hemmer et al. (1994) equity is 0.99.
and in Statement of Financial Accounting Together with the other parameters,
Standards No. 123, ‘Accounting for stock- I use the implied firm value to estimate the
based compensation’ (FASB, 1995). To Black–Scholes partial derivative of each
implement the Hemmer et al. and FASB firm’s stock price with respect to volatility.
techniques, I assume that the expected time Note that because common stock is an
to exercise for all options held is equal to option on firm value, its value is sensitive to
60% of the remaining time-to-maturity, the volatility of firm value. In contrast, a
though the cross-sectional results are not stock option’s value is sensitive to stock-
sensitive to varying this percentage. With return volatility. To make the two convexity
respect to the Hemmer et al. method, I also measures comparable, I compute the
assume that all options have a three-year change in common stock value with respect
vesting period at the grant date. The option to a 0.01 change in the annualized standard
values and convexity measures using these deviation of stock-returns. Since firm value
procedures are generally about 10–25% is the sum of debt value plus equity value,
lower than those using the Black-Scholes portfolio theory suggests that the volatility
formula. However, all cross-sectional results of firm value can be estimated using meas-
are qualitatively unchanged. ures of debt volatility and equity volatility,
As illustrated by Black and Scholes and an assumption about the correlation
(1973) and Smith (1976), the payoff to a between debt value and equity value. For
share of common stock in a levered firm simplicity, I set this correlation equal to
can also be viewed as a call option, where one, though the results are not sensitive to
the option’s underlying asset is the value of this assumption. The sensitivity of common
the firm and the exercise price is the face stock value to equity volatility is then
value of the firm’s liabilities. Unlike stock estimated, at the margin, by holding debt
options, the Black-Scholes parameters for volatility constant while stock-return
common stock are not all readily available. volatility is allowed to vary by 0.01 (see
However, an estimate of the value of the Section A.2. for further details). Finally,
option component is readily observable in I estimate the aggregate sensitivity of a
the stock price. Knowledge of this option CEO’s common stock to equity risk as the
value reduces, by one, the number of Black– partial derivative of stock price with
Scholes parameters that must be estimated. respect to stock-return volatility, multiplied
I estimate parameters for the exercise price, by the number of shares of stock held by
time to maturity, interest rate, dividend yield, the CEO.
and volatility of firm value as described in To gain some insight for the source of
Section A.2. of the appendix. Firm value, convexity inherent in common stock, consider
the sum of market value of equity and how the value of common stock varies as a
market value of debt, and the underlying function of firm value. In an unlevered
asset in this option pricing application, is left firm, the payoff to common stock is a linear
as the free parameter. Using the Black– function of firm value, and the value
Scholes model, I compute the implied of common stock increases by $1 for every
per-share value of each sample firm. As $1 increase in firm value. For a levered
reported in Table 15.6 of the appendix, the firm, an increase in firm value will not
implied per-share firm values are slightly accrue entirely to common stockholders,
smaller than the sum of per-share book but will instead be shared between the
value of debt plus stock price for nearly all debtholders and shareholders, assuming the
firms. This relation is not surprising given probability of financial distress is greater
that the value of an option is strictly than zero. Confirming this notion, a $1
greater than the price of the underlying change in firm value translates into a
asset less the exercise price.The correlation median change in common stock value
between the implied firm value and the of only $0.97, with $0.03 accruing to
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 405
the firms’ debtholders (see Table 15.6 in considerable error. The median aggregate
the appendix). sensitivity of CEOs’ wealth to equity risk
due to options is $29,893. These sensitivi-
4.1. Descriptive statistics on ties range from a minimum of $0 for the
convexity in the wealth–performance 50 CEOs that do not hold options, up to
relation a maximum of $347,256 for the CEO
holding the option portfolio most sensitive
Table 15.2 presents the characteristics to equity risk.
of the stock option portfolios held by the The sensitivity of common stock to
CEOs in the sample. Panel A indicates that equity risk is much smaller than that of
the characteristics of the CEOs’ options options. The median change in per-share
vary considerably. For each CEO, I compute stock price for a 0.01 change in stock-
the mean price-to-strike ratio and mean return volatility is $0.00005, compared
time-to-maturity of the options in their to $0.156 per option. This result is not
portfolio. The price-to-strike ratio is the surprising, given that the median price-
stock price divided by the option’s exercise to-strike ratio for a share of common
price. The mean price-to-strike ratio stock is 1.9, which is much larger than the
indicates, on average, the extent to which price-to-strike ratio of most CEOs’ stock
the CEO’s options are in the money. The options (see Table 15.6 of the appendix).
value of an option with a price-to-strike Even when multiplied by the quantity of
ratio much greater than one increases stock held, the total sensitivity of common
almost linearly with stock price, and as stock value to equity risk is very small for
such, will be quite insensitive to changes in most CEOs, with a median sensitivity of $2.
stock-return volatility. By comparison, an As a result, for most firms, the aggregate
“at the money” option has a price-to-strike sensitivity of CEOs’ stock-based wealth to
ratio of one, and will be more sensitive to equity risk due to options plus stock is
equity risk. The mean price-to-strike ratio driven primarily by stock options. However,
has a median value of 1.3, and ranges from the sensitivity of stock value to equity risk
about 0.4 to 9.0.The mean time-to-maturity is significant for a small number of CEOs.
is the number of years remaining to expira- For approximately 3% of the CEOs, the
tion, on average, for a CEO’s options. The change in stock value for a 0.01 change in
mean time-to-maturity has a median value stock-return volatility exceeds $25,000.
of 7.2 years, and ranges from 1.5 to 16.5 Thus, while the sensitivity of stock value to
years. This variation suggests that per- equity risk can potentially provide managers
option incentive effects are expected to with incentives to shift wealth from bond-
vary widely across CEOs. holders, this effect is expected to be
Estimates of the sensitivity of CEOs’ extremely isolated, and relevant only for
options and stock to equity risk are pre- firms experiencing severe financial distress.
sented in Panel B of Table 15.2. In the first To interpret the economic significance of
row, the sensitivity of option value to equity these sensitivities, one must consider both
risk is averaged over the options in each the sensitivity of a CEO’s wealth to equity
CEO’s portfolio. For the 228 CEOs with risk and his ability to alter the risk of the
options, the median change in value, per firm. I estimate the latter component for
option, for a 0.01 change in the standard each sample firm by first computing the
deviation of stock returns is $0.156, and difference between 1993 stock-return
ranges from $0.001 to $0.748.The consid- volatility and the 1993 volatility of a con-
erable variation in these per-share statistics trol sample obtained from the CRSP tapes,
indicates that the incentive effects of stock and matched on two-digit SIC code and
options are heavily dependent on security- market value of equity. Each sample firm is
specific parameters. Measures that ignore then assumed to experience a change in
this variation are likely to estimate the equity volatility that brings it in line with its
incentive effects of these instruments with industry and size-matched control portfolio.
Table 15.2 The sensitivity of CEOs’ wealth to equity risk and stock price
The data consist of option portfolios and common stockholdings for 278 CEOs selected uniformly from the 1000 largest firms on Compustat, ranked
by market value of equity on December 31, 1988. Price-to-strike ratio is the December 1993 stock price divided by the exercise price of an option.
The mean price-to-strike ratio is the simple average of the price-to-strike ratios for all options in a CEO’s option portfolio. Time-to-maturity is the
remaining time to expiration for an option as of December 31, 1993. The mean time-to-maturity is the simple average of the time-to-maturity for
all options in a CEO’s option portfolio. Sensitivity of wealth to equity risk due to options and stock is the change in the dividend-adjusted
Black–Scholes value of a CEO’s stock option portfolio and common stockholdings for a 0.01 change in the annualized standard deviation of the
firm’s stock returns (see the appendix). The average sensitivity of an option’s value to equity risk is the sensitivity of wealth to equity risk averaged
over all options in a CEO’s option portfolio. Total sensitivity of wealth to equity risk aggregates the per option and per-share sensitivities weighted
by the number of securities held. Sensitivity of wealth to a 1% change in stock price is the change in the dividend-adjusted Black–Scholes value of
stock options or common stockholdings for a 1% change in the value of the firm’s stock price (see the appendix). The average sensitivity of an
option’s value for a 1% change in stock price is the sensitivity of wealth to stock price averaged over all options in a CEO’s option portfolio.
406 DIRECTORS’ REMUNERATION
Standard
Option portfolio characteristic Mean deviation Minimum Q1 Median Q3 Maximum
Panel A:
Mean price-to-strike ratioa 1.50 0.86 0.38 1.07 1.30 1.67 9.00
Mean time-to-maturity per optiona 7.18 1.68 1.50 6.25 7.23 8.17 16.53
Panel B:
Average sensitivity of an option’s value to equity risk ($)a 0.167 0.105 0.001 0.097 0.156 0.212 0.748
Total sensitivity of wealth to equity risk: options ($)a 45,967 51,641 82 14,108 29,893 56,590 347,256
Total sensitivity of wealth to equity risk: options – all CEOs ($) 37,700 49,982 0 4614 20,915 48,164 347,256
Per-share sensitivity of stock price to equity risk: ($) 0.005 0.016 0.000 0.000 0.000 0.002 0.225
Total sensitivity of wealth to equity risk: stock ($) 2857 12,654 0 0 2 240 98,232
Total sensitivity of wealth to equity risk: stock options ($) 40,557 50,777 0 6489 22,664 55,770 347,292
Panel C:
Average sensitivity of an option’s value to a 1% change in stock 0.272 0.173 0.006 0.147 0.231 0.366 1.160
price ($)a
Per-share sensitivity of stock price to a 1% change in stock price ($) 0.393 0.247 0.029 0.248 0.336 0.490 2.548
Total sensitivity of wealth to a 1% change in stock price: options ($) 72,169 104,844 0 3997 36,407 96,250 687,350
Total sensitivity of wealth to a 1% change in stock price: stock ($) 251,995 886,860 0 10,571 37,307 108,341 6,008,957
Total sensitivity of wealth to a 1% change in stock price: 324,164 904,534 353 38,784 90,685 222,387 6,300,594
stock options ($)
The mean change in CEOs’ stock-based up risky, positive NPV projects.This problem
wealth from this change in volatility would is likely to be most severe in firms with
be $324,000, with a standard deviation of substantial investment opportunities (see
$562,000. This mean change in wealth is Milgrom and Roberts, 1992, Chapter 13,
about 20% larger, at $391,000, when for a discussion of potential risk-related
considering only CEOs with option port- agency costs in firms with valuable growth
folios.The magnitude of these wealth effects opportunities).4 As such, the expected
suggests that the convexity generated by loss from valuable projects bypassed by
stock options is potentially large enough to managers is hypothesized to be positively
influence managers’ behavior. related to the proportion of assets that
The maintained hypothesis in this paper represent growth options. By providing
is that firms add convexity to a manager’s managers with incentive schemes that have
incentive scheme to encourage investment convex payoffs, equity holders can reduce
in valuable risk-increasing projects, that is, these risk-related agency costs.The following
to help overcome the risk-aversion effect. hypothesis follows directly: Convexity in the
However, note that even in the absence of relation between managers’ wealth and stock
convexity in the wealth–performance rela- price is positively related to the proportion
tion, managers have incentives to undertake of assets that are growth options.
positive NPV projects when the slope of the I use three proxies to capture variation in
wealth–performance relation is greater than firms’ investment opportunities. These are:
zero. Therefore, it is difficult to precisely (i) the book-to-market ratio, (ii) expenditures
determine how much convexity, at the margin, on research and development, scaled by
is necessary to induce managers to invest in market value of assets, and (iii) a measure of
risk-increasing, positive NPV projects. investment expenditures defined as the sum
To illustrate the differences between of capital expenditures plus acquisitions over
the slope and convexity of the wealth– the most recent three years, divided by
performance relation, Panel C of Table 15.2 market value of assets. Though these vari-
reports descriptive statistics on the sensi- ables are expected to contain information
tivity of CEOs’ wealth to stock price. On about investment opportunities, each has
average, a stock option changes in value by unique limitations as a measure of this
$0.27 for a 1% change in stock price, unobservable underlying construct.Therefore,
compared to $0.39 for a share of common as in Baber et al. (1996) and Gaver and
stock. Thus, in contrast to the convexity Gaver (1993), I employ common factor
induced by each of these securities, a share analysis to construct a single variable that
of stock increases the slope of the wealth– captures variation common to these observ-
performance relation by about 40% more able proxies. Separate results are presented
than a stock option does, on average. On an using the common factor as an alternative
aggregate basis, both stock options and to the above three proxies.
common stock make a substantial contribu- Table 15.3 summarizes the proxies for
tion to the wealth–performance slope. The investment opportunities and the financial
median change in option portfolio value for characteristics of the 278 sample firms. In
a 1% change in stock price is $36,400, general, the sample firms are large, with
compared to a median change of $37,300 median market value of assets equal to
for common stockholdings. $5.4 billion. The range in values for book-
to-market, R&D, and investment expendi-
tures is quite large, suggesting that there is
5. CONVEXITY AND THE considerable variation in investment oppor-
INVESTMENT OPPORTUNITY SET tunities across the sample firms.The factor
score for investment opportunities has a
As illustrated in Section 2, risk-related mean of zero, by construction, and varies
agency problems can cause managers to pass from 0.85 to 2.27.
408 DIRECTORS’ REMUNERATION
The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat,
ranked by market value of equity on December 31, 1988. All financial variables are computed
for fiscal year ending December 31, 1993. Book-to-market ratio is the book value of assets divided
by the sum of book value of liabilities and market value of equity. R&D expenditures is R&D expense
divided by the market value of assets (in %). Investment expenditures is the sum of capital
expenditures plus acquisitions over the period 1991 through 1993, divided by market value of
assets. Factor score is obtained using common factor analysis on the variables Book-to-market,
R&D expenditures, and Investment expenditures. Market value of assets is book value of debt
plus market value of equity. Leverage is calculated as [(book value of assets – book value of equity)/
market value of equity].
Sensitivity of wealth to equity risk a b1(Book-to-market) b2(R&D) b3(Investment expenditures) c1 Log(market value of assets)
c2(Sensitivity of wealth to stock price) c3(Cash compensation) c4 Log(sensitivity of wealth to stock price)
c5 Age error.
All variables are measured as of December 31, 1993. Sensitivity of wealth to equity risk from options is the change in the dividend-adjusted Black-Scholes value
of a CEO’s stock option portfolio for a 0.01 change in the annualized standard deviation of the firm’s stock returns (in $ thousands). Sensitivity of wealth to equity
risk from options and stock is the change in the dividend-adjusted Black-Scholes value of a CEO’s stock option portfolio and common stockholdings for a 0.01
change in the annualized standard deviation of the firm’s stock returns (in $ thousands). Book-to-market ratio is (book value of assets/market vlaue of assets). R&D
expenditures is R&D expense divided by market value of assets (in %). Investment expenditures is the sum of capital expenditures plus acquisitions over the period
1991 through 1993, divided by the market value of assets at December 31, 1993. Factor score is obtained using common factor analysis on the variables Book-to-
market ratio, R&D expenditures, and Investment expenditures. Log[market value of assets] is the natural logarithm of [book value of liabilities market value of
equity]. Sensitivity of wealth to stock price is the change in the dividend-adjusted Black-Scholes value of a CEO’s stock option portfolio and common stockhold-
ings for a 1% change in the value of the firm’s stock price (in $ thousands). Cash compensation is the sum of salary plus cash bonus for 1993 (in $ millions). Age
is the CEO’s age in years. T-statistics in parentheses are computed using White’s heteroskedasticity-consistent standard errors.
Intercept (?) 82.62 (3.67) 112.37a (5.18) 71.05a (2.03) 77.29a (3.40) 103.04a (4.64) 80.80a (2.28)
Book-to-market ratio () 63.72a (4.61) 55.79a (3.76)
R&D expenditures () 3.10a (2.44) 3.29a (2.51)
Investment expenditures () 48.29 (1.63) 46.81 (1.54)
Factor score (investment () 27.76a (5.32) 19.46a (3.96) 26.44a (4.92) 17.12a (3.44)
opportunities)
log(market value of assets) () 18.19a (6.59) 17.13a (6.44) 11.00a (3.54) 17.10a (6.01) 16.26a (6.01) 9.61a (3.08)
Sensitivity of wealth to () 0.00 (0.02) 0.00 (0.18) 0.01a (2.08) 0.00 (0.74) 0.00 (0.85) 0.01b (1.70)
stock price
Cash compensation () 7.06 (1.48) 6.71 (1.44)
Log (sensitivity of wealth () 10.96a (6.11) 12.79a (6.84)
to stock price
Age () 0.70 (1.57) 0.41 (0.87)
Adjusted R2 22.47 20.76 31.77 20.12 19.25 32.59
To interpret the magnitude of the factor the greater the expected risk-aversion effect.
score coefficient, consider two CEOs with Though logarithm is the reported functional
factor scores that differ by the standard form of this variable, the regressions are
deviation of the sample firms’ factor not sensitive to using other concave func-
scores.The coefficient on the factor score in tional forms, such as the square root. Since
Column (5) implies that the compensation the same stock-based securities drive both
schemes for these two CEOs are expected the sensitivity of CEOs’ wealth to equity
to be set so that a ten percentage point risk and to equity value, a positive coeffi-
rise in stock-return volatility increases the cient is predicted.
stock-based wealth of the CEO managing When these variables are included, as
the high growth options firm by about shown in Columns (3) and (6), the coeffi-
$143,000 more than the CEO managing cient on log[sensitivity of wealth to stock
the low growth option firm. price] is significantly positive, indicating that
As noted in Section 2, the sensitivity of securities that increase the risk-aversion
CEOs’ wealth to equity risk, or the wealth effect also increase the wealth effect.
effect, is expected to measure the impact of However, the coefficient on the investment
risk on managers’ utility with error. Missing opportunities factor remains significantly
from this estimate is the risk-aversion positive in both columns, providing further
effect, or the reduction in utility that support for the hypothesis that firms
risk-averse, poorly diversified managers provide managers with incentives to invest
experience when the volatility of their in risky projects when the potential
wealth increases. To examine whether this loss from underinvestment in valuable
omitted variable is influencing the results, I risk-increasing projects is greatest.
include proxies for the risk-aversion effect Yermack (1995) concludes that invest-
in the regressions. ment opportunities are not an important
The dollar value of 1993 cash compen- determinant for grants of CEO stock
sation, salary plus bonus, is included to options. However, the sensitivity of wealth
control for the level of CEOs’ outside to stock price is used as the dependent
wealth. The greater the cash compensation variable in his tests.The results in Table 15.4
that can be invested outside the firm, the suggest that firms are expected to consider
better diversified the CEO is likely to be, the sensitivity of wealth to equity risk, in
and the lower the expected risk-aversion addition to sensitivity of wealth to stock
effect. CEO age is included to control for price, when granting stock options.
manager-specific variation in diversification
of wealth and degree of risk-aversion. 5.2. Specification checks
The results are robust to the following
alternative treatments of the age variable: Corporate hedging theory identifies circum-
(i) including dummy variables for various age stances where firm value can be increased
categories, (ii) interacting age with the other through reductions in risk (e.g., Myers,
independent variables, and (iii) re-estimating 1977; Smith and Stulz, 1985; Smith and
the regressions on subsamples of the CEOs Mayers, 1987, 1990; Froot et al., 1993).
formed by age. The signs of the coefficients All else being equal, firms with strong
on cash compensation and age are difficult incentives to hedge are expected to provide
to predict since it is unclear how the wealth managers with weaker incentives to increase
and risk-aversion effects are likely to be firm risk. To examine the sensitivity of
correlated for these variables.The log of the the reported results to these factors,
sensitivity of CEOs’ wealth to stock price is I include measures of financial leverage,
also added as a proxy for the CEOs’ degree operating performance, cash flow volatility,
of diversification. The more sensitive the and tax-loss-carryforwards in the Table 15.4
CEO’s wealth is to firm performance, the less regressions as proxies for firms’ incentives
well-diversified the CEO is likely to be, and to hedge. The coefficients and significance
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 411
Table 15.5 The relation between stock-return volatility and the sensitivity of CEOs’ wealth to
equity risk
Stock-return volatility is the annualized standard deviation of daily logarithmic stock returns (in
%) over 240 trading days, starting with the last 120 trading days in 1993 and ending with the first
120 trading days in 1994, multiplied by 252(1/2). Sensitivity of wealth to equity risk from options
and stock is the change in the dividend-adjusted Black-Scholes value of a CEO’s stock option
portfolio and common stockholdings (in $ thousands) for a 0.01 change in the annualized standard
deviation of the firm’s stock returns. Adjusted sensitivity of wealth to equity risk due to options
and stock removes the influence of firm-specific stock-return volatility by using the sample mean
standard deviation of stock returns in all Black-Scholes computations, instead of using the
equity risk specific to each firm. Log[market value of assets] is the natural logarithm of [book
value of liabilities market value of equity] at December-end 1993. Leverage is [(Book value of
assets – book value of equity)/market value of equity]. Factor score is obtained using common
factor analysis on the variables Book-to-market ratio, R&D expenditures, and Investment expenditures
as described in Section 5.
T-statistics are in parentheses and are computed using White’s heteroskedasticity-consistent standard errors.
a Statistical significance at the 5% level.
b Statistical significance at the 10% level.
positive. The regression results are robust the Black-Scholes partial derivatives used
to including the book-to-market ratio, R&D to estimate the sensitivity of CEOs’ wealth
expenditures, and investment expenditures to equity risk. To address this concern,
in place of the factor score as proxies for I recompute all Black–Scholes partial
investment opportunities. derivatives using the sample mean standard
Since the standard deviation of stock deviation of stock returns, instead of the
returns is an input in the Black-Scholes volatility specific to each firm. Though
model, it is possible that the regression this adjusted measure of sensitivity of
coefficients are influenced by a mechanical wealth to equity risk is not expected to fully
relation between stock-return volatility and reflect the incentive effects of managers’
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 413
Table 15.6 Description of parameters used in Black–Scholes computations for common stock
The sample consists of 278 CEOs selected uniformly from the 1000 largest firms on Compustat,
ranked by market value of equity on December 31, 1988. All data are for 1993. Per-share stock
price is the stock price on December 31. 1993. Per-share book value of debt is the book value of
liabilities as of December 31, 1993. Per-share market value of assets is per-share stock price plus
per-share book value of debt. Implied per-share market value of assets is computed using the
Black–Scholes model (see Section A.2 of the appendix).The standard deviation of equity returns is
the standard deviation of daily logarithmic stock returns over the last 120 trading days in 1993,
multiplied by 252(1/2). The standard deviation of debt is the standard deviation of monthly
logarithmic returns on the Merrill Lynch corporate bond index that matches the firm’s S&P senior
debt rating, multiplied by 12(1/2).The standard deviation of bond index returns is estimated over the
five-year period ending December 1993.The standard deviation of equity and debt returns are used
to compute the estimated standard deviation of returns on firm value () (see Section A.2 of the
appendix). The risk-free interest rate is the yield, as of December 31st, 1993, on a U.S. Treasury
strip with the same time to maturity as the weighted average maturity of the firm’s liabilities. The
weighted average maturity of liabilities is estimated as described in Section A.2 of the appendix.
Dividend yield is dividends per share paid during 1993, divided by the implied per-share market
value of assets. Price-to-strike ratio is the implied per-share market value/per-share book value of
debt. Sensitivity of stock price to a $1 change in firm value is the change in the dividend-adjusted
Black-Scholes value of common stock for a $1 change in the value of the firm (see Section A.2 of
the appendix).
T the weighted average maturity of the Mikkelson (the referee), Cathy Schrand, Clifford
firm’s liabilities estimated using Smith, Philip Stocken, Jerry Warner, Ross Watts,
Compustat data on corporate liabili- Jerry Zimmerman, workshop participants at
ties maturing in less than 1 year, Baruch College, University of Chicago, Columbia
University, Northwestern University, and The
2 years, 3 years, 4 years, 5 years, and Wharton School, and especially S.P. Kothari for
more than five years. When the firm helpful comments. I gratefully acknowledge
has outstanding debt with different financial support from the American
times to maturity, common stock is Compensation Association. Earlier drafts of this
technically a compound option. That work were titled ‘Compensation, Convexity, and
is, when a portion of the debt matures, the Incentives to Manage Risk’.
the stockholders have the option to 1 A 10 percentage point change represents
pay off that portion of debt and the standard deviation of the sample firms’
purchase an option to buy the firm for stock-return volatility, adjusted for industry and
the remaining book value of debt. To firm size. Stock-return volatility is measured
as the annualized standard deviation of daily
approximate the sensitivity of wealth stock returns.
to firm risk for this compound option, 2 Ofek and Yermack (1997) provide evidence
I make the simplifying assumption that that managers can influence the magnitude of
the firm has a single debt obligation their own stock-based wealth and degree of
with time to maturity equal to the diversification. This suggests a less important
weighted average time to maturity of role for the risk-aversion effect. However, to the
the firm’s debt. I assume a maturity extent that convexity in managers’ incentive
of ten years for the portion of debt schemes comes largely from non-portable
maturing in more than five years. For employee stock options, managers are less likely
most banks, utilities, and insurance to be able to influence the wealth effect.
3 Cuny and Jorion (1995) and Statement of
firms, maturity data for years two and Financial Accounting Standards No. 123,
over is unavailable. I assume these ‘Accounting for stock-based compensation’,
firms have an average maturity on address valuation issues resulting from vesting
long-term debt of 7.5 years, though the and portability restrictions. The difference
results are not sensitive to alternative between Cuny and Jorion’s ‘corrected’ option
maturity assumptions. value and Black/Scholes’ value is smallest when
r ln(1 risk-free interest rate), where options do not have vesting restrictions. Since
the risk-free interest rate is the yield, most of the options valued in this study are not
as of December 31, 1993, on a newly granted options, measurement error due
U.S. Treasury strip with the same time to vesting issues is not likely to be a serious
problem. Failure to consider portability restric-
to maturity as the weighted average tions is likely to lead to upward biased estimates
maturity of the firm’s liabilities. of option values and convexity. However, it is
d ln(1 expected dividend rate on firm much more difficult to determine how portability
value), where the expected dividend restrictions affect cross-sectional variation in
rate is set equal to dividends paid option values and convexity.
during 1993 divided by the implied 4 Note that this risk-related agency problem
market value of the firm. Note that is somewhat different than the well-known
the dividend rate on firm value is not underinvestment problem described by Myers
known ex ante. It is obtained in step (i) (1977). He demonstrates that when fixed claims
when the implied total market value of are present in the capital structure, equity holders
may forgo positive net-present-value projects if
the firm is computed. the gains accrue primarily to fixed claim holders.
The risk-related agency problem described here
does not require debt in the capital structure,
but instead derives from risk-averse managers
NOTES that are poorly diversified with respect to their
firm-specific wealth.
* I thank Ray Ball, Michael Barclay, John 5 Gaver and Gaver (1993) and Baber et al.
Core, Paul Fischer, Dan Gode, Jarrad Harford, (1996) provide further empirical support for
Christian Leuz, Jon Lewellen, John Long, Wayne this hypothesis. Yermack (1995) finds that the
THE SENSITIVITY OF CEO WEALTH TO EQUITY RISK 417
N. K. Chidambaran* and
Nagpurnanand R. Prabhala
EXECUTIVE STOCK OPTION
REPRICING, INTERNAL
GOVERNANCE MECHANISMS,
AND MANAGEMENT TURNOVER†
Source: Journal of Financial Economics, 69(1) (2003): 53–89.
ABSTRACT
We examine firms that reprice their executive stock options and find little evidence that
repricing reflects managerial entrenchment or ineffective governance. Repricing grants are
economically significant, but there is little else unusual about compensation in repricing firms.
Repricers tend to be smaller, younger, rapidly growing firms that experience a deep, sudden
shock to growth and profitability.They are also more concentrated in the technology, trade, and
service sectors and have smaller boards of directors. Repricers have abnormally high CEO
turnover rates, which is inconsistent with the entrenchment hypothesis. Over 40% of repricers
exclude the CEO’s options when they reprice.
We present new evidence on cross-sectional drops is that the control firms are never
determinants of repricing and on top explicitly identified in the data. We use sta-
management turnover in repricing firms. In tistical methods to account for imperfectly
doing so, we shed new light on the relevance observed control samples. Our approach
of the agency and poor governance exploits the information in variables such
explanations underlying much of the public as past returns or option grant prices that
criticism of repricing. We begin by charac- point to plausible control firms, recognizes
terizing the negative shock that leads that the instruments are imperfect, and
to repricing. Repricers experience negative produces a statistical comparison of
returns over the two years before the repricers with (never observed) control
repricing fiscal year. The return shock is firms. We further implement the method-
accompanied by a steep decline in prof- ology by conditioning directly on the
itability and sales growth that is not “underwater” nature of options.
reversed in the next two years.The evidence Second, our analysis contributes to the
provides a clear picture of the typical literature by introducing several new find-
repricer: it is a young and rapidly growing ings on the cross-sectional determinants of
firm that experiences an abrupt and firms’ repricing decisions. While previous
relatively permanent shock to growth work (Brenner et al., 2000; Chance et al.,
and profitability. 2000) finds that firm size is about the only
We analyze repricing in the context of variable that explains why firms reprice, we
overall executive compensation, focusing on uncover evidence that other industry and
top management. We examine whether firm-specific variables matter. Younger
repricings are economically significant firms, rapidly growing firms, firms belonging
compensation events by analyzing the to the technology, trade, and service sectors,
levels of compensation in repricers and the and firms with smaller boards are more
compensation changes that accompany likely to reprice. We also find no evidence
repricing. We find that repricing grants are that repricers have longer-serving managers,
economically significant in relation to both more diffuse stock ownership, or especially
overall pay levels and previous-year option low institutional ownership. Overall, the
grants, and they also significantly exceed evidence provides little support for the notion
option grants in matched peers of repricers. that repricing primarily manifests agency
However, compensation adjustments accom- problems or ineffective governance of firms.
panying repricing tend to be small, and We also develop evidence on top
compensation levels in repricers are similar management turnover in repricers. One
to those in matched peers. Thus, repricings criticism of repricing is that it reflects
are themselves economically significant the inability of shareholders to fire poorly
compensation events, but there is little else performing managers because these man-
unusual about the compensation levels or agers are entrenched. The entrenchment
changes in compensation that accompany hypothesis predicts that top management
repricing. turnover rates in repricers are abnormally
We then investigate cross-sectional low. However, we find the opposite result:
determinants of a firm’s repricing decision. repricers have abnormally high CEO
Given a shock to firm value, some firms attrition rates. There is little evidence that
choose to reprice their executive stock repricers are unable to effect changes in
options while others do not. What industry, management, as an entrenchment hypothesis
financial, and governance characteristics might suggest.
differentiate the two sets of firms? Our The last part of our paper documents
cross-sectional analysis adds to the and analyzes an interesting heterogeneity
literature in two ways. One contribution in executives included in firms’ repricing
is methodological. A key difficulty in com- announcements. In over 40% of all
paring repricers to control firms that do repricing announcements, the CEO is not
not reprice despite similar stock price included in the list of executives repriced.
422 DIRECTORS’ REMUNERATION
We find a pronounced imbalance in a repricing that took place after the fiscal
compensation structures in these “non- year but before the annual proxy filing as
CEO” repricers. Specifically, the CEO has an event pertaining to the proxy’s fiscal
fewer options and more direct shareholdings, year. We discard seven cases because the
while non-CEO executives have relatively repricing is caused by a merger, spinoff, or
more options than shares in these firms. regrant of options in a division instead of
Thus, a negative return shock in non-CEO options in a parent company. In four cases,
repricers weakens incentives of non- the repricing does not relate to named
CEO executives far more than those of executives of the firm but to lower employ-
the CEO. Here, repricing seems to play the ees. In one case, option strike prices are
role of mitigating the intra-management actually increased to the current market
incentive imbalances created by negative price and the maturity is extended, perhaps
return shocks. motivated by tax considerations. In another
We proceed as follows. Section 2 case, the executive bought options at the
describes the repricing sample. We charac- Black–Scholes value from the company.
terize the industry and time trends in Our final repricer sample has 213
repricing and the nature of the return and instances of repricing. The sample is larger
operating shocks that precede repricing. than that used by Brenner et al., who
Section 3 describes repricing grants in the analyze 133 repricings in EXECUCOMP
context of overall compensation policy. between 1992 and 1995. Our sample size
Section 4 conducts a cross-sectional analysis also significantly exceeds that used in
of firms’ decisions to reprice. Section 5 Chance et al. (2000), who analyze 53
analyzes CEO and non-CEO repricing. announcements between 1985 and 1994
Section 6 offers conclusions. identified through a LEXIS–NEXIS search.
For each year in which a firm reprices
executive stock options, we identify the
2. DATA executives whose options are repriced and,
in particular, note whether the CEO’s
Our repricing sample is based on Standard & options are repriced. To identify whether a
Poor’s, 1998 EXECUCOMP database, repriced executive is a CEO or not, we
which contains information for 51,555 match executive names to CEO titles deter-
executives for 1,836 firms for fiscal year mined as described below, and also verify the
1992 to fiscal year 1997. In this six-year data manually in the proxy. We also count
period, EXECUCOMP identifies a total of the number of executives for whom stock
864 executives who are repriced, which options are repriced for each repricing event.
translates into 240 separate firm-level For each firm in EXECUCOMP, we
repricing announcements. Following Brenner obtain dates when an executive assumes
et al. (2000), we read the proxy state- the CEO title and when the executive
ments, 10-K filings, or annual reports to leaves the CEO position. This allows us to
cross-verify the repricing event. For some identify the CEO in charge at the beginning
years, repricing is disclosed only in the of each year. When the dates are not
filings in subsequent years. In general, reported, we use EXECUCOMP’s classifi-
repricing-related reporting has become cation of CEO for that fiscal year. This
both more uniform and more comprehensive procedure identifies CEOs for 10,042
in recent years. firm-years. For each CEO, we obtain the
From the initial sample of 240 instances compensation package and the number of
of repricing in EXECUCOMP, we discard options and shares held. We also obtain the
27 cases; 14 of these contain reporting aggregate number of options and shares
errors. EXECUCOMP sometimes classifies held by all other executives named in the
an ESO repricing that occurred in a prior proxy. If the CEO of a firm changes during
year as belonging to the current year. a fiscal year, we classify the firm-year as
The database, for example, misclassifies having a turnover event. Some observations
EXECUTIVE STOCK OPTION REPRICING 423
for the 1997 fiscal year are missing from of 55 events in fiscal 1996, and 36 events
the 1998 EXECUCOMP data. We fill these in fiscal 1997. The number of executives
in from the 1999 EXECUCOMP database. repriced is relatively stable at a median
For the entire sample of firms, we obtain value of four. Table 16.1 also shows that
return data from the Center for Research in repricings not involving the CEO have
Security Prices (CRSP) and cross-sectional become more common over time, starting
data on firm characteristics from the from only one non-CEO repricing in 1992
annual COMPUSTAT tapes. We obtain to 22 in 1996.
data on the number of members serving on Table 16.2 presents the number of repric-
a company’s board of directors from ing announcements classified by industry.
Investor Responsibility Research Center We classify all firms in EXECUCOMP into
(IRRC). Board-related observations are 18 industry categories. Our primary classi-
available for about 75% of the sample of fication is according to two-digit SIC codes,
11,016 firm-years. Additionally, we obtain but in some instances, we use four-digit SIC
two measures of the fragmentation of firm codes, as this is more informative about the
ownership reported in the SPECTRUM types of companies that engage in repricing.
Disclosure database: institutional owner Unlike Brenner et al. and Chance et al.,
ship and aggregate shareholdings of 5% we find evidence of industry patterns in
block-holders of the firm. Both variables repricing. Firms in the computers and elec-
are available for slightly more than two- tronic parts, software and high-technology,
thirds of the sample. We merge the return, and biotechnology industries account for
firm characteristics, and governance data 80 of 213, or about 37.5%, of all repricing
with the repricing data set. events. Other industry segments with some
concentration of repricers are the trade
2.1. The repricing sample: industry (wholesale and retail) and service segments.
The technology, trade, and service sectors
and time trends
together account for over two-thirds of all
Table 16.1 presents an annual breakdown repricing announcements. On the other
of the repricers in our sample and the hand, repricing is rare in the heavy indus-
number of executives repriced at each tries such as utilities and mining.
announcement. Repricing seems to have The more detailed industry classifications
become more popular over time. There are appear to be helpful in identifying industry
15 repricing events in fiscal 1992, a peak patterns in repricing. Part of the gain in
Table 16.1 presents annual distribution of all repricing events reported in EXECUCOMP. Column 1
shows the total number of repricing events. Columns 2 and 3 show the number of repricing
events when the CEO is repriced and the number of repricing events when the CEO is not repriced,
respectively. Column 4 shows the median (mean) number of repriced executives named in the proxy
statements of the firm.
Number of
1992 15 14 1 4 (3.36)
1993 31 21 10 4 (3.59)
1994 33 17 16 4 (3.24)
1995 43 24 19 3 (4.02)
1996 55 33 22 4 (3.91)
1997 36 18 18 4 (3.87)
TOTAL 213 127 86 4 (3.74)
424 DIRECTORS’ REMUNERATION
Table 16.2 Repricing announcements by industry
Table 16.2 reports the distribution of repricing and non-repricing firms across industry segments.
We classify firms into 18 industry segments based on SIC codes, as detailed in the table.
power comes from working with four-digit from non-software firms in the services
SIC codes rather than two-digit SIC codes. sector.The detailed classifications, however,
This allows us, for instance, to separate out do not eliminate all problems. For instance,
biotech companies within the small manu- firms in the retail sector concentrating on
facturing sector and software companies software and computers might be better
EXECUTIVE STOCK OPTION REPRICING 425
classified as computer-related industries, starting from month 24 but not before.
but these firms are classified as retailers Wilcoxon z (p) values for differences
and share SIC codes all the way down between repricers and non-repricers for the
to four-digit levels with other retailers not periods [24, 18], [18, 12], [12,
in computer-related businesses. Without 6], and [6, 0] are 5.41, 10.12,
adhoc choices on our part, we cannot 11.68, and 6.59, respectively, all signifi-
reclassify these firms. cant at 1%, while Wilcoxon z (p) values
for differences in periods [30, 24]
and [36, 30] are 0.96 (p 0.34) and
2.2. The repricing sample: 1.21 (p 0.23), both insignificant.2
Dissimilarities in returns between repricers
pre-repricing returns
and non-repricers evidently surface about
Repricing firms usually experience a two years prior to the repricing fiscal
significant decline in stock prices over year-end and persist through the year of
some period of time prior to repricing. For repricing. Negative returns are dispersed
instance, Brenner et al. find that three-year over the two-year period even for the
returns prior to repricing are significantly repricing sample, so the data provide less
negative. We characterize this price drop in guidance on the magnitude of negative
six-month intervals over the three years returns or their timing within the two-year
before the repricing fiscal year-end. These period to qualify a non-repricer as a valid
findings help define control firms that control firm. We choose firms with returns
experience large negative price drops yet do of less than 15% in months [24, 6]
not reprice their ESOs. Table 16.3 reports or 30% in [6, 0] as our control sample.
the distributional features of six-month The former group of firms experience a
buy-and-hold returns covering three years relatively permanent shock to prices,
going back from time zero for repricers and while the latter group represents firms with
all non-repricers, where year zero is the large negative shocks in a recent period.
repricing fiscal year-end for repricers and We recognize that the definition is adhoc so
every fiscal year-end for every firm is a the control sample is probably imperfect
time zero for non-repricers. with errors of both inclusion and exclusion.
Repricers experience lower returns than (We address this issue through econometric
non-repricers in all six-month periods methods in Section 3.)
Table 16.3 presents data on the distribution of stock returns for repricing and non-repricing firms.
A repricer is a firm on the EXECUCOMP database that makes a repricing announcement in a fiscal
year, and a non-repricer is a firm on the EXECUCOMP database that does not make such an
announcement. Each row reports buy-and-hold returns (in percent) for several six-month periods
prior to the fiscal year-end of the repricing year, which is month zero. For each period, we report the
first quartile, median, third quartile, and the mean of the returns distribution.
2.3. The repricing sample: EBITDA ratio for all firms in the same
operating performance industry group and fiscal year. As recom-
mended by Barber and Lyon, our analysis
Panels A and B of Table 16.4 present data focuses on median rather than mean
on the operating performance of repricers abnormal performance, as mean ratios can
from year 2 through year 2, where year be distorted by outliers.
zero is the repricing fiscal year-end for Repricers perform at industry levels in
repricers and every fiscal year end for every year 2, when the median abnormal
firm is a year zero for non-repricers. Our EBITDA ratio equals 0.52%, not signif-
metric for judging operating performance is icantly different from zero; the Wilcoxon
the “EBITDA” ratio, i.e., earnings before z(p) 0.44 (0.65). However, the operat-
interest, taxes, and depreciation in year t ing performance of repricers tails off rather
(COMPUSTAT item # 13) divided by the sharply in the next two years.The abnormal
average book value of assets (COMPUSTAT EBITDA ratio in year zero is negative
item # 6) in years t and t 1 (Barber and (median 6.71%) and significant at
Lyon, 1996). Following Barber and Lyon, 1% (Wilcoxon z(p) 7.77 (0.00)) and
we compute the abnormal EBITDA ratio the performance decline is not reversed in
as the raw number less the median the next two years. The decline in repricer
Table 16.4 reports data on operating performance of three groups of firms. Repricers are firms on the
EXECUCOMP database that make a repricing announcement during a fiscal year. Non-repricers are firms
that do not make such an announcement during the fiscal year. Control firms are firms with a return not
exceeding 15% in months 24 through 6 or 6 through zero, where zero denotes the fiscal year-end
month. The columns in Table 16.4 report characteristics for the fiscal year and a window of two years
preceding and following the fiscal year. For each group of firms, we report EBITDA, defined as earnings
before interest, taxes, and depreciation (COMPUSTAT item # 13) as a percentage of the average book
value of assets (COMPUSTAT item # 6) in the fiscal year and the year preceding, abnormal EBITDA,
which equals EBITDA minus industry median EBITDA, sales, percentage sales growth, and industry-
adjusted sales growth (sales growth minus industry median sales growth). In each case, we report the
median (mean) of the relevant characteristic.
A: EBITDA
Repricers 15.79 (15.16) 13.02 (11.35) 9.58 (7.53) 11.31 (7.79) 11.35 (7.70)
Non-repricers 14.45 (14.78) 14.53 (14.84) 14.74 (15.05) 14.67 (14.93) 14.68 (14.85)
Control firms 15.42 (14.67) 12.47 (11.76) 11.41 (9.81) 11.67 (9.88) 12.78 (11.08)
B: Abnormal EBITDA
Repricers 0.58 (0.91) 3.49 (4.82) 6.71 (8.53) 5.20 (8.43) 4.88 (7.90)
Non-repricers 0.00 (0.56) 0.00 (0.58) 0.02 (0.57) 0.01 (0.46) 0.01 (0.32)
Control firms 0.20 (0.77) 2.29 (3.63) 3.60 (5.64) 3.29 (5.54) 2.20 (4.37)
C: Sales
Repricers 222 (639) 276 (723) 320 (717) 334 (709) 355 (796)
Non-repricers 652 (2627) 708 (2768) 803 (2994) 923 (3265) 984 (3417)
Control firms 380 (1475) 417 (1544) 462 (1652) 500 (1814) 540 (1954)
D: Sales growth
Repricers 28.16 (47.12) 22.22 (53.7) 3.61 (13.26) 5.85 (13.73) 8.06 (27.28)
Non-repricers 8.93 (17.30) 9.13 (17.43) 10.07 (18.71) 10.15 (18.15) 10.39 (17.52)
Control firms 14.77 (33.91) 9.38 (20.06) 6.57 (16.24) 6.87 (15.04) 8.36 (16.53)
E: Industry-adjusted sales growth
Repricers 13.78 (34.43) 7.60 (40.89) 7.26 (0.82) 5.14 (2.09) 1.11 (15.84)
Non-repricers 0.07 (7.40) 0.00 (7.35) 0.02 (7.86) 0.00 (7.38) 0.00 (6.51)
Control firms 4.17 (22.31) 1.17 (8.76) 4.33 (4.38) 3.5 (3.29) 2.48 (5.07)
EXECUTIVE STOCK OPTION REPRICING 427
Table 16.5 reports compensation levels and changes for a sample of 126 CEOs whose stock options
are repriced between 1992 and 1997 and corresponding compensation variables for firms matched
by size decile, industry, and fiscal year. Panel A reports salary, bonus, the Black–Scholes value of
option grants, and total dollar compensation in the fiscal year prior to repricing. Panel B reports
the dollar changes in salary and bonus between the repricing and prior year. Panel C reports the
Black–Scholes value of (a) the total repricing-year grants, (b) fresh replacement options granted
in lieu of the options canceled in repricing, and (c) new option grants. Panel C also gives the
change in the Black–Scholes value of the new option grants relative to the previous-year grants.
Columns 1 and 2 report data for repricers and matched firms, respectively, while Column 3 gives
Wilcoxon z (p) values for testing differences between the two. For each characteristic, we report
the median (mean).
NO SIGNIFICANT STOCK
PRICE DECLINE
1–p1
REPRICE p1*p2
Figure 16.1 Model of the repricing decision as a sequential two-stage process, where p1 is the
probability that firms experience a significant stock price decline and p2 is the proba-
bility that firms choose to reprice conditional on such a decline. Non-repricers either
do not experience a significant price decline [probability (1 p1)] or choose not to
reprice despite such a price decline [probability (p1 *(1 p2))].
1800
1600
1400
1200
1000
800
600
400
200
0
1993 1994 1995 1996 1997
Figure 16.2 Median annual values by year for salary, bonus, option grants, and total compensation,
across all firms in EXECUCOMP.
430 DIRECTORS’ REMUNERATION
3500
3000
2500
2000
1500
1000
500
0
Decile 1 Decile 2 Decile 3 Decile 4 Decile 5 Decile 6 Decile 7 Decile 8 Decile 9 Decile 10
Figure 16.3 Median annual values by size decile for salary, bonus, option grants, and total
compensation, across all firms in EXECUCOMP. Firms are classified into ten size
deciles (based on the book value of assets) of firms. Decile 1 and Decile 10 consist of
the smallest and the largest firms, respectively.
1200
1000
800
600
400
200
0
Trade Services Technology Other
Figure 16.4 Median annual values by industry for salary, bonus, option grants, and total
compensation, across all firms in EXECUCOMP. Firms are classified into four
industries: trade, services, technology, and other.
for the top-decile firms versus $503,800 Industry: The data in Section 2.1 and
for the lowest-decile firms. Similar trends Section 3 show that repricers tend to con-
are also apparent in all components of centrate in specific industries, particularly
CEO compensation. trade, services, and technology. Figure 16.4
EXECUTIVE STOCK OPTION REPRICING 431
depicts total compensation and its elements remain somewhat lower for repriced CEOs
in these three sectors and corresponding (median $71, 030) than in matched
compensation variables in all other indus- firms (median $159, 200). This finding
tries taken together. Salary, bonus, and is not surprising, following Core and Guay
total compensation are lower in the (2001) and Yermack (1995), who argue
technology, services, and trade sectors that firms facing high cash needs and with
compared to other sectors, and the differ- costly external finance reduce their use of
ences are significant at 1% or better cash compensation. Repricers are small
(z and p values not reported here). and rapidly growing firms that have experi-
Additionally, option grants are significantly enced severe, permanent operating shocks,
more dominant forms of compensation in the and plausibly need to conserve resources to
technology sector, where options constitute recoup. Finally, repriced CEOs have about
a median of 31% of pay versus 14–20% the same levels of overall compensation
for other industry sectors. as in firms matched by industry, size, and
The mechanics of adjusting for size, fiscal year.
industry, and time are as follows. We divide
all firms in EXECUCOMP, year by year,
3.2. Compensation changes
into ten size deciles formed on the basis of
the book value of assets (from COMPUS- In this section, we examine adjustments in
TAT) at the beginning of the fiscal year. We compensation variables that tend to
also classify firms as belonging to technol- accompany repricing. Panel B in Table 16.5
ogy, trade, services, or other industry reports median and mean salary and bonus
sectors. For each firm, we then compute changes for repriced CEOs in the fiscal
matched peer salary, bonus, option grant, year of repricing and corresponding statis-
and total compensation as the median of tics for industry-, size-, and year-matched
the relevant CEO compensation variable for firms. The median salary change accompa-
all firms in the same size decile, industry nying repricing is $11,850 versus $18,340
group, and fiscal year. Panel A in Table 16.5 for matched non-repricing peers, and a
reports compensation variables for similar difference exists for the means.
repriced CEOs and matched peers. It also The difference is economically small but
reports Wilcoxon z (p) values for testing significant at 5% (Wilcoxon z (p)
differences between repriced CEOs and 2.00 (0.05)). The median bonus change
their matched peers. is zero versus $7,600 for matched peers.
The matching procedure does attenuate Again, the differences are economically
some of the differences between repricers small but statistically significant at 5%
and all non-repricers. Median total dollar (Wilcoxon z (p) 2.11 (0.04)). The
compensation (TDC) of repriced CEOs low levels of bonuses and their apparent
($941,700) is closer to the TDC of the downward rigidity suggest that repriced
peer-matched sample ($916,500) than CEOs have bonuses that amount to floor
that of all non-repricers ($1,197,000) and amounts and are probably not a signifi-
TDC differences are no longer significant. cant component of discretionary annual
Likewise, salaries (median $375, 000) performance-linked incentive pay.
are closer to and not significantly different Dealing with the third element of com-
from peer-matched salaries (median pensation, option grants, is more intricate.
$369, 500), while they are significantly Option grants in the repricing year consist
lower than salaries in all non-repricers of two distinct components: (a) options
(median $450, 000). Previous-year stock granted in lieu of the canceled (repriced)
grants (median $247, 400) tend to be options; and (b) additional new grants over
somewhat higher than those in matched and above the repriced options. The two
firms (median $151,000), but as before, components must be disentangled to assess
the difference is only significant at 10% the magnitude of the repricing grant and the
(Wilcoxon z (p) 1.67 (0.09)). Bonuses size of new grants that accompany repricing.
432 DIRECTORS’ REMUNERATION
This breakup, however, faces two obstacles. $123,100. This is somewhat lower than
First, option grants reported in proxy matched firm grants, as seen in Table 16.5,
statements (and EXECUCOMP) do not and also lower than previous-year grants
separately identify the number of repriced (median $247, 400), although the dif-
options. A second issue is the lack of ferences are only significant at about 10%.
uniformity in reporting practices. Some The median ratio of current-year to previous-
firms seem to regard repricing as just an year grants is 67% and is not significantly
alteration of existing options.These firms do different from 100% (Wilcoxon z (p)
not include repriced options in the repricing- 1.38 (0.19)). Thus, firms appear to
year options grant. Other firms view repricing maintain or reduce the number of new
as a cancelation of old options and regrant options granted in the fiscal year of
of fresh replacement options. These firms repricing. The value of repriced options
do include the replacement options as (median $870, 100) significantly exceeds
part of the repricing-year option grants. the value of the previous-year grant
Thus, straight grant numbers reported (median $247, 400) with Wilcoxon
in EXECUCOMP and proxies are not z (p) 4.36 (0.00). Likewise, the ratio of
comparable across firms. the number of repriced options to previous-
We read proxies or annual reports of year grants has a median of 182%
repricers to obtain the number of repriced (mean 284%) and significantly exceeds
options and also use these proxies to ascer- 100% (Wilcoxon z (p) 5.11 (0.00)).
tain whether repriced options are included The evidence suggests that repricing is in
in the grants table. Sufficient information is fact a significant addition rather than an
available for 109 out of 127 cases, mostly alternative to options granted in the normal
in footnotes to the option grants table. In course by firms.
about 75% of the sample, repricing-year A related question of some interest is
grants include repriced options. Here, we whether large option grants are associated
deflate the total repricing-year grant by with poorly performing firms in general, in
the number of repriced options included which case we should not view repricing as
in the grant to obtain new option grants. In a special event. To address this issue, we
other cases (about two dozen), repricing- examine option grants for the control firms
year option grants explicitly exclude repriced with negative returns used in Section 2 and
options. Here, new options directly equal Table 16.3. The median (mean) current-
grant figures reported in EXECUCOMP and year grants for this sample equal $247,600
we obtain the number of repriced options ($807,600). These are slightly below the
separately by reading proxy statements. previous-year grants for the same firms
Panel C in Table 16.5 reports the value (median (mean) equal to $264,300
of total grants, repriced options, and new ($889,400)), and also significantly below
options, and the change in option grant rel- the value of options repriced.5 Repricing is
ative to the previous year’s option grant. therefore special; it is not the case that
The median Black–Scholes value of the total poorly performing firms that do not reprice
repricing-year grant, which incorporates instead give large option grants in lieu of
both repriced options and new grants, is repricing.
$1,222,000.This far exceeds both matched An alternative to repricing is to issue
grant values (median $218,700) and new options but leave the old, out-of-the-
previous-year grants (median $247, 400). money options untouched. Firms might be
Our findings are consistent with those of reluctant to use this alternative because the
Brenner et al., who also find that the total number of options are capped at
EXECUCOMP reported Black–Scholes levels pre-authorized by shareholders. It can
value of repricing-year grants is unusually be difficult for firms to make large grants
high for repriced executives. similar to those involved in ESO repricing
Turning to the components of the total without seeking specific shareholder approval
repricing-year grants, the median value of for increasing the cap, whereas repricing
the new grants accompanying repricing is preserves decision-making at the board level
EXECUTIVE STOCK OPTION REPRICING 433
without having to get shareholder approval.6 and by introducing new variables into the
Leaving old options untouched also has the cross-sectional analysis. The expanded
disadvantage that it could adversely affect analysis permits, in particular, a more
the future diluted EPS of the firm. This detailed investigation of alternative expla-
alternative can be more attractive, however, nations for repricing, particularly the
following the October 1999 FASB ruling agency explanation that underlies much of
that companies must expense costs of the public criticism of repricing.
repricing because repriced options are Section 4.1 discusses univariate compar-
deemed to have “variable” strike prices. isons of repricers with non-repricers and
Taxes are unlikely to play a role in this with control firms that experience negative
choice, as tax events in ESOs are triggered returns but do not reprice. Section 4.2
only at eventual option exercise. reports estimates of multivariate specifica-
Finally, the above analysis focuses on tions in which repricers are compared
compensation changes of repriced CEOs. to control firms, accounting for the unob-
However, several firms reprice other servability of control firms. Section 4.3
members of the top management team but discusses an alternate definition of control
not the CEO. We check whether CEOs in firms that conditions directly on the
these firms receive backdoor compensation underwater nature of outstanding options.
through unusual salary, bonus, or grant Section 4.4 offers a summary of the results
changes. The data do not support this in relation to alternative explanations for
hypothesis. For non-CEO repricers, the ESO repricing.
median (mean) change in CEO salary,
bonus, and option grants is $12,290
($13,600), $0 ($72, 370), and 4.1. Univariate comparisons
$22,270 ($1, 050, 000), respectively, Table 16.6 presents cross-sectional data
versus $17,570 ($20,340), $8,000 for repricers, all non-repricers, and control
($16,460), and $57,690 ($232,600), firms. Column 1 reports the data for
respectively, for matched peers, with repricers, Column 2 reports data for all
Wilcoxon z (p) values for differences equal non repricers, and Column 3 reports data
to 1.66 (0.10), 3.52 (0.00), and for control firms. Column 4 reports the
0.36 (0.72), respectively. There is little nonparametric Wilcoxon (or Mann–
evidence that firms give backdoor compen- Whitney) z-statistics and p-values for
sation through other means as a substitute testing differences between repricers and
for a conventional strike reset repricing. the control sample. The variables in Table
In sum, compensation levels in repricers 16.6 are organized into four categories:
match those in peers, and changes in com- (i) CEO turnover in repricers; (ii) three
pensation variables accompanying repricing proxies for firm size, a focus of much
are economically small.The repricing grant, previous work; (iii) data on several other
however, is large and comparable to annual firm and governance characteristics includ-
total compensation levels.Thus, the repricing ing firm age, ownership, and board size;
grant itself is significant to the repriced and (iv) data on executive equity and
executive but there is little else unusual in option holdings.
the accompanying compensation changes.
Table 16.6 reports the number of CEO turnovers and several firm-specific characteristics of three
groups of firms. Repricers are firms on the EXECUCOMP database that make a repricing announce-
ment during a fiscal year. Non-repricers do not make such an announcement during the fiscal year.
Control firms are firms with returns not exceeding 15% in months 24 through 6 or 6 through
zero, where zero denotes the fiscal year-end month. The last column reports z and p values for
Wilcoxon tests that compare characteristics of repricers with those of control firms.The firm-specific
characteristics are the annual sales, book value of assets, market value of equity, shares and executive
stock options held by all executives named in the firm’s proxy statement, the firm’s age (defined as the
time in months since listing), the tenure of the CEO (in years), the aggregate institutional ownership
and aggregate ownership of 5% block-holders in a firm, and the number of members in a firm’s board
of directors. For each characteristic, we report the median (mean).
Besides providing a test of the Other explanations for small firm repricing
entrenchment-agency explanation for do not rely on agency hypotheses. Small
repricing, the turnover findings also have firms tend to be younger firms with less
interesting implications for contingent- well developed lines of management, so
claims style valuation models of executive the costs of replacing managers could be
stock options (see, e.g., Carpenter, 1998). higher for these firms. Renegotiation of
ESO valuation models attempt to incorpo- compensation contracts might also be
rate features of executive stock options easier in small firms because these firms
that are not present in standard option val- are organizationally less complex, which
uation models. One stream of this literature makes for faster decision-making.
accounts for the repricing feature of execu- It is not clear which of the above
tive stock options (e.g., Brenner et al.). A explanations accounts for the size effect,
different set of papers deals with turnover but the data certainly suggest that we should
and its effect on ESO valuation (e.g., Cuny control for size in the multivariate analysis.
and Jorion, 1995). Our evidence suggests Following Brenner et al., we use sales as a
that ESO valuation models should incorpo- proxy for firm size. As the empirical distri-
rate both repricing and turnover features bution of firm size is skewed, we use the
simultaneously, because conditions that natural logarithm of firm size in regression
lead firms to reprice also appear to specifications.
increase the probability of CEO turnover.
Valuation models that ignore the probability
4.1.3. Firm age
of higher turnover while incorporating the
repricing feature can overstate the value of We measure the age of the firm as the
the reset feature in executive stock options. number of months since the first price for
the firm is reported in the CRSP tapes.
Younger firms are less likely to have well
4.1.2. Firm size
developed lines of succession, so manage-
We report three proxies for firm size— ment turnover is likely to be costly for
the book value of all assets of a firm, the young firms. If executive retention moti-
firm’s sales, and the market value of a vates repricing, young firms might be more
firm’s equity. We measure all variables as likely to reprice. The evidence in Table 16.6
of the beginning of the fiscal year. By all is consistent with this explanation. Repricers
three metrics of size, repricers are smaller tend to be younger, with a median time
than non-repricers and control firms. For since listing of about seven years versus
instance, the median sales of all non- about 19 years for all non-repricers and
repricers are about 2.0–2.5 times the 11 years for control firms.
median repricer sales, as in Brenner et al.
It is difficult to unambiguously interpret 4.1.4. CEO tenure
why small firms are more likely to reprice.
One possibility is that repricing reflects Long-serving CEOs are perhaps more
agency problems. Small firms perhaps face entrenched and have more pliable boards
less scrutiny from the financial community of directors in place. If an entrenchment-
and institutional ownership, which could expropriation hypothesis explains repricing,
curb rent-seeking behavior, is perhaps lower repricers should have high CEO tenure. The
in small firms. On the other hand, agency data in Table 16.6 provide little support
problems are more pronounced in larger, for such a hypothesis. CEO tenure is
mature firms that have more diffuse own- not statistically different between
ership and greater separation between repricers and control firms, or between
ownership and control. A finding that small repricers and the universe of non-repricers.
firms reprice more often is actually incon- In each case, the median CEO tenure is
sistent with the idea that repricing reflects about seven years.This result is also invari-
shareholder–manager agency problems. ant to alternate definitions of long-serving
436 DIRECTORS’ REMUNERATION
CEOs, such as CEOs who have served for at with greater market valuation of firms,
least five years. suggesting that smaller boards are more
shareholder value oriented. If repricing
reflects shareholder–manager agency prob-
4.1.5. Institutional and block lems, repricers should have greater board
ownership sizes, to the extent that board size is
symptomatic of the existence of such
We examine the aggregate level of institu- agency problems.
tional ownership in a firm. Firms with low The last row in Table 16.6 reports
institutional ownership are perhaps subject median and mean board sizes of repricers,
to less intense monitoring, so the balance of non-repricers, and control firms. We find
power between shareholders and managers little evidence that repricers have larger
may tilt more towards management in these boards: in fact, the evidence suggests
firms. If poor monitoring explains repricing, exactly the opposite. The median board in
repricing will be more concentrated in firms repricers has seven members compared to
with low institutional holdings. We find no nine members for all non-repricers and
evidence to support this contention. Repricers eight for control firms. The differences in
have median institutional ownership of board size are significant at 1%. To the
about 49.05%, which is slightly but not extent that small boards reflect greater
significantly higher than the 47.34% orientation towards shareholder value
institutional ownership in control maximization, repricing is unlikely to be an
firms. Institutional ownership is somewhat outcome of ineffective governance.
higher in the universe of all non-repricers,
consistent with higher institutional
involvement in large firms. 4.1.7. Equity and option holdings
In addition to institutional ownership, of executives
we also obtain one other proxy for owner-
ship fragmentation available in the CD Firms in which executives have large option
SPECTRUM database—the aggregate holdings as compared to direct equity
shareholdings of all 5% block-holders in a ownership are probably more likely to
company. This variable is inversely related reprice. One explanation is that negative
to the fragmentation in a firm’s ownership. return shocks cause greater misalignment in
If shareholder expropriation explains managerial incentives when executives have
repricing, firms with more diffuse and more options rather than straight equity.
fragmented ownership should be more Another explanation is that, whatever be
likely to reprice. We find little support for the reason for ESO repricing, it is worth
this hypothesis. The median aggregate the effort only when option holdings are
block holding in repricers (26.12%) is not sufficiently large.
significantly different from that in control Table 16.6 reports the number of options
firms (23.69%). and shares held by all named executives
appearing in company proxy statements.
We normalize all share prices to $25 so
4.1.6. Board size firms with beginning-of-year share prices
greater than $25 have the number of share
The number of members in a firm’s board and options scaled up, while those with
of directors can be viewed in two ways in share prices below $25 are scaled down, to
relation to repricing. One is that smaller make share and option holdings more com-
boards offer greater flexibility and speed in parable across firms.The data in Table 16.6
renegotiating contracts with managers. show no significant difference between the
This suggests that smaller boards should be shares held by all executives in repricing
more likely to reprice. A different perspective firms from those held by all executives in
of board size is offered by Yermack (1996), control firms. The data do indicate that
who finds that small boards are associated repricers have more options outstanding
EXECUTIVE STOCK OPTION REPRICING 437
relative to control firms. Therefore, stock median repricer market value but control
options are not lower when executive share- firms are only 5% larger than repricers.
holdings are high, as in Yermack (1995). Absent control firm comparisons, it is hard
Options seem to be important elements of to say whether repricer versus non-repricer
compensation and incentives in repricers differences reflects characteristics peculiar
regardless of executive shareholdings in to repricers or those generally associated
these firms. with firms that experience large price drops.
Are repricers the types of firms in which
one might expect greater use of options in 4.2. Multivariate tests
executive compensation? The characteristics
of repricers in Table 16.6 fit the profile of In this section, we report evidence from
firms in which we expect options to be multivariate tests to simultaneously assess
used more extensively. Repricers are small, the role of industry- and firm-specific
young, and rapidly growing firms that have factors in differentiating repricers from
high internal demands for cash, and non- firms that experience negative return
cash compensation (such as stock options) shocks but elect not to reprice.
is more likely to be used in such firms. Implementing such a comparison is
Further, since repricers are young firms, afflicted by an identification problem.
executives are less likely to have exercised Figure 16.1 illustrates this issue. While we
their option grants. An industry effect in can observe firms that make a repricing
repricers—specifically, their concentration announcement, non-repricers could either
in the technology industry—also explains be firms that never experience significant
the more widespread use of options. In the price declines or control firms that experi-
1990s, technology firms simultaneously ence such declines but choose not to
witnessed rapid growth and a short supply reprice. We could identify the latter by
of management talent, and used options specifying that control firms should have
widely as “golden handcuffs” to facilitate experienced past returns below a negative
management retention since options almost return cutoff, say r*, as in Table 16.6.
always vest over periods of three to four However, such control samples would
years. For instance, Auspex Systems Inc. probably be imperfect because the cutoff
writes in its proxy statement dated return and the period over which it is
November 20, 1997 that “...competition defined are essentially arbitrary choices.
for qualified employees in the technology The control sample could also require more
market is extremely intense, and, due to the complex definitions involving multiple
rapid growth of many successful companies return cutoffs over several prior periods.
in this sector, such competition is increas- Finally, including some firms as controls
ing . . .” and “. . . the Board of Directors while altogether ignoring others ignores
believes that having a stock [option] plan variation within a selected control sample.
in place is vital to retaining, motivating, and For instance, a firm with a return of
rewarding employees, executives and con- 40% is probably more likely to belong to
sultants, by providing them with long-term the control sample than a firm with a
equity participation in the company relating “borderline” return of 15%. However,
directly to the financial performance and this differential is not reflected in the
long-term growth of the company.” standard control sample procedure, whereby
The evidence developed thus far emphasize both firms receive equal weights as members
two main issues. First, there are several of the selected control sample.
systematic differences between repricers and We use statistical methods based on
both the universe of all non-repricers “partial observability” probit models
and control firms. The evidence also under- (e.g., Abowd and Farber, 1982) to address
lines the value of comparisons with control unobservability of control samples. Roughly
firms. For instance, the median market speaking, the technique compares repricers
value of all non-repricers is 216% of to a “weighted average” of non-repricers,
438 DIRECTORS’ REMUNERATION
with higher weights assigned to firms that weights, while returns in [24, 18] and
are more likely to be appropriate control [6, 0] are less important in precipitating
firms, i.e., firms with more negative histories a repricing decision. Weights on returns in
of past returns. The weights, however, need periods prior to t 24 are not signifi-
not be specified by the researcher; the cant, so we do not include these in the final
technique estimates the weighting function specifications.
endogenously as part of the likelihood The second-stage estimates identify
function and produces a statistical what types of firms are more likely to
comparisons of repricers with the (never respond to negative return shocks by
observed) control firms. repricing their executive stock options. In
Formally, the procedure takes as input the baseline specification, Model 1, we
two sets of variables. In the first stage, the include four industry dummies consisting
input is a vector of instruments, say x1i of a technology group (our industry groups
(e.g., past six-month returns), that help 8, 17, and 18), services (group 15), trade
determine whether firm i belongs to the (groups 13 and 14), and industrial
appropriate control sample of firms that manufacturing (group 7). To the industry
have negative returns but choose not to dummies, we add firm size, proxied by the
reprice. In the second stage, the input is a natural logarithm of sales. We also include
second set of variables, say x2i (such firm age (our proxy being the time since
as firm size, age, board size, etc.) that first listing on the CRSP tapes), which is
potentially differentiate repricers from negatively related to management turnover
non-repricers, conditional on a negative costs. Finally, we add the number of options
shock.The likelihood function for the model held by all executives listed in EXECUCOMP
can be written as for the firm, as a proxy for the demand
for repricing. The baseline specification
L(x1, x2) Repricers[p1(x1)*p2(x2)] estimates confirm that repricing concen-
Non-Repricers[1 p1(x1)*p2(x2)] (1) trates in technology, services, trade, and
manufacturing relative to all other sectors
The likelihood function is derived by not- such as construction and utilities. Dummies
ing that repricers have received a negative for technology, services, and trade are
return shock large enough to precipitate con- 2.5–3.5 times the dummy for manufactur-
sideration of repricing (probability p1) and ing, indicating a greater preponderance of
have chosen to reprice (probability p2) in repricing in the first three sectors relative
response to this shock, to give a compound to the latter. As in Brenner et al., firm size
probability of p1 *p2, the first term in is important: small firms are more likely to
Eq. (1). Firms that have not announced a reprice conditional on a negative return
repricing either do not receive a large shock. Firm age is also significant and
negative return shock (probability 1 p1), remains so across all specifications. The
or receive such a shock but choose not to number of options is positively related to
reprice (probability P1 *(1 p2)). The total the conditional probability of repricing.
probability of these two possibilities is Models 2–4 sequentially add other
1 p1 p1 *(1 p2) 1 p1 *p2. The explanatory variables to the baseline speci-
likelihood function follows. fication. Model 2 adds the change in the
Table 16.7 provides estimates of several industry-adjusted EBITDA ratio (similar
two-stage models fitted to the repricing results obtain with differences in unad-
firms. The first-stage estimates determine justed EBITDA ratios) from year 2 to
what types of firms are likely to be in the the year of repricing, and the change in
control sample. As expected, the procedure sales growth from year 2 through the
weights returns in six-month periods repricing year. As indicated in Table 16.4,
between t 24 and t 0 negatively. profitability and sales growth shocks are
Returns in the time interval [12, 6] and more severe for repricers compared to
[18, 12] have the largest (absolute) control firms, and we test this proposition
EXECUTIVE STOCK OPTION REPRICING 439
Table 16.7 reports estimates of a two-stage probit model fitted to firms on the 1998
EXECUCOMP database, to compare repricers with firms that face similar return shocks
but choose not to reprice. The dependent variable is one if a firm announces a repricing
in a fiscal year and zero otherwise. Independent variables for the first stage include six-
month returns over four periods beginning in month 24, where month zero denotes the
fiscal year-end. Independent variables for the second stage are dummies for membership
in the technology, services, trade, and manufacturing sectors, the natural logarithm of
annual sales, the natural logarithm of the number of options held by all named executives,
firm age (years since first listing), the change in the EBITDA ratio and annual sales
growth over the last two fiscal years, and the number of members in a firm’s board of
directors. Numbers in parentheses denote t-statistics.
Variables Model 1 Model 2 Model 3 Model 4
First stage
Intercept 1.28 (12.86) 1.15 (11.19) 1.33 (14.11) 1.23 (10.84)
Return [24, 18] 0.38 (4.36) 0.37 (3.91) 0.41 (4.17) 0.44 (10.84)
Return [18, 12] 0.97 (13.37) 0.88 (11.72) 1.14 (9.31) 1.08 (9.10)
Return [12, 6] 1.33 (14.10) 1.24 (11.44) 1.43 (9.86) 1.36 (8.41)
Return [6,0] 0.49 (4.51) 0.40 (3.47) 0.49 (3.14) 0.42 (3.14)
Second stage
Intercept 1.36 (2.13) 1.34 (2.23) 1.62 (1.77) 1.28 (1.46)
Technology 1.71 (4.14) 1.49 (4.46) 1.88 (3.30) 1.69 (3.58)
Services 1.45 (3.85) 1.23 (3.70) 2.15 (3.72) 1.55 (3.22)
Trade 1.45 (4.51) 1.42 (4.66) 1.58 (3.59) 1.44 (3.59)
Manufacturing 0.45 (1.97) 0.44 (1.86) 0.61 (1.76) 0.58 (1.79)
Log (sales) 0.34 (3.61) 0.40 (4.17) 0.14 (1.14) 0.21 (1.67)
Firm age (years) 0.05 (5.06) 0.05 (4.82) 0.06 (4.26) 0.05 (4.21)
Log (#options) 0.07 (2.36) 0.06 (2.32) 0.05 (1.14) 0.04 (1.10)
EBITDA 2.25 (1.98) 1.22 (0.71)
Sales growth 0.64 (2.67) 0.72 (2.05)
Board size 0.16 (2.91) 0.10 (1.94)
Model significance
2 (d.f.) 434 (12) 491 (13) 392 (13) 394 (15)
industry dummy variables for technology, provides additional justification for the
services, and trade continue to be significant, 40% out-of-moneyness underlying the
as does firm age. Variables other than firm Appendix A computations.8
size, notably industry membership and firm We reestimate the two-step probit speci-
age, do matter in explaining the conditional fication of Section 4.2, using the strike
probability of repricing given a negative price ratio instead of past returns as the
return shock. first-step instrument pointing towards
plausible control firms. The results are
4.3. Control sample based on qualitatively similar to those in Table 16.7
and are not reported here. As expected, we
approximate moneyness7
find that the strike price ratio has a nega-
Section 4.2 uses prior stock returns to tive coefficient and is significant at better
identify control firms that do not reprice than 1%. This indicates that firms with
despite negative return shocks that plausi- lower strike price ratios, i.e., those with
bly cause their options to go underwater. more out of money options, are more likely
This section attempts to further refine to be at the node at which a repricing deci-
control sample comparisons by conditioning sion must be made. With regard to the sec-
more directly on the underwater nature of ond-step estimates, the significant changes
outstanding ESOs. are in firm size and the change in the
The key obstacle in assessing the money- EBITDA ratio in the full specification.
ness of outstanding ESOs is that disclosures While firm size altogether loses its mar-
do not report data on strike prices of ginal (10%) significance, the change in the
outstanding options. This is, of course, a EBITDA ratio now becomes significant at
well known issue afflicting all empirical better than 1%. (The results are available
research on executive stock options. upon request.)
However, an approximation is possible
based on option grant prices reported in 4.4. Summary of cross-sectional
proxies (see also Carter and Lynch, 2001). results
Since options are normally granted at-
the-money, the difference between strikes Our cross-sectional analysis provides an
of current grants and lagged grants is a extensive investigation into factors that
plausible proxy for the moneyness of explain repricing. We contribute several
previously granted options. new results to the repricing literature by
Accordingly, for each firm-year we examining a richer slate of variables. Two
compute the weighted average of strike aspects of our results deserve special com-
prices of option grants, with weights for a ment. First, we show that several variables
given strike price proportional to the other than firm size explain why firms
number of options granted at that price. reprice. Our results also provide a more
From Table 16.3, return shocks to repricers thorough evaluation of agency-based criti-
are spread over 24 months prior to the cisms of repricing, and they cast some
repricing fiscal year-end. Hence, the ratio of doubt on the notion that repricing primarily
the current-year strike to the two-year-back manifests shareholder–manager agency
lagged strike is a rough measure of the problems or ineffective governance of
moneyness of previously granted options. firms. Specifically,
For our sample of repricers, median and
mean strike price ratios equal 56.27% • Repricers are young and rapidly
and 69.20%, respectively, suggesting that growing firms concentrated in the
the median repricer has options that are technology, services, and trade sectors
about 40% out-of-the-money. This accords (together accounting for over 70%
well with the 40% figure reported in of all repricings). These are not
both Brenner et al. and Chance et al., and traditionally viewed as the types of
EXECUTIVE STOCK OPTION REPRICING 441
Table 16.8 reports data on turnovers and several characteristics of firms in the 1998 EXECUCOMP
database that announce a repricing between 1992 and 1997. Column 1 reports data for firms that
reprice their CEO, while Column 2 gives data for firms that reprice executives other than CEOs.
Column 3 reports z and p values for Wilcoxon tests that compare the two sets of firms.
Characteristics reported here include annual sales, book value of assets, market value of equity,
shares and options held by the CEO and other executives in the proxy statement, the difference
between the options (as a fraction of option plus equity) of the CEO and other executives, the firm’s
age (months since first listing), CEO tenure in years, the aggregate ownership of institutions and 5%
block-holders, and the number of members in a firm’s board of directors. For each characteristic, we
report the median (mean).
Non-CEO Wilcoxon z
Variable CEO repricers repricers (p-value)
repricing fiscal year, with respect to whom Institutional ownership is actually higher in
firms must make repricing and turnover CEO repricers, though not significantly so,
decisions. Columns 1 and 2 give characteris- while block ownership is comparable to con-
tics of CEO repricers and non-CEO repricers, trol firms. There is no compelling evidence
respectively. Column 3 reports Wilcoxon of poor governance in either sample.
z and p values for testing differences The data do, however, show pronounced
in characteristics of CEO repricers and asymmetries in the structure of the equity
non-CEO repricers. As in Section 4, the holdings of CEOs relative to other execu-
cross-sectional data in Table 16.8 are tives in CEO versus non-CEO repricers. In
organized into four categories: CEO turnover, non-CEO repricers, CEOs have more straight
firm size, other firm characteristics, and equity compared to options, while other
equity and options holdings. executives have relatively more options
The CEO turnover data emphasize that compared to equity. In CEO repricers, the
on average, both CEO repricers and non- tilt of equity versus options for CEOs
CEO repricers are not particularly averse compared to other executives is not as
to initiating CEO changes. In neither CEO pronounced. The asymmetry in CEO and
repricers nor non-CEO repricers do we find other executives’ equity holdings relative
evidence of abnormally low turnover rates. to options is captured by the variable
For non-CEO repricers, the one- and two- “% Options (non-CEO) – % Options
year attrition rates are 30% and 36%, (CEO)” in Table 16.8. To compute this
respectively.The one- and two-year attrition variable, we first calculate the aggregate
rates for CEO repricers are 11.81% and number of options held by executives other
18.89%, respectively.The one- and two-year than the CEO, and express this as a per-
turnover rates for these firms significantly centage of the total number of options and
exceed turnover rates for all non-repricers, shares they hold. We then compute the
which are 6.29% and 12.57%, respectively same measure for the CEO, and subtract
(from Table 16.6). For both samples, CEO the option percentage for the CEO from the
turnover rates are also at least as high option percentage for non-CEO executives.
as the rates for control firms, which are As Table 16.8 shows, the difference in the
10.26% and 18.21%, respectively. There option percentage is positive for both CEO
is little evidence that turnover rates in repricers and non-CEO repricers, which
either sample are abnormally low, as an indicates that executives have greater pro-
entrenchment hypothesis would suggest. portions of their equitylinked compensation
Other cross-sectional characteristics of in options compared to CEOs. However, the
CEO repricers and non-CEO repricers also variable is significantly larger for non-CEO
reveal no evidence of expropriation or repricers (median 21.99%) than for
poor governance. We find no significant dif- CEO repricers (median 4.78%). This
ferences in size, as measured by sales, indicates that for the sample of non-CEO
assets, or market value of equity, between repricers, executives other than the CEO
CEO repricers and non-CEO repricers. CEO have more options relative to straight
repricers tend to be older (87 months) than equity holdings, while CEO repricers have
non-CEO repricers (78 months), but both more equity relative to options.
sets of firms are much younger than the The imbalance in options relative to equity
control group of firms (126 months, from between the CEO and other executives
Table 16.6). We also detect no statistically provides one possible explanation for the
significant differences in CEO tenure: CEO incidence of non-CEO repricing. A price
repricers have a median tenure of 6.32 years decline has a differential impact on the
versus 7.48 years for non-CEO repricers and value of options and the value of shares.
6.56 years for control firms. Board sizes in Further, when the price decline is such
both samples are roughly equal, with a that the options end up out-of-the-money,
median board size of seven members; executives may place very little value on
both types of firms have small boards. them (Hall and Murphy, 2000). Therefore,
444 DIRECTORS’ REMUNERATION
if executives have more options compared that exclude the CEO from repricing are
to equity, a negative return shock misaligns more likely to experience CEO turnover.
incentives for non-CEO executives more Table 16.9 reports probit estimates in
than it does for the CEO. The CEO’s incen- which the incidence of CEO turnover in
tives are less adversely affected because the repricers is modeled as a function of a
CEO has relatively more equity compared firm’s decision to include the CEO in the
to options. Non-CEO repricing should, there- list of executives repriced. Controls include
fore, be more common when equity-linked firm size, industry dummies for technology,
compensation is option-oriented for execu- service, trade, and manufacturing, and
tives but straight equity-oriented for CEOs, the number of shares held by the CEO,
which is what we find. This would suggest since firms with high CEO shareholdings
that repricing has a “re-incentivizing” role, are less likely to experience CEO turnover
in which it corrects incentive imbalances (see, e.g., Denis et al., 1997). Column 1
within the top management team that arise reports estimates of a specification that
from a negative return shock. includes a zero/one CEO repricing dummy
In computing the variable “options,” we as an explanatory variable. The coefficient
normalize the current stock price level to for CEO repricing is negative and signifi-
$25. This accounts for differences in stock cant at 5%.
price levels across firms, so options on a The dummy variable specification we
stock trading at $30 are counted as being report in the first column of Table 9 ignores
three times as valuable as options on a stock the endogeneity of the CEO repricing
trading at $10. However, the normalization decision. Ignoring selection-bias potentially
does not account for other differences such leads to inconsistent parameter estimates
as differences in option maturities across (see, e.g., Greene, 1993, Chapter 22). We
firms or across-firm variation in stock return correct for the selection-bias using the
volatilities. These differences could be two-step procedure suggested in Heckman
important and result in variation in option (1979). In the first step, we estimate a
values across firms (see, e.g., Guay, probit model for the CEO repricing deci-
1999). Appendix A presents two alternate sion. In the second step, the inverse Mills
approaches to calculating the dollar value ratio from the probit estimate replaces the
of option holdings and recalculates CEO repricing dummy in the turnover regres-
the variable “% Options (non-CEO) – % sion. Columns 2 and 3 report the results
Options (CEO)” under these measures. The from the two-step procedure. Column 2
option-equity asymmetry between CEOs and reports the turnover probit estimates with
non-CEO executives in CEO repricers the selection-bias correction. Column 3
and non-CEO repricers persists under the reports estimates of the first-step probit
alternate measures of compensation asym- model for whether the CEO is repriced with
metry as well. past returns, industry, firm age, and size,
and the difference between the option
percentage of the CEO and other executives
5.2. CEO repricing and turnover as explanatory variables. The negative sign
in ESO repricers for repricing persists in this selection-bias
As Table 16.8 shows, the rates of CEO corrected specification as well. Thus, ESO
turnover for non-CEO repricers is higher repricers that reprice their CEOs are less
than the rates of CEO turnover for CEO likely to witness CEO turnover.
repricers. The difference in CEO attrition
rates suggest that within firms that reprice
their ESOs, the decision to reprice a CEO is 6. CONCLUSION
related to the incidence of CEO turnover.
Specifically, repricers that include the CEO The repricing of executive stock options has
among repriced executives are less likely to received considerable attention recently,
witness CEO turnover, while ESO repricers much of it unflattering. Institutional
EXECUTIVE STOCK OPTION REPRICING 445
Table 16.9 reports estimates of probit models fitted to 213 firms on the 1998 EXECUCOMP
database that reprice their executive stock options. The dependent variable is one if there is CEO
turnover in the repricing year and zero otherwise. Independent variables include six-month returns
for four non-overlapping periods beginning in month 24 where month zero is the repricing fiscal
year-end, industry dummies for technology, services, trade, and manufacturing sectors, the natural
logarithm of sales, firm age (years since listing), the natural logarithm of the shares held by
the beginning-of-year CEO, a zero/one dummy variable for CEO repricing, and the inverse Mills
ratio based on probit estimates in Column 3. The dependent variable in Column 3 is one for a CEO
repricing and zero otherwise, and independent variables include, additionally, the option percentage
(the number of executive options divided by the options plus the number of shares) for the non-CEO
executives minus that for the CEO. Numbers in parentheses denote t-statistics.
Model 2
Independent variables
Intercept 1.99 (2.51) 2.46 (3.19) 0.85 (1.83)
Return [6, 0] 0.27 (0.92) 0.19 (0.63) 0.08 (0.32)
Return [12, 6] 0.81 (1.78) 0.89 (1.97) 0.35 (1.17)
Return [18, 12] 0.70 (1.71) 0.63 (1.53) 0.15 (0.61)
Return [24, 18] 0.02 (0.10) 0.02 (0.07) 0.32 (1.39)
Technology 0.86 (1.34) 0.89 (1.37) 0.55 (1.57)
Services 0.66 (0.94) 0.66 (0.93) 0.06 (0.16)
Trade 1.43 (2.12) 1.36 (2.01) 0.31 (0.74)
Manufacturing 0.43 (0.63) 0.34 (0.49) 0.38 (1.00)
Log (sales) 0.14 (1.42) 0.15 (1.56) 0.05 (0.65)
Log (# CEO shares) 0.11 (3.77) 0.10 (3.03)
Age firm 0.04 (1.39)
% Options (non-CEO) –
% Options (CEO) 0.64 (2.65)
CEO repriced 0.66 (2.69)
Inverse Mills ratio 0.40 (2.56)
Model significance
χ211 (p-value) 38.41 (0.00) 35 (0.00) 22.88 (0.02)
investors, in particular, often frown upon the literature in three ways. First, we
this practice, as it seems to reward present evidence on whether repricings are
managers despite poor performance, pre- economically significant compensation
cisely when they perhaps ought to be fired events. We find that while repricing grants
instead. It is therefore natural to inquire are large, and in fact, comparable to total
whether repricing represents an indictment annual compensation levels, compensation
of poor shareholder governance systems, changes accompanying repricing tend to be
reflecting entrenched and possibly over- small, and compensation levels in repricers
compensated managers expropriating are not abnormally high. Thus, repricing
shareholder wealth. is itself significant to the repriced execu-
We contribute to the repricing debate tive, but there is little else unusual in the
by developing new and extensive empirical compensation patterns in repricers.
evidence on over 200 repricing events Our second contribution is methodolog-
between 1992 and 1997. Besides a larger ical. In comparing repricers with control
sample of repricings, our paper adds to firms that do not reprice despite negative
446 DIRECTORS’ REMUNERATION
returns, a key issue is that control firms are We develop some first evidence on CEO
not explicitly identified in the data except turnover in repricers, in perhaps the most
through imperfect conditioning instruments direct examination of the entrenchment
such as past returns or option grant prices. explanation for repricing. If repricing
We implement statistical methods to account reflects the inability of shareholders to fire
for unobserved control firms.The technique poorly performing managers because
exploits information in the conditioning managers are entrenched, top management
instruments but recognizes that the instru- turnover in repricers should be abnormally
ments are imperfect pointers and produces low. In fact, repricers have abnormally
a comparison between repricers and (never high rates of top management turnover.
observed) control firms. We also implement Over 25% of firms making a repricing
the technique by conditioning directly on the announcement experience a CEO change
underwater nature of options rather than either in the year of repricing or the year
through poor past returns. after. This attrition rate exceeds that in
Finally, we expand the literature through both the universe of non-repricers and
a richer cross-sectional analysis of the control firms of non-repricers with poor
repricing decision. We present several new past returns. We thus find no evidence
findings that shed light on alternative that repricers are especially reluctant to
explanations for repricing. In particular, we initiate top management changes, as an
offer an extensive evaluation of the agency entrenchment hypothesis suggests.
and ineffective governance explanations that Finally, our paper presents some new
underlie much of the public criticism of evidence on within-sample variation in
repricing. We find, as in previous work, that repricers. We find that a significant 40%
repricers are small firms. However, unlike proportion of repricings are “non-CEO”
previous work, we find that other variables repricings, in which repricers do not include
matter in explaining the repricing decision, their CEOs in the list of executives repriced.
even after controlling for firm size. In these instances, repricing may redress
We find that the typical repricer is a within-management incentive distortions
young, rapidly growing firm that experiences created by negative return shocks. We
a steep and somewhat abrupt shock to its also document that within repricers, CEO
growth that is not reversed in the following turnover is lower at firms that include the
two years. Repricers are more likely to belong CEO in their repricing. Thus, the inclusion
to the technology sector, which accounts for of the CEO in a repricing lowers the
close to 40% of all repricings in our sample. probability of turnover while exclusion of
Repricers are also more likely to come the CEO is likely to be associated with
from the services and trade sectors, and CEO departure.
together the three sectors account for over Repricing is a key governance issue for
two-thirds of all repricing announcements. institutional investors, who must decide
There is no evidence that institutional own- whether to support this compensation prac-
ership in repricers is low or that the tice. Much of the controversy surrounding
ownership structure is especially fragmented repricing arises because repricing appears
in repricers, or that CEOs in repricers are to confer rewards upon poorly performing
entrenched by virtue of being long-tenured. executives, perhaps symptomatic of agency
Finally, repricers tend to have smaller problems, entrenchment, or poor governance
boards of directors. Collectively, our cross- of firms. The broader picture, however, tells
sectional results reveal no evidence that a different story. Cross-sectional character-
repricing is concentrated in firms whose istics of repricers, their compensation
shareholder–manager agency problems are policies, the incidence of non-CEO repricing,
especially prominent, nor do the results and the high rate of top management
reveal any compelling evidence that repricers turnover in repricers do not support the
are poorly governed. In fact, by at least one idea that repricing is primarily a manifes-
metric, board size, repricers appear to be tation of poor governance or managerial
better governed firms. entrenchment.
EXECUTIVE STOCK OPTION REPRICING 447
(non-CEO) – % Options (CEO).” The approval (see, e.g., the case of General Data
median (mean) difference in option percent- Comm Industries, Investor Relations Business,
age between CEOs and non-CEOs is equal February 15, 1999).
to 2.31% (5.04%) for CEO repricers, 2 Prior returns of repricers and non-repricers
overlap in calendar time. Positive correlation
which is not significantly different from
across two samples tends to understate signifi-
zero (Wilcoxon z (p) 0.71 (0.41)), cance, but correlation within samples can
while the difference for non-CEO repricers overstate significance levels. To mitigate within-
has a median (mean) value of 15.97% sample correlation, we also compute returns as an
(14.02%) and is significant at 5% excess over median industry, time period, and size
(Wilcoxon z (p) 2.09 (0.04)). The ratio decile matched portfolio returns. Patterns in these
difference between CEOs and non-CEOs is excess returns are similar to those in raw returns.
significant at 5% (Wilcoxon z (p) 2.17 Differences between repricer and matched returns
(0.03)). The option-equity asymmetry for the six prior periods going back from [6, 0]
between CEOs and non-CEO executives in have z statistics of 4.38, 10.05, 8.30,
3.51, 0.25, and 0.99, respectively; the first
CEO repricers and non-CEO repricers
four are significant at 1% but not the last two
remains under this third measure of (p 0.80 and 0.32, respectively).
compensation asymmetry as well. 3 The abnormal EBITDA and sales growth
statistics for non-repricers show that industry
adjustment on the lines of Barber and Lyon
NOTES (1996) produces well-behaved statistics. The
median abnormal operating performance (Panel
† We are grateful to G. William Schwert (the B) and abnormal sales growth (Panel E) are
editor) and two anonymous referees for very close to zero and economically and statistically
helpful comments. We also thank Mary Ellen insignificant. The mean statistics are positive,
Carter, Jonathan Karpoff, Scott Lee, Kai Li, suggesting that the metrics are positively
Arvind Mahajan, Vojislav Maksimovic, Thomas skewed, as in Barber and Lyon (1996).
Noe, Gordon Phillips, S. Abraham Ravid, 4 We are grateful to an anonymous referee for
K. Geert Rouwenhorst, and seminar partici- motivating and developing this section
pants at the Tuck/JFE Contemporaneous 5 While current-year grants in these control
Corporate Governance Conference, the 2000 firms are lower than the repriced option grants,
AFA and 2001 AFE meetings, Louisiana they exceed the new option grants in repricers;
State University, Texas A&M University, median (mean) $123, 100 ($612, 200). This
Tulane University, University of Massachusetts, can be attributed to the fact that repricers scale
Amherst, University of Virginia, and Yale back grants of new options in the repricing
University for their comments. Kristina Minnick fiscal year.
and Subhankar Nayak provided excellent 6 For example, the board in Bell Sports Corp.
research assistance. Special thanks to Kose sought ways to stay within the cap.The company
John and David Yermack for extensive feedback writes in its November 1996 proxy that senior
on an earlier draft. managers with long tenure were asked to cancel
* Corresponding author. Rutgers Business 40% of existing stock options to increase stock
School-Newark and New Brunswick, Rutgers options available for grants to other employees.
University, 94 Rockafeller Road, Piscataway, 7 We are grateful to an anonymous referee for
NJ 70118, USA. Tel.: 1-732-445-4446; suggestions on developing and implementing
fax: 1-732-445-2333. E-mail address: this analysis.
chiddi@rci.rutgers.edu (N.K. Chidambaran), 8 The requirement of lagged strike prices
nprabhal@rhsmith.umd.edu (N.R. Prabhala). means that we lose data for fiscal 1992 and
1 In one unsuccessful instance, the State of 1993, the first two years of data in EXECU-
Wisconsin Investment Board (SWIB) attempted COMP. We develop a second strike price ratio
to force a computer networking company, Shiva proxy by augmenting the data with the two-year
Corporation, to include such a proposal in the back market price of the stock from CRSP
annual proxy statement, but failed to do so. whenever lagged strikes are missing. Repricers
More recently, however, SWIB has successfully have similar median and mean strike price
pushed through binding bylaw resolutions ratios (58.61% and 72.05%, respectively)
that block repricing without shareholder under this alternative measure.
EXECUTIVE STOCK OPTION REPRICING 449
ABSTRACT
levels has increased. Popular publications despite high performance. The article cites
regularly reveal the compensation of the top 1997 German salaries of top executives
executives and particularly the chief executive including:
officers (CEO) of the large corporations.
The Securities and Exchange Commission ● Jurgen Schremp of DaimlerChrysler
(SEC) requires companies to reveal at 3 million DM;
compensation components for top execu- ● Bernd Pischetsrieder of BMW at
tives. Since US executives are the highest 3.5 million DM;
paid in the world, the US has the largest ● Ulrich Hartmann of VEBA at
economy, and data are readily available in 2.9 million DM; and
the US, most popular and academic attention ● Heinrich Von Pierer of Siemens at
has focused almost exclusively on the United 2 million DM.
States. However, there has been increased
interest in executive compensation in other These compensation levels seem low
countries, and particularly in Germany.2 compared to equivalent executives in the
Several bodies of literature have US such as:
developed in order to determine whether
management is taking advantage of the ● Bob Eaton of DaimlerChrysler at
lack of ownership control. The first body of 20 million DM;
literature examines the relationship between ● John Welch of GE at 69 million DM;
executive compensation and corporate ● Andrew Grove of Intel at 89 million
performance, while the second relates DM; and
corporate performance to the degree of ● Richard Scrushy of Healthsouth at
control that owners have over managers. Our 181 million DM.
study contributes to the relatively smaller
third body of literature relating corporate Although this difference between the two
control directly to executive compensation countries has been narrowing recently as
and thus directly tests the agency issue. German firms have increasingly turned to
Section 2 reviews some of the most granting stock options to top executives
relevant studies to date. (Kroll, 1999), corporate control systems
Though we know less about executive remain quite different in the two countries.3
compensation in countries other than the One main difference is that many German
US, researchers are currently investigating firms have a 2-tiered board system and
very similar issues in some of these coun- also many firms have representation of
tries, although data availability has banks on the supervisory board (SB) of the
impeded this process. Germany has the firm. All the firms in our sample do in fact
third largest economy in the world and have a 2-tiered board system. For many
consequently is of considerable interest. reasons then, it is quite important to
Internationalization has progressed identify the factors affecting executive
recently at a rapid pace and German firms compensation in Germany and to assess
are more commonly merging with firms the effect of corporate control mechanisms
from other countries. Executive compensa- on executive compensation.
tion becomes a particularly important issue In order to analyze this issue we have
when a German firm merges with an compiled a unique and comprehensive data
American firm, such as in the case of the set in order to analyze German executive
Daimler–Benz merger with Chrysler. compensation. This data set is more exten-
Corporate structure differs greatly between sive than data used in previous studies of
the two countries. Do American CEOs get German compensation. This enables us to
paid more than their German counterparts? contribute to the literature, based largely
If so, can this be justified? A recent article on American data, by analyzing the factors
in the German magazine Focus has claimed affecting executive compensation and the
that German executives are underpaid effects of ownership control on executive
452 DIRECTORS’ REMUNERATION
and size, that have been found to be Increased attention has been paid to
important determinants of executive corporate governance issues recently in
compensation. other countries. However, there are only a
A second line of research in executive limited number of academic empirical
compensation relates corporate perform- investigations of corporate governance in
ance to ownership dispersion. Demsetz and other countries. Germany has probably
Lehn (1985) find no relationship between been the country in which the greatest
profitability and ownership concentration interest has been shown. Germany has more
using a linear model. However, Morck et al. of a bank dominated financial system in
(1988) find a significant non-monotonic contrast to the market dominated system
relationship between both ownership con- in the United States. A number of studies,
centration and profit rates and ownership such as Rubach and Sebora (1998), Kim
concentration and Tobin’s Q—a measure of (1995), and Emmons and Schmid (1998),
market valuation. As ownership concentra- have discussed corporate governance issues
tion increases, Tobin’s Q rises, falls, and in Germany and compared them to the
then rises again. United States. Chirinko and Elston (1996)
The last series of studies is most closely systematically evaluate German bank influ-
related to our study. These studies directly ence and find that bank influence does not
test the agency problem with respect to lower finance costs nor have any discern-
executive compensation. Stigler and able effect on profitability. However, they
Friedland (1983) examine the relationship also find that control problems are
between shareholder control and executive addressed by concentrated ownership and
compensation for a sample of 92 firms for bank influence. Weigand and Lehmann
1937–1938. They find no significant (1998) examine the impact of ownership
relationship between the average salary of structure on the performance of German
the top three executives and the percent of firms, as measured by return on assets,
stock held by the top 20 shareholders. from 1991 to 1996. They find that corpo-
Santerre and Neun (1989) replicate this rate structure only affects performance for
study, and after including profits as an firms not quoted on the stock exchange. In
additional independent variable, find a addition, they show that blockholder own-
negative relationship. Both Santerre and ership results in higher returns and that
Neun (1986) and Dyl (1988) use more there are systematic interactive influences
recent data to test the agency problem with of ownership concentration and the
respect to executive compensation. Using identity of owners on performance for the
only data for CEOs, both studies find a non-quoted companies. Franks and Mayer
negative relationship between ownership (1998), however, do not find a significant
concentration and CEO compensation. relationship between board turnover and
More recently, Goldberg and Idson (1995) patterns of ownership.
examined the agency question for Fortune Schwalbach (1991) was one of the first
500 companies. This study estimated the studies to examine the effects of manage-
relationship between the dispersion of ment interests and profits on compensation
corporate ownership and the compensation in Germany. Using 468 firms in 1988 from
of top executives across the executive the Bonn Data, he shows that size and
hierarchy and for different components of industry effects are important in determin-
the compensation package in contrast to ing executive compensation. Conyon and
the more limited previous studies.The study Schwalbach (1999) examine executive pay
found a statistically significant agency effect, outcomes in the UK and Germany. Their
though it was small relative to company study finds a positive and significant
size. The effects were greatest for salaries, relationship between pay and company
the most liquid form of remuneration, and performance in both countries. Finally,
the executive with the strongest effect using 83 firms from 1968–1990 and 220
was the Chairman of the Board. firms from 1988–1992, Schwalbach and
454 DIRECTORS’ REMUNERATION
Grasshoff (1997) find that firm size and responsibilities of American boards of
industry are important determinants of directors.The SB appoints the MB and also
compensation for the management board sets the compensation of the members of
but that performance itself plays only a the MB. The MB is the top management
minor role. of the corporation and has responsibility for
The two previous studies most similar to the day-to-day operation of the corporation.
our current study are FitzRoy and Commercial banks have a significant influ-
Schwalbach (1990) and Schmid (1997). ence on corporate governance in contrast to
The first study, using annual data for 95 the United States where the banks are
firms from 1969–1985, finds a negative prohibited in general from directly owning
effect of concentrated ownership on the equity in corporations.
average annual salary of the management In Germany only certain legal identities
board. Ownership is measured by a are required to have supervisory and
Herfindahl index of equity capital distribu- managing boards. These include the com-
tion in 1982. Since these data are only mon AG (Aktiengesellschaft) or stock held
available for one year, the same ownership firm, as well as the more rare corporate
distribution variable is employed for every form, the KgaA (Kommanditgesellschaft
year for each company in the sample. Schmid auf Aktien).4 German law dictates that if
(1997) provides an alternative approach to an AG has fewer than 2000 employees,
the issue by examining the relationship for then the employees elect one third of the
1991 between compensation on the two members of the SB and two thirds are
boards of 110 of the largest 120 traded elected by the shareholders. If the AG has
German companies and shareholder struc- more than 2000 employees, the shareholders
ture and firm performance. He tests various appoint half of the SB while the firm
hypotheses and finds evidence that the employees select the other half. The chair-
compensation of both boards is affected by person of the SB is elected by a two-thirds
performance and by the firms’ shareholder vote of all board members for a maximum
structures. His measure of ownership term of five years, with the possibility of
concentration uses a Herfindahl index for reappointment. The law dictates the size of
110 firms in 1991 and has some advan- the SB for firms with more than 2000
tages over the measure that we use here. employees. The SB must have 12 members
However, his study only covers one year for firms with 2000–10,000 employees,
while our study contributes to the literature 16 members for firms with 10,000–20,000
by conducting a dynamic analysis of corpo- employees, and 20 members for larger
rate governance effects on compensation firms with over 20,000 employees. The
for a broad panel of German firms over a owner appointed members are frequently
17-year time period. bank representatives and the employee
members are frequently labor representa-
tives. Since SB members are paid an
3. GERMAN CORPORATE honorarium fee, and all members are paid
GOVERNANCE STRUCTURE the same, we are more interested in the
relationship of MB compensation and
There are two main aspects of the German performance. Moreover, the members of the
corporate governance structure that are dif- MB are comparable to the top executives
ferent from the system in the United States. including CEOs of American corporations
The first is that many large companies in that have been studied previously.
Germany have a 2-tiered board structure A controversial attribute of the German
comprised of the Aufsichtrat or SB and the corporate governance system is the extent
Vorstand or managing board (MB). We are to which the German universal banks
primarily interested in the MB rather than impact the governance of the firm. While
the SB. The SB has oversight responsibil- little empirical evidence exists to measure
ities for the corporation similar to the the precise extent of corporate control of
EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 455
board and compensation for the average definition in Section 5, and 0 otherwise.The
board member. These measures are chosen refutable hypothesis is that banks will exert
because the actual compensation is not influence and moderate executive compen-
available for each individual board member. sation. Therefore, a negative coefficient is
SB members are compensated equally, but expected.
MB members are paid differentially. We also test the effects on executive com-
However, these compensation differentials pensation of different types of ownership.
do not approach the differentials among We employ dummy variables to indicate
American executives. whether the firm is owned by another firm,
We hypothesize a positive relationship owned by foreign interests, owned by a bank,
between log Compensation and ROE and or family owned. Government control, mixed
also a positive relationship between log ownership, and management control (no
Compensation and log Sales.This is consis- large blocks) comprise the omitted cate-
tent with the findings of previous studies gory. Because bank influence has a close
that have found that executives are paid relationship with bank ownership, we run
more when the company is more profitable regressions including these dummy variables
and when the company is larger. both with and without the bank influence
We cannot employ a Herfindahl index as variable. We expect executive compensation
done by Schmid (1997) and FitzRoy and to be less when there is strong ownership
Schwalbach (1990) because complete data control in these four categories.
on ownership structure are not available Since agency problems may differ by
for each firm for the 17-year span of our industry, there may be some heterogeneity
data panel. This is because those who own in compensation practices across firms. A
less than 25% of a firm are not required to final analysis examines bank influence by
disclose ownership in Germany. We do, industry group.
however, have complete interval data on
ownership percentages greater than 25%
for each firm for every year of the study. 5. DATA
We therefore construct an alternative
concentration measure based on the One of the greatest impediments to meas-
ranked categorical source data on owner- uring the impact of corporate governance
ship concentration. The highest numbered on executive compensation in Germany has
concentration category represents the high- been the lack of reliable and comprehensive
est degree of owner concentration. Initially data.This study uses two non-commercially
we develop a variable that incorporates the available sources of unique data.The first is
rank of the concentration class, and for this a firm-level database that tracks the
variable we would expect a negative coeffi- financial performance of a comprehensive
cient in the regression model because the set of German firms from 1961 to 1986.
agency problem is expected to be most The second data source is derived from
important when ownership is more dispersed. various sources, including Commerzbank’s
That is, in the categories with greater Wer Gehort zu Wem?, Handbuch der
dispersion, management may be able to Grossunternehemen, Leitende Maenner und
extract greater compensation. Since this Frauen der Wirtschaft, Aktienfuehrer,
formulation using the ownership categories various annual reports of the firm, and
might assume a particular relationship, Boehm (1992). The combined information
we alternatively employ a model with in these collected data exceeds the depth
dummy variables for each category, omit- and breadth of commercially available
ting the category with the most dispersed data for Germany, enabling us to directly
ownership. examine the importance of corporate
The bank influence variable is a dummy governance on compensation. The bank
variable that takes a value of 1 when a firm influence variable for example is a composite
is determined to be bank influenced, per our measure using information on percent of
EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 457
bank ownership of the firm, percent of because of the substantial changes in the
bankers on the SB and whether they are German accounting laws, which make it
chair of the SB, and total number of votes impossible to compare the data without
exercised by banks at annual shareholder considerable adjustment.12 The sample
meetings.5 time period of 1970–1986 is important not
We operationalized this information by only as point of comparison with current
defining a dummy bank influence variable. stock of US studies available for this period,
A firm is characterized as bank influenced but also to understand the pay–performance
rather than independent if the bank owns relationship during this period for Germany.
more than 25% of the shares of the firm, The number of firms in the sample is also
if total votes of banks at shareholder meet- fairly representative because the German
ings (including proxy votes) are greater exchange is considerably smaller than its
than 50%, or if total votes are between American counterpart. For example in
25% and 50% and the chair of the SB is 1980 there were only about 459 listed
a banker.6 This discrete measure is possibly firms incorporated as AG and KgaA.13
an imperfect measure of the influence of For these 91 firms we have annual firm-
banks on firms, however it does succeed in level information detailing the identity of
incorporating all of the available information firm owners. These discrete data are
on bank–firm relations for each of these reported in mutually exclusive categories
firms over each of the 17 years, thereby defining the identity and concentration of
enabling a dynamic empirical analysis of the ownership structure for the firms.
compensation behavior.7 Firms categorized as “Firm Controlled”
As mentioned above, the data used also are those in which another domestic firm
include a subset of the Bonn Data that is either owns more than 50% of the out-
based on a collection of financial reports of standing shares or another firm owns at
about 700 German industrial corporations least 25% and no one else owns more than
quoted on the German stock exchange.8 25%. “Foreign Controlled” firms are those
Because of mergers, bankruptcies, acquisi- with more than 50% ownership by foreign-
tions, changes in legal status, and double ers.“Bank Controlled” firms are those with
listing of consolidated and non-consolidated more than 50% ownership by financial
information, about 295 unconsolidated institutions. “Family Controlled” firms are
firms remained in 28 industry groups as of those with more than 50% ownership by
1986.9 We had information on the owner- family members, groups, or individuals.
ship identity and concentration for exactly “Government Controlled” firms are those
100 firms over the full time period. This with more than 50% ownership by any
was reduced to 91 firms in the analysis county or state. “Management Controlled”
where we had both balance sheet and bank firms are those which have no block owner-
relationship data.Thus we used all firms for ship percentage over 25%. Finally “Mixed
which the appropriate data were available Control” denotes a situation where differ-
rather than using a random sample of ent types of the above owners exist, each
firms.10 owning anywhere from 25% to 50% of the
We need to focus on these listed firms in shares.We combine the last three groups as
order to study the phenomenon of executive the omitted category in the empirical
compensation since these are the types of analysis. Most firms in this omitted group
firms for which supervisory and managing have no block ownership over 25%.
boards are mandatory. We used data from The concentration of firm ownership is
1970 to 1986 because prior to 1970 there measured from one to five, where five
were too many missing key variables.11 defines the highest degree of concentration,
Even for this time period, some variables with a single stockholder holding more
were missing, thus requiring running than 75% of the firm’s shares. If two or
regressions on smaller subsets of the full three stockholders hold more than 75%
sample. Data after 1986 are not used of the shares and the firm does not qualify
458 DIRECTORS’ REMUNERATION
for the higher concentration level, then The number of observations at the bottom
concentration is set at four. Concentration of the table indicates the maximum num-
is set at three if a single stockholder holds ber of observations for each variable;
more than 50% of the shares, and two, if however, the actual regressions generally
two or three stockholders own more than used fewer observations because of missing
50% of the shares. Concentration is set at data. Note that there appear to be substan-
one for all cases in which the concentration tial differences across variable means for
level is lower than for category two.14 Note the levels of ownership concentration. This
that if the firm satisfies two or more con- justifies the separate analysis of each
centration levels, the firm is placed in the category of ownership concentration. It
higher concentration category.15 SB and should be noted that average compensation
MB pay is defined as total salaries paid to per board member is lower for the higher
supervisory and managing board members concentration ownership categories 3–5.
by the firm for which they provided this Thus without controlling for other factors,
service.16 Sales is measured as sales of the there is some evidence that higher levels of
firm net of taxes.The data for variables are ownership concentration help to control the
measured in terms of logs of millions of agency problem.
Deutsche marks. The estimations were run Table 17.2 contains the results of OLS
with both time and industry dummy fixed-effect regressions on the log of com-
variables in order to control for macroeco- pensation measures of the two boards. We
nomic shocks and industry-specific effects are most interested in the results dealing
in the model. with the MB. Panel A of Table 17.2 reports
results for regressions using the log of
average salary of a management board
6. EMPIRICAL RESULTS member and the log of total salaries of the
management board members of each firm
The means and standard deviations of the as the dependent variables. The results for
main variables are presented in Table 17.1. both measures of compensation are quite
Variable All firms Bank Independent Conc 5 Conc 4 Conc 3 Conc 2 Conc 1
influenced (high) (low)
Standard deviation in parenthesis below means. Net sales, total assets, and average board salaries are in logs of millions of DM.
Conc stands for ownership concentration: Conc 5 highest concentration, Conc 1 lowest concentration. Bank influence is a
composite measure of bank influence on the firm, incorporating information on: Ownership structure and concentration, proxy
voting by banks, bank representation, and position on SB. Data sample covers 17 years for 100 firms; however, not all firms
have data for each year for every variable. Therefore, actual observations used in calculations vary.
EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 459
Table 17.2 OLS fixed-effects regressions of (Panel A) MB salaries and (Panel B) firm-level SB
salaries
Ownership
Variable ROE Sales concentration Bank influence Adj R2
Panel A
Average MB 0.049* 0.210* 0.036* 0.108* 0.4184
member salary (2.326) (20.403) (3.419) (2.338)
Firm total MB 0.027 0.502* 0.0841* 0.191* 0.8072
salary (1.100) (42.408) (7.020) (3.594)
Panel B
Average SB 0.094* 0.1577* 0.114* 0.002 0.3425
member salary (2.648) (9.096) (6.471) (0.019)
Firm total SB 0.111* 0.410* 0.098* 0.1310** 0.5304
salary (2.937) (22.079) (5.185) (1.570)
All equations use time and firm dummies, and White’s error correction for obtaining heteroskedastic parameter
estimates.
*t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365.
similar indicating that we are capturing the indicating that better performance and
same compensation process with either larger size lead to greater compensation
definition and can proceed to report results for the members of the SB board. However,
only for the average salary per board mem- the bank influence variable is negative and
ber in the following tables. The ownership not significant in the average SB member
concentration variable has a negative salary equation and positive and significant
coefficient and is statistically significant in at the 10% level in the firm total SB salary
both equations as predicted by the agency equation. Since bankers sit on the SB they
hypothesis. Firms with more concentrated may have conflicting motives regarding
ownership provide less compensation for controlling compensation. Clearly bank
MB members. ROE has a positive and influence is more important in controlling
significant coefficient for the average MB compensation on the management board.
member salary but is insignificant for the In results not reported here, we have
firm total MB salary.The Sales variable has substituted ROA for ROE and Assets or
a positive and statistically significant coeffi- Employees for Sales. The results including
cient in both equations indicating that larger these variables were essentially the same as
size leads to greater compensation for the those reported here.
members of the MB.17 The bank influence In Tables 17.3 and 17.4 we add addi-
dummy variable is negative in both cases at tional explanatory variables to the analysis.
the 5% level. Bank influence appears to Instead of using a single summary variable
reduce management compensation. to measure ownership concentration, we
Panel B of Table 17.2 presents results employ four dummy variables in order to
for the SB. In both the average salary examine the importance of the different
equation and the total salary equation, the ownership concentration categories and
ownership concentration variable again has omit the least concentrated category,
a negative coefficient that is statistically Conc 1. We also include dummy variables
significant. Greater ownership concentration for groups that have ownership control.18
leads to lower compensation for SB mem- Table 17.3 analyzes management board
bers, confirming the hypothesis that lack of salaries and Table 17.4 analyzes SB
ownership control permits greater personal salaries. In both tables three equations are
rewards. Both ROE and Sales are positive estimated using OLS fixed-firm effects
and significant in the two regressions, and generalized method of moments using
460 DIRECTORS’ REMUNERATION
Table 17.3 Managing Board (MB) salaries
All equations use time and firm dummies, and average MB salary.
OLS fixed-effects model reports estimates corrected for heteroskedasticity.
GMM heteroskedastic-consistent estimates use instruments of lagged ROE, ROA, and Sales variables from
t 2, . . . , t 7.
The Sargan statistic tests for over identifying restrictions in the model.
Regressions 2 and 3 do not have the Bank ownership variable because of the high correlation with the composite
bank-influence variable.
*t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365.
instrumental variables (GMM IV). All The results in Tables 17.3 and 17.4
reported coefficients are heteroscedastic- provide strong support for the agency
consistent parameter estimates. Time dum- hypothesis, though the traditional variables,
mies were also included in the regressions. ROE and Sales, do not perform as well as
The first equation in each set of equations before. For the regressions with the MB
includes the concentration dummies and average salary, ROE has a positive but
four ownership dummy variables but non-significant coefficient while Sales is
excludes the bank influence variable positive and significant in half of the equa-
because it is highly correlated with bank tions but negative and significant in the
ownership. The second equation excludes other half of the equations. The bank influ-
the concentration dummies and the third ence variable however, has a negative and
excludes the ownership dummies. significant coefficient in all four equations
EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 461
All equations use time and firm dummies, and average SB salary.
OLS fixed-effects model reports estimates corrected for heteroskedasticity.
GMM heteroskedastic-consistent estimates use instruments of lagged ROE, ROA, and sales variables from
t 2, . . . , t 7.
The Sargan statistic tests for over identifying restrictions in the model.
Regressions 2 and 3 do not have the Bank ownership variable because of the high correlation with the composite
bank-influence variable.
*t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1365.
Electronic and
Dependent Metals and Chemicals precision Motor Food and
Variables minerals and fiber instruments vehicles tobacco Textiles Other
All equations use time and firm dummies, and average MB salary.
OLS fixed-effects model reports estimates corrected for heteroskedasticity.
* t-Value indicates variable significant at 5% level and **significant at the 10% level. Observations 1683.
Table 17.5 reports regressions for MB the United States. There is evidence that
salaries by industry using three independ- banks serve as monitors of executive com-
ent variables. While ROE is insignificant in pensation for the MB and some evidence
all five equations, Sales is positive and with respect to the SB. Block ownership by
statistically significant in three of the five various groups also restrains compensation.
industries. Bank influence is negative and The results of this study are consistent
significant in five of the seven industries. with findings of Schmid (1997) and indi-
Thus, even within most industries banks cate that in a different institutional setting
appear to temper compensation when they the same types of economic problems relat-
have significant influence. ing to executive compensation hold—that
is, there is evidence of agency problems
caused by the separation of ownership and
7. CONCLUSION control. When ownership is dispersed
management can obtain greater monetary
We have analyzed data on the compensa- compensation. Unfortunately, we do not
tion of members of the supervisory and have information on non-monetary compen-
managing boards of large German corpora- sation, but we might also expect that lack
tions in order to satisfy the heightened of control by ownership may enable man-
interest in the compensation of German agement to extract greater non-monetary
executives. Most of the empirical work on compensation.
executive compensation has dealt with the This study shows that the agency impact
United States. This study extends one on executive compensation in Germany is
important aspect of the research done on similar to that in the United States. In
American executives to Germany. We find, addition we find that bank influence
as do studies on the US, that the greater the reduces executive compensation and that
ownership concentration the less the ability ownership structure can affect executive
of executives to extract higher levels of compensation practices.
compensation. The agency problem caused
by the separation of ownership and control
appears to exist in Germany as well as ACKNOWLEDGEMENTS
elsewhere. We also find that performance
and size are generally positively related to We would like to thank Joachim
compensation, similar to previous studies in Schwalbach, Frank Seger, and three
EXECUTIVE COMPENSATION AND AGENCY COSTS IN GERMANY 463
ABSTRACT
We examine a sample of firms that adopt “target ownership plans”, under which
managers are required to own a minimum amount of stock. We find that prior to plan
adoption, such firms exhibit low managerial equity ownership and low stock price
performance. Managerial equity ownership increases significantly in the two years following
plan adoption. We also observe that excess accounting returns and stock returns are
higher after the plan is adopted.Thus, for our sample of firms, the required increases in
the level of managerial equity ownership result in improvements in firm performance.
option holdings do not satisfy the ownership options is delayed.Thus, by explicitly linking
requirements). Seven percent of the plans in future equity compensation to a specific
our sample express the target as a number minimum ownership goal, these target
of shares. The remaining plans express the ownership plans are designed to motivate
target as a percentage of salary, so that executives to increase their equity ownership
the target becomes more difficult to attain and to maintain this increase.
if the stock price decreases. Finally, the We hypothesize that target ownership
typical disclosure specifies the maximum plans are adopted when the board of
time allowed to achieve the ownership goal. directors recognizes that the firm has a
Adopting firms indicate that their governance problem. The board adopts
motivation for imposing minimum equity the plan in order to move the firm to a
ownership levels is to ensure that their more appropriate governance structure. We
managers have the appropriate incentives assume that stock returns that are lower
to increase shareholder value. For example, than industry benchmark returns are
the 1993 proxy statement of Morrison evidence of potential governance problems.
Restaurants Inc. says: The United Shareholders Association
(1992) uses this approach to detect
Believing that equity ownership plays a governance problems, and the approach is
key role in aligning the interests of verified empirically by Core et al. (1999).
Company personnel with Company Another way to detect actual or potential
stockholders, the Company encourages governance problems is to determine
all employees to make a personal invest- whether a firm’s managerial stock ownership
ment in Company stock. The Company’s is low relative to some comparison group.
goal is that 10 percent of the Common This approach is similar to the method used
Stock will be owned by employees by the by governance monitoring groups such as
year 2000 and that 80 percent of Institutional Shareholder Services (1993).
employees with more than two years of Thus, we predict that if a firm has low
experience with the Company will own managerial stock ownership and poor stock
Common Stock. (Proxy statement dated price performance, the board infers that the
August 26, 1993, p. 12) level of managerial equity is insufficient to
motivate good performance. One strategy
Thus, target ownership plans are for mitigating this perceived governance
designed to address the contention of some problem is to adopt a target ownership
researchers and governance activists that plan that requires managers to own a min-
stock ownership of senior-level executives is imum level of stock.Thus, our first research
“too small” (e.g., Jensen and Murphy, hypothesis is as follows:
1990; Jensen, 1993; Norton, 1995).
In addition to the implicit requirement H1. The likelihood of adopting a target
created by the public announcement of the ownership plan is negatively related to
ownership targets, 52 (27%) of our prior stock price performance and
sample firms state an explicit penalty for managerial equity ownership levels.
executives who do not meet the ownership
target. The penalty falls into one of three We hypothesize that the board acts in
categories, each of which involves changing the interests of shareholders to mitigate a
some aspect of the executive’s equity perceived governance problem, and for that
compensation. When executives do not reason it adopts a “substantive” target
meet their targets, either (1) a fraction of ownership plan that mandates ownership
their annual cash pay is paid as restricted increases by executives.
stock, (2) their grants of options, restricted An alternative to this hypothesis is that
stock and cash long-term incentives are management controls the board and
reduced or eliminated, or (3) the vesting of convinces the board to adopt a plan that
their outstanding restricted stock and does not actually force the managers to
470 DIRECTORS’ REMUNERATION
increase their equity ownership. The most firms, and performance will improve
purpose of this “symbolic” plan is to make if managerial equity ownership increases.
it appear that managers have taken steps For example, McConnell and Servaes
to improve governance, but without actu- (1990) find evidence of a positive relation
ally having to bear the costs of increased between increases in ownership and firm
ownership. This hypothesis assumes that performance so long as managerial owner-
managers believe that outside investors are ship is less than 50%. Demsetz and Lehn
unable to discriminate between symbolic (1985) predict, and Himmelberg et al.
changes and substantive changes (e.g., (1999) find, that there is no relation
Pfeffer, 1981). between equilibrium levels of ownership
Another explanation for an association and firm performance. In other words, once
between target ownership plan adoption the researcher recognizes the endogenous
and equity ownership increases is that nature of equity ownership, there is little
management has private information about evidence to indicate a relation between
future firm performance and encourages equilibrium levels of ownership and firm
the board to adopt a target ownership plan. performance (Himmelberg et al.).
However, if this were true, we would not In contrast to this prior research, we
expect changes in managers’ relative owner- examine firms that are below equilibrium
ship positions after plan adoption, because levels of managerial equity ownership. We
rational managers would take advantage of predict that as equity ownership rises, there
their superior information and increase will be an increase in operating perform-
their stockholdings prior to the announce- ance. In addition, if the stock market does
ment of the plan. Because managers expect not completely anticipate this contractual
to achieve the ownership targets, it will not change, the firm’s stock price will increase
matter to them whether the plan includes a when the firm announces to the market
penalty for noncompliance. that it has adopted a target ownership
To distinguish between these competing plan. Thus, our third research hypothesis is
explanations, we examine changes in man- as follows:
agerial equity ownership following the
adoption of the plan, and our second H3. Adopting a target ownership plan
research hypothesis is as follows: will have a positive impact on subse-
quent operating and stock market per-
H2. Managerial equity ownership formance.
increases following the adoption of a
target ownership plan.
ownership plan adopters have December 1992 FYEs, 35.5% have 1993 FYEs,
31, 1991 fiscal year-ends. We include plan 36.9% have 1994 FYEs, and 9.2% have
adopters with fiscal years ending up to 1995 FYEs. Table 18.1 describes the final
and including December 31, 1995. sample, which comprises 195 firms across
Approximately 1.5% of the sample firms 40 different industrial and service sectors
have 1991 fiscal year-ends, 16.9% have of the economy.There is some concentration
The sample consists of 195 firms that adopt target ownership plans. This table lists the number of
sample firms in each two-digit standard industrial classification (SIC) code, and a description of
that industry
1 2 Agriculture–Crops
10 1 Mining
13 4 Petroleum and natural gas
16 1 Heavy construction
20 8 Foods
22 2 Textile
23 1 Apparel
24 1 Lumber
25 2 Furniture
26 2 Paper
27 5 Publishing and printing
28 20 Chemical
29 4 Petroleum refining
30 3 Tires and rubber
32 2 Stone, metal, and glass
33 3 Steel and nonferrous metals
34 1 Metal fabricating
35 15 Machinery
36 9 Electrical equipment
37 9 Automobile and aerospace equipment
38 8 Instruments and measuring equipment
39 2 Miscellaneous manufacturing
40 3 Railroad
41 1 Bus transit
44 2 Water transportation
48 9 Telephone and broadcast media
49 12 Electric and gas services
50 3 Durable wholesale
51 5 Nondurable wholesale
52 1 Building materials
53 3 Retail merchandise
57 1 Retail home furniture and equipment
58 3 Eating and drinking places
60 16 Commercial banking
61 6 Credit institutions
63 14 Insurance
64 1 Insurance agents
73 8 Business services
75 1 Automobile rental
80 1 Health services
472 DIRECTORS’ REMUNERATION
Timeline
Plan
adoption
Proxy date
announcing plan adoption
The sample consists of 195 firms adopting target ownership plans. Panel A contains descriptive
statistics for the ownership multiples (stock value divided by salary) required for the CEOs and
the other top executives, and the period allowed for the executives to reach this target. Number is
the number of the 195 adopters whose proxy statements provide data. Panel B gives descriptive
statistics for the actual multiple for the CEO and the other top executives. For the CEO, the multiple
is (value of stock owned/salary). The other executives are the most highly compensated executives
as a group, excluding the CEO.The multiple for the other top executives is a weighted average equal
to the total value of stock owned by the other executives divided by their total salary. We compute
the multiples by using the salary and shares owned as disclosed in the proxy statement for the
adoption year 0, and the stock price at the end of year 0. Panel C gives descriptive statistics for
the number of firms for which the actual multiple for the CEO and the other top executives is less
than the minimum specified by the firm. When no minimum is specified, we impute the minimum
using the median values shown in Panel A
Number Percent
below below
Number minimum minimum
median of 2.4 times their salary in stock. to the actual plan minimum. To do these
Because the distributions are skewed by comparisons, we use the plan targets if
some very large observations, the mean disclosed (Panel A, Table 18.2), and if not
values are much higher than the medians, disclosed we impute a target equal to the
at 32.2 and 4.7 times salary, respectively. medians of 4.0 and 2.5 for the CEO and
The median values suggest that it is a other executives, respectively. In Panel C,
minority of CEOs who have not already we show that 38% of CEOs are below
attained the target, and a minority of other minimum and that 49% of other executives
executive teams that have not attained the (on a weighted average basis) are below
minimum. To provide more direct evidence, minimum. Because the weighted average
we compare the actual ownership multiple calculation can be distorted by one executive
474 DIRECTORS’ REMUNERATION
with very large ownership, we also examine enables us to interpret the residual as the
whether each other executive meets the percentage by which actual ownership devi-
minimum. The results of this analysis indi- ates from expected ownership. We compute
cate that for 81% of the firms, at least one this variable for the CEO and for the other
other top executive does not meet his or her top executives.
target. Accordingly, while only 38% of the Following Demsetz and Lehn (1985)
firms have CEOs that do not exceed their and Baker and Hall (1998), we expect that
minimums, 84% of the firms have a CEO managerial ownership will increase at a
or at least one other top executive who decreasing rate as firm size increases. We
does not meet his or her minimum. follow prior researchers in measuring size
with the natural logarithm of the market
value of equity at the end of year 0, denoted
4. ANALYSIS OF TARGET OWNER- as log(MV equity). Prior researchers such
as Murphy (1999) find that executive
SHIP PLAN ADOPTION
salaries also increase at a decreasing rate
as firm size increases. Because log(stock
In this section, we describe how we value/salary) is equal to log(stock value)–
construct a test of Hypothesis 1, which log(salary), we do not predict the direction
predicts that target ownership adopters of the association between this variable and
have lower pre-plan returns and ownership log(MV equity).
than a control sample. We define our We expect that equity ownership will
measures, describe how we construct the increase at a decreasing rate as monitoring
control sample, and then describe our test.
costs increase. We also follow Demsetz and
Lehn in using stock return volatility as a
4.1. Measures for benchmark- proxy for noise that increases monitoring
adjusted returns and ownership costs. We capture the hypothesized concave
relation between increases in noise and
We measure firm performance as the increases in equity ownership by including
stock price return in the two years (years -2 the square of stock return volatility.
and -1 in Figure 18.1) preceding the adop- We measure stock volatility by using the
tion year 0, less the median stock price standard deviation of daily stock price
performance during the same time period returns over the six months before the
for the firms on the 1998 Compustat file fiscal year-end. Because several of our
that have the same two-digit SIC code. We firms did not trade for all of year 0, we
label the resulting variable “prior industry- maximize our sample size by measuring
adjusted returns”, and we assume that daily stock volatility over the six months
governance problems are an inverse function prior to the fiscal year-end. We predict that
of prior industry-adjusted returns. stock volatility will be positively associated
We compute our ownership benchmark with log(stock value/salary), and that stock
by constructing a regression model compa- volatility squared will be negatively
rable to that used by Demsetz and Lehn associated with log(stock value/salary).
(1985). Because target ownership plans Similar to Smith and Watts (1992), we
require executives to own a certain value of expect that CEO equity ownership will be
common stock directly (exclusive of stock greater for firms with larger investment
options) that is a multiple of that executive’s opportunity sets. Like Smith and Watts, we
base salary, we examine the ownership use the book value of assets divided by the
multiple computed at year 0 as described market value of assets as a proxy for
above.2 To normalize the distribution of growth opportunities, and expect that firms
this highly skewed variable, we transform it with greater growth opportunities will have
by the natural logarithm and use log(stock lower book-to-market ratios.We measure the
value/salary) as the dependent variable in book-to-market ratio at the end of year 0.
our ownership regression.This transformation We expect to find a negative relation
MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 475
between the book-to-market ratio and of 19.4% and 17.9%, respectively. All of
log(stock value/salary). the reported statistical tests are two-tailed.
To control for industry factors and other The log of the ownership multiple has a
unspecified determinants that might affect statistically positive relation to firm size.
the level of equity ownership, we also As we expected, we find a negative relation
include 23 industry indicator variables. To to the book-to-market ratio (or a positive
capture potential temporal differences, we relation to the investment opportunities
include five indicator variables that corre- confronting the firm). We also observe that
spond to the years of data collected from the log of the ownership multiple has a
our sample of plan adopters and concave relation to the standard deviation
ExecuComp. The resulting benchmark of stock returns. We find less managerial
model for the level of stock ownership is as ownership for those firms with very low
follows: or very high stock volatility. We label
the residual from the benchmark model the
Log(stock value/salary)it 0 “stock value residual”, and we assume that
1Log(MV equity)it governance problems are an inverse function
2 Stock volatilityit of the stock value residual.
3 Stock volatility
squaredit
4.2. Results
4 Book-to-marketit
1 . . . 5 year Because Hypothesis 1 predicts differences
indicatorsit between adopting and non-adopting firms,
1 . . . 23 industry we require a set of control firms that have
indicatorsit it . (1) not adopted a target ownership plan. We
create a control sample by deleting from
We estimate this benchmark model using the sample described above all data for
ordinary least squares. We create an 1996 and 1997 and all firms that are
estimation sample by pooling annual own- included in our sample of target ownership
ership data for the target ownership plan plan adopters. (Because we have the full
adopters for years 0, 1, and 2 with sample of target ownership adopters prior
ExecuComp data on managerial ownership to 1996, we know that none of our control
for firm-years from 1992 to 1997. Because firms have adopted plans.) A total of 4,498
ExecuComp has no 1991 fiscal year data, firm-years on ExecuComp from 1992 to
we include the 1991 adopters coded with a 1995 constitute our control sample.
year indicator of 1992. The ExecuComp Because some firms are missing data on
database meets our requirement for an prior industry-adjusted returns, the target
accurate, convenient data source for mana- ownership sample reduces to 170 observa-
gerial equity ownership. We include data tions and the control sample to 4,022
for the years 1996 and 1997 so that we observations. Again, to mitigate the influ-
can use these residuals in later tests ence of outliers, we set the upper- and
(described in Section 5.1) of whether the lower-most percentiles for each independent
managerial equity ownership of the 1994 variable equal to the values at the first and
and 1995 plan adopters increases in the 99th percentiles in each year, respectively.
two years following adoption. To mitigate Univariate results are consistent with
the influence of outliers, we set the upper- Hypothesis 1. We find that prior industry-
and lower-most percentiles for each variable adjusted returns for the target ownership
equal to the values at the first and 99th plan sample are significantly lower than
percentiles in each year, respectively. those for the control sample by a mean of
Table 18.3 presents the estimation results, 4.5 percentage points (p0.002) and
which indicate that the models for the CEO median of 0.7 percentage points
and other top executives are statistically (p0.05).3 In addition, the CEO stock
significant (p0.001), with adjusted R2s value residual for the target ownership plan
476 DIRECTORS’ REMUNERATION
Table 18.3 OLS regression models of log(stock value/salary)t
This table summarizes regression results from estimating Eq. (1). The sample consists of 7,373
firm-year observations for years 1992–1997 for the target ownership plan adopters and for firms
included on ExecuComp. We base t-statistics (in parentheses) on OLS standard errors. Log(CEO
stock value/salary) is the natural logarithm of the value of stock owned by the CEO divided by the
CEO’s salary. Log(other execs’ stock value/salary) is the natural logarithm of the total value of
stock owned by the other executives divided by the total salary of the other executives. The other
executives are the most highly compensated executives, excluding the CEO. We compute these
variables by using the salary and shares owned as disclosed in the year t proxy statement, and the
stock price at the end of year t. We compute all the explanatory variables at or for the period ending
at year t. Log(MV equity) is the natural logarithm of the market value of equity. Stock volatility is
the standard deviation of daily stock price returns over six months. Stock volatility squared is the
square of stock volatility, and book-to-market is the book value of assets divided by the market value
of assets. Coefficients on an intercept, five year indicators, and 23 industry indicators are not shown.
Dependent variable
This table summarizes estimation results of Eq. (2), which is a logistic model in which the
dependent variable is equal to one if a firm adopts a target ownership plan and zero otherwise. The
sample consists of 4,192 firm-year observations from 1992 to 1996. Pred. Prob. is the change in
the predicted probability that occurs when the independent variable increases from its first to third
quartile value, and is evaluated at the mean values of the remaining independent variables. We base
t-statistics (in parentheses) on maximum likelihood standard errors. The prior industry-adjusted
return is the stock price return in the two years preceding the fiscal year in which the plan is
adopted, less the median stock price performance during the same time period for all firms con-
tained in the 1998 Compustat file that have the same two-digit SIC code. The stock value residual
is the residual obtained from estimating the model described by Eq. (1). We measure the stock value
residual at the end of the year in which the plan is adopted. Coefficients on the intercept and three
year indicator variables are not shown.
stock value residual becomes more whether this difference indicates a significant
significant in Column 3 (p0.01). increase. For the sample of firms for which
Although not reported, we also obtain we can obtain proxy statement data on
qualitatively similar results if we select the CEO ownership two years after plan adop-
control sample by using a random match tion (n 174), we observe that the CEO
that approximates the proportion of target stock value residual increases by a mean
ownership adoptions per year. For example, (median) of 14.1% (27.5%). The mean is
because 26.3% of the adoptions occur in marginally significant (p 0.13), and the
the Compustat data year 1993, we randomly median is significant at a 0.001 level. For
select 26.3% of all firms on the 1993 those firms with data on top executive
ExecuComp file as a control group for the ownership changes for the two years after
1993 adoption group, and so forth. Finally, the plan adoption (n 173), we find that
our results are robust to our inclusion of the other executives’ stock value residual
firm size and industry indicators as increases by a mean (median) of 18.2%
additional control variables. (11.7%), which is statistically significant
Thus, consistent with Hypothesis 1, our at a 0.07 (0.01) level.
results suggest that boards respond to There are three possible limitations to
governance problems (as measured by low these results. First, the increases that we
relative stock price returns and low relative document above might reflect mean rever-
managerial equity ownership) by adopting sion in executive ownership, in which case
target ownership plans for senior-level one would expect to observe increases for a
executives. sample of firms known to have a low level
of ownership relative to the population. In
fact, when we regress the change in owner-
ship for the full sample on the beginning
5. CONSEQUENCES OF TARGET residual, we find a negative and significant
OWNERSHIP PLAN ADOPTION coefficient on the beginning residual, which
means that firms with lower ownership
The majority of this section discusses how experience greater subsequent increases.
we conduct tests of our hypothesis that We address this limitation by matching
performance improves for target ownership each sample firm to the control firm with
plan adopters. However, before we examine the closest ownership residual in the
the performance consequences of plan adoption year. We choose from among
adoption, we must first determine if target those control firms for which we can obtain
ownership plans actually cause executives proxy statement data on ownership two
to increase their level of stock holdings, as years after the plan adoption. We then
we predict in our second hypotheses. If we compute the two-year change in the stock
do not find that managerial equity value residual for each sample firm and
ownership increases after the adoption of compare it to the two-year change in the
target ownership plans, we would not stock value residual for the matched
expect to find improvements in performance control firm.
associated with the plans. Benchmarking sample firm increases
against control firm increases also
addresses a second limitation, which is that
5.1. Changes in managerial stock managers’ equity ownership is likely to
ownership increase with the time they have spent at
the firm. This increase over time can occur
To test whether equity ownership increases either because of mechanical reasons
following adoption, we use the stock value related to the accumulation of stock
residuals described in Section 4.1. We through stock compensation plans (i.e., the
examine the difference between the year 2 exercise of stock options) or because of
residual and the year 0 residual, and test economic reasons, such as the need to
MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 479
prevent horizon problems with CEOs near Moreover, the ownership residuals for
retirement. Palia (1998) and Core and Guay these CEOs are no longer statistically
(1999) find evidence that CEO ownership different from zero, indicating that these
increases with the CEO’s tenure. CEOs’ ownership levels have increased to
We find that the sample firm increases equilibrium levels. We obtain qualitatively
are significantly greater than those of the the same results if we examine CEO stock
control firms. Paired t-tests (z-tests) indi- value residualtenure instead. The sample
cate that the sample firm mean (median) firms’ mean (median) increase of 15.6%
increase of 14.1% (27.5%) in the CEO (3.2%) is significantly greater than both
stock value residual is significantly greater zero and the control firms’ mean (median)
than the control firm mean (median) increase of 10.1% (10.1%), and the
increase of 9.3% (1.7%) at a 0.04 year 2 ownership residuals for these CEOs
(0.02) level. The sample firm mean are not statistically different from zero.
(median) increase of 18.2% (11.7%) in This consistent evidence of ownership
the other executives’ stock value residual is increases for the CEO and for the other
also greater than the control firm mean top executives supports Hypothesis 2 that
(median) increase of 12.0% (3.9%) target ownership plans are followed by
at a 0.02 (0.02) level. significant increases in managers’ equity
As a second means of controlling for ownership.
CEO ownership increases over time, we add
CEO tenure as an additional control vari-
able in the regression model for log(CEO 5.2. Post-adoption operating
stock value/salary) described by Eq. (1). We performance: methodology
denote the residual from this model as CEO and results
stock value residualtenure. Data on tenure for
the other executives are not available from To examine whether plan adoption
most firms’ proxy disclosures. Therefore, we improves firm performance, we first look at
cannot control for tenure in the ownership whether the accounting return on assets
model for the other executives. For the (ROA) is statistically positive over the two
sample firms, we observe that the mean years (years 1 and 2) following the adop-
(median) CEO stock value residualtenure tion of the target ownership plan. For
increase of 12.4% (5.0%) is significantly accounting returns, we use the industry and
greater than zero at a 0.12 (0.09) level, and performance match suggested by Barber
also significantly greater than the control and Lyon (1996) to develop a comparison
firm mean (median) change of 15.8% benchmark. Barber and Lyon find that this
(12.1%) at a 0.004 (0.001) level. approach generates well-specified models
Another limitation on our results is that that can test future abnormal performance
the changes in CEO ownership from year 0 for firms that “ . . . as a group, have histori-
to year 2 might be contaminated if a cally experienced especially good or poor
sample firm changes CEO during year 1 or performance” (p. 378), or whose perform-
year 2. We delete the 42 firms with CEO ance “differs only slightly” from the popu-
turnover, and examine separately the sam- lation (p. 396). Using this benchmark is
ple of firms with the same CEO from the important, because in the two years prior
year of adoption until at least two years to adoption, our sample firms underper-
after the plan adoption (n 132). For form the control firms’ ROA by a mean of
these firms, we observe that the CEO stock 0.9 percentage points (p0.01) and
value residual increases by a mean median of 0.7 percentage points (p0.01).
(median) of 39.9% (30.3%), which is sta- Also, by examining the future performance
tistically different from zero at a p 0.01 of historical poor performers that are less
level. This increase is also greater than the likely to survive for two future years, this
12.2% (12.0%) increase of the control matching procedure mitigates any potential
firms with the same CEO at a p0.01 level.5 sample selection bias.
480 DIRECTORS’ REMUNERATION
To implement this procedure, we select a the ROA next closest to the sample firm
comparison firm for each adopting firm. in the year of adoption.We still require that
We match firms on their two-digit standard the control firm ROA be within 90% and
industrial classification (SIC) codes, and 110% of the sample firm ROA. When we
then select that firm with an ROA closest cannot find a matching firm in the same
to the sample firm in the year prior to industry, we follow Barber and Lyon
adoption. We require that the control firm’s (1996) and select the firm with the closest
ROA be within 90% and 110% of the ROA regardless of its SIC code. As dis-
sample firm’s ROA. We are unable to cussed below, Vijh (1997) uses a similar
match five adopting firms because they splicing method when he computes excess
have no Compustat data for year 0. stock returns.
Sometimes, either the sample firm or a Although Barber and Lyon (1996) base
control firm is missing data in year 1 or their ROA calculation on operating income
year 2, either because it was acquired or before depreciation expense (Compustat
for some other reason. All of the 190 data item 13), this data item is not avail-
sample firms with year 0 data also have able on Compustat for certain financial
data available for year 1, but nine of the institutions. To maximize our sample size,
firms have no data available for year 2, so we also compute an ROA using operating
we do not compute the excess ROA for income after depreciation expense
these firms. If a matching firm has no data (Compustat data item 178).
available for either year 1 or year 2, we use Table 18.5, Panel A, presents the
the firm that is in the same SIC code with accounting performance comparisons. The
Table 18.5 Two-year post-adoption excess operating and stock performance for target ownership firms
The sample consists of 190 target ownership firms. We calculate excess ROA by using the
matched-firm approach of Barber and Lyon (1996), where the matching firm is the firm in the same
industry with the closest prior operating performance, and by using both operating income after
depreciation and operating income before depreciation. We calculate excess stock returns using the
matched-firm approach of Barber and Lyon (1997), where each sample firm is matched to the
non-sample firm with the closest book-to-market ratio within that subset of firms whose market
value lies between 70% and 130% of the sample firm market value.
Excess returns:
First six months 190 3.8% 0.086 2.9% 0.041
of year 1
Year 1 190 5.7% 0.161 5.7% 0.160
Years 1 and 2 190 5.3% 0.442 7.9% 0.171
MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 481
first line of each half of the panel indicates zero (p 0.10). These short-window results
that our matching procedure is successful do not support Hypothesis 3.
in creating a control sample whose prior- Although target ownership plans lead to
year performance is insignificantly different improvements in operating performance, we
from that of our test sample. In the first might not see excess stock returns in the
year after the contractual change and for short time period around the proxy date.
the cumulated two years after adoption, the There are two reasons for this. First,
target ownership sample has an ROA that although our searches of the Dow Jones
is significantly greater than the control and PR newswires reveal no public discus-
firms. The adopting firms statistically sion of the plans prior to the proxy date,
outperform the ROA (after depreciation) of managers might have privately communi-
the benchmark firms by a mean (median) cated the news of the adoption of a target
of 1.2 (0.5) percentage points. In the two ownership plan to large shareholders.Thus,
years after adoption, the adopting firms by the proxy date, the expected benefits of
statistically outperform the compounded the plan might already be impounded into
ROA of the benchmark firms by a mean the stock price.7
(median) of 1.8 (0.8) percentage points. Second, even if the market knows that
The bottom half of Panel A shows that the plan exists, target ownership plans rep-
our results are also robust for a reduced resent an innovation, and the positive or
sample for which we use ROA before negative performance consequences of this
depreciation. Again, the adopting firms innovation become clear only with the pas-
statistically outperform the ROA of the sage of time. In this case, when there are
benchmark firms by a mean (median) of improvements in operating performance,
1.2 (0.6) percentage points. In the two the market will be surprised and there will
years after adoption, the adopting firms be positive abnormal returns for a period
statistically outperform the compounded following the plan adoption. For example,
ROA of the benchmark firms by a mean the stock market might react to changes in
(median) of 1.4 (0.7) percentage points. managerial investment and financing deci-
These results are consistent with sions, rather than to the contractual change
Hypothesis 3 that target ownership plans disclosed in the proxy statement. We
are followed by significant increases in explore these possibilities by assessing
operating performance. whether stock price returns are statisti-
cally positive over the six-, 12-, and 24-
5.3. Post-adoption stock month periods starting in the year after
performance: methodology and target ownership plan adoption (year 1).
Barber and Lyon (1997) and Lyon et al.
results
(1999) show that the use of a control firm
We assess stock price performance in the matched on size and book-to-market in
announcement window around the release of computing buy-and-hold excess returns
the proxy statement and for the six-, 12-, and (BHERs) for each sample produces test
24-month periods beginning in year 1, the statistics that are well specified in random
year after target ownership plan adoption. samples of firms and for almost all non-
We first examine stock market returns in the random samples of firms. We use their
three-day window surrounding the date of methodology to develop our group of con-
the proxy statement in which the target trol firms. We obtain the matching firms
ownership plan is disclosed.6 Using standard from the list of all Center for Research in
event study methodology (e.g., Brickley, Securities Prices firms with price data as
1986) and the statistical tests described in of the end of June of year 0 (where year 0
Patell (1976), we find that over the three- is the year of target ownership plan adop-
day window, the cumulative average excess tion). We compute market value in June of
return for the adopting firms is 0.15%, year 0, and the book-to-market ratio by
which is statistically indistinguishable from using the last book value reported prior to
482 DIRECTORS’ REMUNERATION
June of year 0. Using the subset of firms positive stock value residuals, 70 of the
whose market value lies between 70% and firms have other top executives with
130% of the sample firm value, we pair positive stock value residuals, and 38 of the
each sample firm with the firm that has the firms have both CEOs and other executives
closest book-to-market ratio. (We note that with positive stock value residuals. This
our control group is closely matched to our observation raises two issues. First, we
test sample: on average the percentage wish to establish that our results are not
difference in firm size is 4.0% and the driven by observations that already have
percentage difference in the book-to-market appropriate equity incentives (for which we
ratio is -0.0%.) We compute the BHERs expect the incentive effects of establishing
starting the first month of year 1, the fiscal minimum ownership levels would be lower).
year in which the firm releases the proxy A second issue is that if an executive has a
statement announcement of the plan. positive stock value residual, the plan could
Because the proxy is always released within create perverse incentives by forcing him or
the first six months of the fiscal year-end, her to own “too much” stock.
all of our BHER measures can capture any We address the first issue in two ways.
announcement effects related to the plan First, we delete the 38 firms that have
adoption. positive stock value residuals for both the
If either a sample or a control firm stops CEO and the other executives as a group.
trading because it was acquired (or for We then examine the performance of the
some other reason) before the end of our remaining 157 firms and find that these
accumulation period, we follow Vijh (1997) firms experience large, significant increases
by ending the excess return calculation with in ownership for both the CEO and the
the delisting month. If a control firm other executives as a group in the two years
delists, we use the return for a firm that has after plan adoption. For this subsample,
the book-to-market ratio next closest to the all means and medians of the excess
sample firm in the year prior to adoption performance variables remain positive and
and a firm size within 70% and 130% of significant, with two exceptions. First, while
the sample firm size. excess ROA using operating income after
Table 18.5, Panel B, presents the stock depreciation remains significant, excess
price performance results.The sample firms ROA using operating income before
statistically outperform the control group depreciation loses significance (this vari-
for the first six months of the fiscal year of able is available for fewer observations).
adoption with a mean (median) BHER of Second, the mean excess stock return for
3.8% (2.9%). Although the BHERs for the first six months becomes insignificant
the 12- and 24-month periods are positive, (p 0.152), although the median remains
they are insignificant at conventional levels. significant (p 0.056). Thus, our results
Combined with our findings of significant are robust to deleting the subset of firms
and positive operating performance, the with positive stock value residuals.
BHER results suggest that the market Second, the plan may not require an
reacts favorably to the adoption of the increase in ownership even if the ownership
target ownership plan, and prices its residuals are negative (i.e., the plan is not
expected benefits in the six-month period binding). If the plan is not binding on the
around plan adoption. These stock price top executives, one might expect the
results support Hypothesis 3. incentive effects of establishing minimum
ownership levels to be lower.To address this
5.4. Sensitivity analysis issue, we delete from the sample of 157
firms with negative residuals the 15 (42)
Our sample comprises firms with relatively firms for which the plan does not bind on at
low levels of management stock ownership least the CEO or one of the other top exec-
prior to the adoption of the new program. utives (the CEO or the other top executives
However, 70 of the firms have CEOs with as a group), resulting in a subsample of
MANDATORY INCREASES IN EXECUTIVE STOCK OWNERSHIP 483
142 (115) firms. The inference for the ownership plans does not impose excess
subsample of 142 firms is the same as that ownership on the executives. Moreover, we
reported above for the 157-firm subsample: find no evidence that firms with CEOs or
all of the excess performance variables other executives with positive stock value
remain positive and the same variables are residuals have excess operating or stock
significant. For the smaller 115-firm price performance that is significantly
subsample for which the plan is binding on lower than the remainder of the sample.
the CEO or the top executives as a group, Finally, we note the results of two addi-
the inference is qualitatively similar to that tional sensitivity tests. The two-year CEO
reported above for the 157-firm subsam- turnover rate of 24.1% for the target
ple: all of the excess performance variables ownership adopters is greater than the
remain positive, and the same variables are turnover rate of the control sample of
significant except that the mean excess 19%, and the difference is marginally
two-year ROA using operating income after significant (p 0.12 from a binomial test
depreciation becomes insignificant of differences). Our accounting and stock
(p 0.141), although the median remains price performance results are qualitatively
significant (p 0.014). Thus, with the the same as in Table 18.5 if we examine
exception that excess ROA using operating only those firms with the same CEO over
income before depreciation loses signifi- the time period used for developing the
cance in these subsamples, our results are performance tests. Finally, we note that we
robust to examining the subsamples of have a concentration of financial firms in
firms obtained by deleting firms with our sample (as shown in Table 18.1), and
nonpositive stock value residuals and then these firms have different accounting
by deleting firms without binding plans. conventions and operate in a very specific
The finding that a subset of firms has industrial sector. We obtain qualitatively
positive ownership residuals raises another the same performance results if we delete
question. If a firm’s executives already own the 36 financial firms and examine
a large amount of stock prior to the target separately the non-financial adopting firms;
ownership plan adoption, the plan could the only difference is that the six-month
force the executives to own “too much” excess return is only marginally significant
stock, which could create perverse incentives (p 0.15).
and lead to lower firm performance.
Although this is a possibility, target owner-
ship plans impose a floor on executive own- 6. SUMMARY AND CONCLUSION
ership levels, and do not require increases
in ownership levels for executives whose We construct a powerful test of the
holdings are already above this floor. Of the hypothesis that re-contracting to require
70 sample firms with positive CEO stock managers to increase equity ownership from
value residuals, none of the firms’ owner- suboptimal levels will improve incentives
ship targets require increases in ownership and increase performance. We implement
by the CEO. Similarly, of the 70 sample this test by examining a sample of firms
firms with positive other executive stock that adopt minimum ownership levels for
value residuals, none of the firms’ ownership executive officers. Because most managers
targets require increases by all of the other in our sample have low ownership, these
executives. However, eight of the targets plans generally require increases in mana-
require increases in ownership by some of gerial ownership. We find that firms that
the other executives. Finally, the actual adopt target ownership plans show lower
increases in ownership for these firms stock price performance than do their
either are significantly less than zero or are industry peers in the time period prior to
not different from zero, depending on the plan adoption. These firms also have
measure used for the analysis. Thus, it managers who own lower levels of equity
appears that the design of these target relative to our benchmark model, which is
484 DIRECTORS’ REMUNERATION
5 The other executives’ stock value residual Core, J., Holthausen, R., Larcker, D., 1999.
for the 130 of these observations with available Corporate governance, CEO compensation,
data increased by a mean (median) of 16.5% and firm performance. Journal of Financial
(14.2%), which is significantly greater than Economics 51, 371–406.
both zero and the -19.9% (5.6%) change for DeAngelo, H., DeAngelo, L., Skinner, D., 2000.
the control sample.
Special dividends and the evolution of divi-
6 There is some debate on the desirability of
using proxy statement release dates for dend signaling. Journal of Financial
detecting the shareholder value consequences Economics 57, 309–354.
associated with changes in compensation Demsetz, H., Lehn, K., 1985. The structure of
contracts; Gaver et al. (1992) discuss this point. corporate ownership: causes and conse-
7 As noted above, most of the target owner- quences. Journal of Political Economy 93,
ship plans are adopted in year 0, which raises 1155–1177.
the possibility that the leakage could have Gaver, J., Gaver, K., Battistel, G., 1992.The stock
occurred prior to the start of our event window market reaction to performance plan adop-
at the beginning of year 1. To address this tion. The Accounting Review 67, 172–182.
possibility, we examine excess returns for the six
Hewitt Associates, 1993. Survey Findings:
months ending with the start of year 1 for the
set of firms (with available return data) that do Executive Stock Ownership Guidelines.
not state that they adopted early in year 0. We Hewitt Associates, Lincolnshire, Ill.
find no significant excess returns for either the Himmelberg, C., Hubbard, G., Palia, D., 1999.
sample of 73 firms that specifically disclose Understanding the determinants of managerial
that they adopted in year 0, or for the 152-firm ownership and the link between ownership
subsample that does not state that they adopted and performance. Journal of Financial
in year 1.This finding is consistent with the lack Economics 53, 353–384.
of disclosure of the plans and with the Institutional Shareholder Services, 1993. The
conjecture that the board action to adopt these ISS Proxy Voting Manual. Institutional
plans occurs late in year 0.
Shareholder Services, Washington, DC.
Jensen, M., 1993. The modern industrial
revolution, exit, and the failure of internal con-
REFERENCES trol systems. Journal of Finance 48, 831–880.
Jensen, M., Murphy, K., 1990. Performance pay
Baker, G., Hall, B., 1998. CEO incentives and and top-management incentives. Journal of
firm size. NBER Working paper 6868. Political Economy 98, 225–264.
Barber, B., Lyon, J., 1996. Detecting abnormal Kole, S., 1996. Managerial incentives and firm
operating performance: the empirical power performance: incentives or rewards?
and specification of test statistics. Journal of Advances in Financial Economics 2,
Financial Economics 41, 359–399. 119–149.
Barber, B., Lyon, J., 1997. Detecting long-run Lyon, J., Barber, B., Tsai, C., 1999. Improved
stock returns: the empirical power and methods for tests of long-run abnormal stock
specification of test statistics. Journal of returns. Journal of Finance 54, 341–372.
Financial Economics 43, 341–372. McConnell, J., Servaes, H., 1990. Additional
Brickley, J., 1986. Interpreting common stock evidence on equity ownership and corporate
returns around proxy statement disclosures value. Journal of Financial Economics 27,
and annual shareholder meetings. Journal of 595–612.
Financial and Quantitative Analysis 21, McMillan, J., Sabow, S., 1994. Encouraging
343–349. executive stock ownership. The Corporate
Brill, J., 1993. Stock ownership guidelines for Board (January/February), 16–20.
executives. The Corporate Executive Morck, R., Shleifer, A., Vishny, R., 1988.
(March–April), 381–385. Management ownership and market valua-
Core, J., Guay, W., 1999. The use of equity tion: an empirical analysis. Journal of
grants to manage optimal equity incentive Financial Economics 20, 293–315.
levels. Journal of Accounting and Economics Murphy, K., 1999. Executive compensation. In:
28, 151–184. Ashenfelter, O., Card, D. (Eds.), Handbook of
486 DIRECTORS’ REMUNERATION
Labor Economics, Vol. 3. North-Holland, Smith, C., Watts, R., 1992. The investment
Amsterdam. opportunity set and corporate financing,
Norton, L., 1995. Velvet handcuffs. Barron’s dividends and compensation policies. Journal
(July 24), 21–24. of Financial Economics 32, 263–292.
Palia, D., 1998. The endogeneity of managerial Towers, P., 1993. Stocking up in the 90s: how
compensation in firm valuation: a solution. companies are creating managers/owners.
Working paper, Columbia University. CompScan.
Patell, J., 1976. Corporate forecasts of earnings United Shareholders Association, 1992. 1992
per share and stock price behavior: empirical Shareholder 1,000: rating corporate
tests. Journal of Accounting Research 14, America’s responsiveness to shareholders.
246–276. Vijh, A., 1997. Long-term returns for equity
Pfeffer, J., 1981. Power in Organizations. carveouts. Journal of Financial Economics
Pitman, Boston. 51, 273–308.
Reese, J., 1993. Buy stock—Or die. Fortune Young, L., 1995. The owner mentality. The Wall
(August 23), 14–15. Street Journal (April 12), R12.
Part 5
Governance, performance
and financial strategy
INTRODUCTION
corporate decisions more aligned to shareholder interests, thus testifying to the greater
effectiveness of the governance regime introduced by the Code.
Entrenched managers may not only enjoy private benefits of control at the expense of
shareholders and capture the pay setting process to award themselves excessive compen-
sation but also prevent shareholders from finding out what they are up to. The earnings
they report may have been ‘managed’ and therefore lack transparency. Whether such
creative accounting is feasible depends on the corporate governance structure including
shareholder rights (see Gompers et al., (Ch.20) on shareholder rights as a measure of
corporate governance). Christian Leuz, Dhananjay Nanda and Peter Wysocki (Ch.24)
propose an explanation for the systematic differences they observe in earnings manage-
ment across 31 countries. Insiders, in an attempt to protect their private control benefits,
use earnings management to conceal firm performance from outsiders. Thus, earnings
management is expected to decrease in investor protection because strong protection limits
insiders’ ability to acquire private control benefits, which reduces their incentives to mask
firm performance.The findings are consistent with this prediction and suggest an endogenous
link between corporate governance and the quality of reported earnings.
REFERENCE
Sudarsanam, P. Sudi (2003), Creating value from mergers and acquisitions. The challenges, an
international and integrated perspective, New York: FT Prentice Hall.
Chapter 19
ABSTRACT
Although there has been an intensive debate on the relative merits of different systems of
corporate governance, empirical evidence on the link between corporate governance and firm
performance almost exclusively refers to the market-oriented Anglo-Saxon system. This paper
therefore investigates the more network- or bank-oriented German system. In panel regressions
for 361 German corporations over the time period 1991 to 1996, we find ownership concen-
tration to affect profitability significantly negatively. However, this effect depends intricately on
stock market exposure, the location of control rights, and the time horizon (short-run vs. long-
run). We conclude from our results that (1) the presence of large shareholders does not
necessarily enhance profitability, (2) ownership concentration seems to be sub-optimal for many
German corporations, and, finally, (3) having financial institutions as largest shareholders of
traded corporations improves corporate performance.
Shareholders are stupid and impertinent – stupid, because they give their funds to some-
body else without adequate control, and impertinent, because they clamor for a dividend
as a reward for their stupidity.
Carl Fürstenberg (1850–1933), German financier
attention has been paid in the corporate thwarted any serious attempts of hostile
governance literature to ownership concen- takeovers in Germany.6 In view of high unem-
tration as a key to more effective corporate ployment and sluggish growth, critics have
governance and shareholder value maximiza- pointed to the German network-orientation
tion. The presence of large shareholders as a root cause for entrepreneurial inertia,
may curb managerial discretion, reduce risk aversion, and low investment in emerging
agency costs and enhance performance new technologies and infant industries.7
(Stiglitz, 1985; Shleifer and Vishny, 1986). In the remainder of the paper we discuss
The existing empirical evidence on the impact in Section 2 the link between corporate
of ownership structures on corporate per- governance and firm performance in more
formance refers almost exclusively to detail. Section 3 presents our empirical
Anglo-Saxon firms and is rather mixed.2 analysis that employs a panel data set of
This paper therefore focuses on German 361 German corporations over the time
corporations. We investigate empirically period 1991 to 1996. Our main finding is a
how ownership concentration, the location significantly negative effect of ownership
of control rights, board representation of concentration on profitability. However, this
owners, and stock market exposure affect effect depends intricately on stock market
firm profitability (return on total assets). exposure, the location of control rights, and
Ownership concentration is low in the the time horizon chosen (short-run vs. long-
Anglo-Saxon countries which rely heavily run). Section 4 summarizes and concludes.
on stock markets to channel the flow of
capital, control its efficient use, and assure
outside investors of maximizing the return 2. CORPORATE GOVERNANCE AND
on their investments.3 The Anglo-Saxon FIRM PERFORMANCE
financial system has been criticized for
short-termism, neglect of interests other 2.1. The governed versus the
than shareholders’, and inefficiency in managed corporation
delivering effective corporate governance.4
By contrast, concentrated ownership is The separation of ownership
a salient feature of the German system.
and control
Non-incorporated small and medium-sized
firms tightly held by individuals or families A prime element of corporate governance is
dominate the firm population. German the alignment of shareholders’ interests
corporations are typically in the hands of with the interests of managers hired to run
large blockholders who often command a the firm. The major concern is whether
super-majority interest. Widely dispersed managers pursue their own interests (pet
outside shareholdings as in the USA or UK projects, empire building, perks etc.) rather
are rare. Ownership structures hardly change than maximize shareholder value. Public
over time, implying that large shareholders corporations with widely dispersed outside
stick to their blockholdings even in bad times. shareholdings may be particularly prone to
An active market for corporate control does managerial discretion. Such corporations
not exist yet despite the recent takeover- have a clear separation of control and
battle between Mannesmann and Vodafone. ownership. Shareholders delegate decision-
Although improving lately, the German making to managers and control to a
stock market is still relatively small supervisory board. Managers are put in
regarding listings and market capitaliza- place because it is expected that their
tion.5 Close ties between industrial firms superior entrepreneurial skills allow them
and financial institutions (banks) (e.g., via to better achieve success for the firm than
cross-holdings, long-term lender-borrower the owners. The supervisory board is
relations) apparently foster access to debt responsible for selecting, monitoring, and
capital, thus reducing the need to attract replacing senior managers.
equity capital on the stock market. This An extensive literature has discussed the
network-like structure has effectively pros and cons of separating ownership from
DOES THE GOVERNED CORPORATION PERFORM BETTER? 493
(1994) this outcome is reversed because depend on who they are. In other words, the
of risk-averse large shareholders. Recent location of control rights can be a more
contributions by Pagano and Röell (1998) important determinant of the degree of
and Bolton and von Thadden (1998a,b) control exerted by owners than ownership
stress the existence of a trade-off between concentration. Insider (internal) control
control and liquidity. As the incentive and may represent a higher degree of control
the ability to monitor increase with share at any given level of blockholdings than
concentration, control will be more stringent outsider (external) control (Cubbin and
and effective but this comes at the cost of Leech, 1983; Mayer, 1992). Family inter-
liquidity. Constraining liquidity could affect ests, allied industrial firms, banks, and
profitability negatively if the firm cannot holdings companies are frequently identi-
fully exploit its investment opportunities. fied as insiders, while the shareholders of
This reasoning may only apply to smaller diffusely held firms (e.g., institutional
firms with the intention to go public. In investors such as pensions funds) are
general, mature large corporations hardly viewed as outsiders.
face financing constraints and finance It is not obvious though that insider
investment overwhelmingly from retained control enhances performance. For instance,
earnings.11 Cash-flush corporations may individuals or families are often large
thus rather overinvest to keep funds in the blockholders because they are the founders
firm instead of “wasting” them on the of the company or the heirs. Given an
shareholders (Jensen, 1986; Hellwig, 2000). emotional involvement, these owners may
Strong owners could prevent managers be more strongly interested in the success
from stashing money away and thus reduce of the firm than investors for which the
both free cash flow and inefficient invest- firm is just one piece in their portfolios.
ments. Profitability should then be enhanced, A large shareholder “personally” attached
but this is not for sure, since large share- to the firm possibly motivates managers
holders may overdo monitoring. As shown and employees to more commitment and
by Burkart, Gromb, and Panunzi (1997), determination so that firm performance
too much monitoring can stifle managerial improves. The opposite may hold if running
initiative. The governed corporation may the firm is impeded because the owners
then miss out on profitable investment engage in infighting or take on a know-all
opportunities. attitude and try to “mastermind” decision-
Large shareholders can inflict substantial making. In this respect, investors without a
costs on other shareholders as well as stake- personal interest may be better for corporate
holders (managers, employees, creditors) in performance. Banks can be such neutral
the form of an expropriation-like redistrib- investors. If they are more efficient monitors
ution of wealth if their interests diverge (Diamond, 1984, 1991) and reduce agency
from those of other investors or stakeholders costs, performance should be enhanced.
(Shleifer and Vishny, 1997). Further, large Their actual governance role, either as own-
share- or stakeholders and managers may ers, board representatives or proxy voters,
collude to keep minority shareholders at bay and the impact on corporate performance
(Hellwig, 2000). Finally, large shareholders has been disputed.12 Not much is known
closely associated with the management about industrial firms and holdings as
may refrain from attempts to dismiss blockholders of other industrial firms.
underperforming managers if such an Empirical evidence on large diversified US
intervention puts their own reputation at corporations (Lang and Stulz, 1994; Shin
risk (Mayer, 1996). and Stulz, 1998) suggests that members of
business groups may underperform because
The identity of owners internal capital markets do not work
efficiently. In short, it is open to empirical
The commitment of owners and their testing whether the identity of owners
willingness to intervene may crucially really matters.
DOES THE GOVERNED CORPORATION PERFORM BETTER? 495
impact of ownership concentration on the reporting firms with complete and consistent
return on total assets for a sample of large data. According to their main field of
quoted corporations. Thus profitability economic activity, the sample firms were
first decreases in ownership concentration assigned to 30 different two-digit industries,
and then, at higher levels of concentration, among them machinery (76 firms), chemi-
rises again.17 Becht (1999) reports a cals and pharmaceuticals (60 firms), the
negative effect of ownership concentration electronic products industry (56 firms),
on liquidity (measured by the ratio of and iron and steel (37 firms).
turnover to market capitalization) which Our preferred measure of corporate
supports the hypothesis of a control-liquidity performance is the return on total assets
trade-off.18 (ROA). We also report summary statistics
In sum, “governed” (more concentrated for the return on equity (ROE) but ROE
or owner-controlled) German firms seem to comparisons across firms may be distorted
have enjoyed higher returns during the by the leverage effect and differences in
1970s and early 1980s (Cable, Gorton and the user cost of capital. For both ROA
Schmid, Weigand). The positive impact and ROE the numerator is gross profits,
vanishes or turns even negative when data calculated as turnover minus expenses for
from the late 1980s and the 1990s are personnel and materials.19 Equity capital
included (Becht, Chirinko and Elston, is defined as shareholders’ equity plus
Franks and Mayer, Goergen, Gedajlovic reserves and pension liabilities.20
and Shapiro). The opening of markets and The firm-specific variables are constructed
increased international competition may from balance sheet and profit-and-loss
have altered the profitability-ownership account data as collected from either the
concentration relation since the late Hoppenstedt Bilanzdatenbank (a commer-
1980s. In the subsequent section we cially sold data source), the Bundesanzeiger
explore this conjecture using a large panel (a federal gazette, in which corporations
data set of both quoted and non-quoted are obliged by law to publish their annual
German corporations for the 1990s. financial statements), or annual reports
directly obtained from the corporations.21
We used unconsolidated company data
3. OWNERSHIP CONCENTRATION, whenever available and excluded pure
INSIDER CONTROL, AND holding companies.22
PERFORMANCE OF GERMAN FIRMS Secondary sources were consulted for
identifying owners, share distributions, and
3.1. Data, variables, and sample the composition of managing and supervisory
characteristics boards. The sources are Commerzbank’s
Wer gehört zu wem? (Who owns whom?,
The data set contains 361 firms from the 16th–18th edition), Bayerische Hypotheken-
German mining and manufacturing indus- und Wechselbank’s Wegweiser durch
tries. Although 300 of these firms have deutsche Aktiengesellschaften (Guide of
the legal form of stock corporations German Stock Corporations, “Hypo-
(Aktiengesellschaften), only 183 were offi- Guide”, annual issues 1988–1996), and
cially listed and traded on German stock Hoppenstedt’s Börsenführer (Stock Market
exchanges during the observation period. Guide, annual issues, 1988–1998). In
As we are interested in the importance of combining these data sources it was
stock market exposure, our sample also possible to obtain a fairly precise picture of
includes non-traded stock corporations as voting stock ownership.23 For the purpose
well as some limited liability corporations of this study, we have defined a ‘large’
(GmbH, 54 firms) and limited commercial shareholder as one who controls at least
partnerships (GmbH & Co. KG, KGaA, 5 per cent of a firm’s voting capital. This
7 firms).The time period covered is 1991 to cut-off point of when a shareholder is large
1996, which yielded the largest number of rather than small is less important than in
DOES THE GOVERNED CORPORATION PERFORM BETTER? 497
Table 19.1 Ownership structures: sample of 361 German manufacturing firms, 1991–1996
The sample consists of 361 German firms which operated in 30 industries of the mining and manufacturing
sector during the period 1991 to 1996. Of the sample firms 300 have the legal form of stock corporations
(Aktiengesellschaften) with 183 listed and traded on German stock exchanges. The data set further includes
54 limited liability corporations (GmbH) and 7 limited commercial partnerships (GmbH & Co. KG, KGaA). For
identification of owners and share distributions we used Commerzbank, Wer gehört zu wem? (Who owns
whom?, editions 16th to 18th, publication years 1988, 1992, 1994), Bayerische Hypotheken- und Wechselbank,
Wegweiser durch deutsche Aktiengesellschaften (“Hypo-Guide”, Guide of German Stock Corporations, annual
issues 1988–1996), and Hoppenstedt’s Börsenführer (annual issues, 1988–1998). We define a large shareholder
as controlling at least 5 per cent of a corporation’s voting capital.
498 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
corporate charter). The Herfindahl index, them to the group of firms controlled by
defined as the sum of squared individual families or (voting pools of) individuals.
stakes, has the advantage of accounting for For some firms such as Bayer,
an asymmetric dispersion of shares among Mannesmann, Schering, Siemens, or VEBA
different shareholders.24 more than 90% of the voting capital is
Ownership concentration may not suffice generally reported as “widely dispersed”.
as an indicator of the degree of “gover- The study of Baums and Fraune (1995)
nance”. The identity of owners may play shows that in the 1992 annual shareholders’
a more crucial role. Officially reported meetings banks controlled an aggregate of
shareholdings often do not reflect the more than 90% of those firms’ voting
true extent of voting control exercised by capital via associated investment compa-
owners of German corporations (Becht nies and proxy votes. These sample firms
and Boehmer, 1997; Boehmer, 1999). were therefore added to the group of
Complex cross-holding arrangements and firms with financial institutions as largest
unreported fiduciary and dormant voting shareholders.
rights complicate the identification of To relate back to the discussion on the
actual controllers. Nevertheless, given the managed vs. the governed firm, or the
information available, we attempt to assess insider vs. outsider model, corporations
the relevance of the owner’s identity for the having individuals or families (FAMILY),
link between profitability and ownership banks (FININST), or different independent
concentration. large shareholders (MIX) may be inter-
For the purpose of grouping firms by the preted as potentially governed by committed
location of control rights five broad cate- insiders. By contrast, for firms owned by
gories of direct owners can be identified: another industrial firm (INDFIRM) it
another industrial firm or holding company might be managers rather than the ultimate
(INDFIRM), families or (voting pools of) owners who control the managers of the
individuals (FAMILY), financial institutions owned company. Firms of this group,
(banks or bank-owned investment compa- mostly subsidiaries of large quoted stock
nies, insurance companies, FININST), corporations (e.g., BASF, Thyssen), may
different large shareholders (MIX, e.g., then be considered as managed. However,
industrial firms and investment companies), we make no attempt here to argue that
and foreign owners (FOREIGN). A sixth a certain location of control rights indeed
group, CHANGE, is defined to account implies “more” or “less” governance.
for 26 firms for which a turnover of Although we took great efforts to trace
blockholdings from one of these owner ownership back to the ultimate owners for
categories to another was observed during all firms, it is not entirely clear whether
the observation period. these ultimate owners really “govern”, or
The groups are mutually exclusive. A whether hired managers nonetheless exercise
firm that, for instance, had another industrial control even if ownership concentration
firm as largest shareholder in some years is high and firms are owned by families
and a bank in other years only appears in for example.
the CHANGE group but not in the groups To investigate whether stock market
INDFIRM or FININST. Due to the lack of exposure makes a difference for the potential
detailed ownership information, the foreign- link between governance indicators and
owned firms were also treated as a separate corporate performance we distinguish
group. A few firms are owned by foundations between corporations traded on the stock
which have no owner in a strict sense. It is exchange (QUOTED) and non-traded or
thus unclear who really controls these firms. non-stock corporations (NON-QUOTED).
As the foundations are often close to the Table 19.2A presents the mean and
founding families of the firms, as in the median values of selected variables for the
case of Bosch GmbH (a leading electronic full sample and the groups of quoted and
products firm), it seems tenable to assign non-quoted firms. The last column reports
DOES THE GOVERNED CORPORATION PERFORM BETTER? 499
Table 19.2A Summary statistics: sample of 361 German manufacturing firms, 1991–1996.
Sample split by stock market exposure
Calculations are based on data from annual reports of 361 German corporations (see Table 19.1). The
sample split is based on stock market quotation for at least three out of the six sample years. Return on total
assets (ROA) is gross profits (calculated from the firms’ profit-and-loss statement as turnover minus expenses
for personnel and materials) divided by the book value of total assets (as reported in the firms’ balance sheet).
Return on equity is gross profits divided by equity (calculated as shareholders’ equity plus reserves and
pension liabilities). Ownership concentration is measured as the sum of squared individual percentage stakes,
thus bounded to lie in the interval [0; 10,000], and alternatively, as the percentage stake of the largest
shareholder. Board representation is an indicator variable taking on unit value if an identified owner served
on the board of executive directors in a certain year, otherwise it has zero value. Turnover is in millions DM.
Employment is the reporting year average number of full-time employees as provided in the firms’ annual
report or in secondary sources. In some cases end-of-the-year figures had to be used because information
on average employment was not available. We measure absolute firm size as the natural logarithm (log) of
total assets to reduce the skewness of the firm size distribution. Firm growth is the annual logarithmic change
in real turnover (turnover deflated by the 1995 GDP deflator, source: German Statistical Office). Capital
intensity is defined as log (total assets/employees). Capital structure is equity (as defined above) over total
capital (book values). Market concentration is measured by a standard Herfindahl index at the two-digit
industry level of the German industry classification for the manufacturing sector (“SYPRO”, source: German
Statistical Office).
Means and medians of all variables are calculated for the period 1991 to 1996 with the exception of firm
growth which, due to its definition, is available for the period 1992 to 1996. We compare the variables’ means
and medians between the two groups of firms, quoted versus non-quoted firms. The last column of the table
provides the t-statistic for the test of equal group means and the Mann– Whitney statistic for the test of equal
group medians.
*** ** *
/ / Significant at the 0.01/0.05/0.10 error level respectively.
500 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
statistics for testing the hypotheses of firms had significantly higher growth of
equal group means (t-statistic) and medians turnover than non-quoted firms.
(Mann–Whitney statistic). The table shows Table 19.2B contains summary statistics
that quoted firms have significantly higher of selected variables for the groups of firms
average returns on total assets and also as classified by the identity of the largest
higher median returns on equity.The quoted owner. For any pair of owner categories we
firms are less levered. Further, as measured conducted pairwise t- and median tests for
by turnover, employment, and the natural the variables listed in Table 19.2B but do
log of total assets, the quoted firms are not report these results separately.25 The
substantially larger on average but the tests reveal that the INDFIRM firms have
median firms do not differ significantly in significantly lower rates of return than any
turnover or total assets. Finally, quoted other owner group except for the equity
Table 19.2B Summary statistics: sample of 361 German manufacturing firms, 1991–1996.
Sample split by the location of control rights
Calculations are based on data from annual reports and secondary sources of 361 German manufacturing firms
(see Table 19.1). The sample split is based on the location of control rights (voting stock). We define a large
shareholder as controlling at least 5 per cent of a corporation’s voting capital and distinguish six identities of
large shareholders: 1. INDFIRM is defined as firms having another independent industrial firm or a holding
company as largest shareholder (e.g., THYSSEN GUSS AG, subsidiary of THYSSEN concern). 2. FAMILY is
defined as firms having (pools of) individuals or families as largest shareholders (e.g., BAUSCH AG, BOSCH
GmbH). 3. FININST is defined as firms having banks, insurance companies, or associated investment companies
as largest shareholders (e.g., LINDE, larger stakes owned by Deutsche Bank and Commerzbank), or, having
widely dispersed shareholdings, but banks control at least 75 per cent of the voting capital through proxy voting
rights (e.g., BAYER). 4. MIX is defined as firms having different independent large shareholders (e.g., BOSCH-
SIEMENS Hausgeräte GmbH, owned equally by BOSCH GmbH and SIEMENS AG). 5. FOREIGN is defined
as firms having foreign companies as largest shareholders (e.g., OPEL AG, owned by GM). 6. CHANGE is
defined as firms which experienced a change in the identity of blockholders through turnovers of blocks from
one of the owner categories 1–5 to another (e.g., AQUA SIGNAL AG). See Table 19.2A for definitions of the
other variables.
DOES THE GOVERNED CORPORATION PERFORM BETTER? 501
Table 19.4A ignores stock market exposure. concentration are highly significant. Larger
Column 2 then presents the robust estimates firms and firms with a higher share of equity
for the quoted firms, while column 3 gives capital (lower leverage) have significantly
the coefficient differences with respect to the lower returns, which is consistent with diver-
(not separately reported) regression coeffi- sification and risk-return considerations.
cients of the non-quoted firms.We also report Firm growth affects ROA positively which
the group means of the firm-specific effects. supports the argument that growing firms
The fixed effects measure the average have better investment opportunities.
systematic difference in ROA across groups The negative coefficient on capital intensity
of firms not accounted for by the model’s seems to imply decreasing returns to scale.
explanatory variables. There is no statisti- Given the above-mentioned fact that the
cally significant difference in the means of time effects become insignificant when
these firm-specific intercepts between the capital intensity enters the regression, this
traded and non-traded firms.31 Stock variable is apparently a proxy for capital
market exposure does also not change the utilization and thus picks up business cycle
profitability-ownership concentration rela- effects. Market concentration has a
tionship. The negative impact of ownership positive but statistically insignificant
concentration is a bit weaker for the quoted effect. Stock market exposure does not
firms (0.0027) than for the non-quoted change the impact of the IO variables on
firms (0.0049) but still significant at the ROA in any significant way.
0.05 error level. The coefficient difference
(0.0022) is not statistically significant.
Does the identity of owners matter?
These results contradict both the hypothesis
of the beneficial large shareholder as As in the case of the industry indicators, we
well as the Demsetz–Lehn hypothesis of no cannot consider the identity of owners in a
ownership effect. However, the estimated fixed-effects regression directly, since the
coefficient on OC for the full (non-split) indicators for the location of control rights
sample implies a quantitatively small effect. lack time series variation and thus will be
Evaluated at the sample means, doubling perfectly correlated with the firm-specific
the unbounded Herfindahl index would lower effects.We therefore follow two approaches.
ROA by about 8%. In the first approach, we investigate the
There is a notable difference between the indirect impact of the location of control
two groups with respect to board represen- rights on ROA by interacting the identity
tation.The coefficient on B is insignificantly indicators with all time-varying right-hand
positive (and remains so when ownership side variables of (1).32 The second approach
concentration is excluded from the regres- homes in on the estimated fixed firm
sions) for the non-split sample and for the effects.The fixed effects can be interpreted as
quoted firms but significantly positive for the autonomous component of ROA neither
the non-traded firms. For the non-traded explained by the explanatory variables nor
firms the presence of the largest owner on by purely stochastic disturbances. By
the board of executive directors thus regressing the fixed effects on the identity
improves profitability. Assuming that infor- indicators we obtain a direct multivariate
mational problems are more pronounced test on how the location of control rights
for non-traded firms this result implies that affects ROA, after having controlled for
tighter control through the owner reduces the time-varying explanatory variables’
agency costs and improves firm perform- influence.
ance.The estimated coefficient on B for the The regression results from the indirect
non-quoted firms suggests an increase of approach are contained in Table 19.4B.The
average ROA by about 3% if average group of firms with another firm as largest
board representation raises by 1%. shareholder, INDFIRM, serves as the base
For the IO variables we find that all group. Column 1 presents the coefficient
coefficients except for the one on market estimates for this base group, while the
Table 19.3A Panel regression estimates of the return on total assets equation, 1992–1996. Full sample
Test statistic
Extended model
Specification test Null hypothesis Parsimonious model Accept/Reject null
***
Model A vs. Model B No fixed industry effects F(29, 325) 56.24 Reject
F(29, 330) 58.79*** Reject
Model C vs. Model D No fixed industry effects F(29, 1762) 240.79*** Reject
F(29, 1767) 228.71*** Reject
Model C vs. Model E No fixed firm effects F(360, 1431) 16.01*** Reject
F(360, 1436) 14.29*** Reject
506 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
The estimating sample contains 361 German firms as described in Table I. The extended model is given by
ROAit b1OCit b2Bit b3Sit b4Git b5Kit b6CI, t1 b7Hjit uit,
uit ai aj t it, (1)
in which ROA, OC, B, S, G, K, C, and H are the return on total assets, ownership concentration (unbounded Herfindahl index), board representation, firm size, firm growth,
capital intensity, capital structure, and market concentration of firm i operating in industry j at time t. See Table 19.2A for definitions of variables. In the parsimonious model
we only include ownership concentration and board representation as right-hand side variables and ignore all variables implied by the industrial organization literature. The
regression disturbance uit can be decomposed into firm-, industry- and time-specific effects, ai, aj, t respectively, and a classical white noise regression disturbance it.
Model A is a standard cross-section regression with the regression variables averaged over the observation period (361 observations). Model B adds 29 industry dum-
mies to Model A. Models C to H use the pooled cross-section time-series data (1805 pooled observations). Model C restricts the firm-specific effects to be constant and
identical for all firms. Model D adds 29 industry dummies to Model C. Model E assumes fixed (constant) firm-specific effects (361 firm-specific intercepts). Other time-
constant effects (e.g., industries dummies) cannot be included. Model F assumes random firm-specific effects (i.e., a common intercept from which individual firms may
deviate randomly). Model G adds 29 industry dummies to Model F. Model H includes fixed firm-specific effects as Model E but treats OC as an endogenous right-hand side
variable (i.e., being correlated with it). Models A, B, C and D are estimated by OLS. Model E is Within-OLS (LSDV, Least Squares Dummy Variables estimator). Models
F and G use GLS. Model H is Within-2SLS using the fitted values of OC from the reduced-form (first-stage) regression of OC on all exogenous right-hand side variables in
(1), B, S, G, K, C, H plus the number of large shareholders as an additional instrument included to identify the regression equation. The number of large shareholders has
been excluded from the second-stage OC regression (to be presented in Table 19.3B). We tested alternative exclusions of variables, since picking a specific variable is –
admittedly – arbitrary. The results were not significantly different when any other pair of the IO variables was excluded.
We present specification test statistics for these models. A standard F-test is employed to test a restricted model (e.g., no industry or no time effects) against an unre-
stricted model (allowing for such effects).To test for random firm-specific effects versus no such effects we use the Lagrange multiplier test suggested by Breusch and Pagan
(1980). Hausman’s (1978) specification test is used to decide on fixed versus random firm-specific effects.The same test, known in this case under the name “Wu-Hausman”
(Wu, 1972) is also applied to compare the Within-OLS estimates (which are inefficient if OC in Equation (1) is endogenous) with the Within-2SLS estimates (which are
efficient if OC is indeed endogenous). “Durbin–Watson” tests for first-order serial correlation (AR1) of panel regression residuals (Bhargava et al., 1982). See Hsiao (1986)
or Baltagi (1995) for a detailed discussion of panel regression models and specification tests. The estimates’ absolute t-statistics as provided in parentheses are based on
White’s (1980) heteroskedasticity-robust variance-covariance estimator.
*** ** *
/ / Significant at the 0.01/0.05/0.10 error level respectively.
DOES THE GOVERNED CORPORATION PERFORM BETTER? 507
508 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Test statistic
Extended model
Specification test Null hypothesis Parsimonious model Accept/Reject null
Model A vs. Model B No fixed industry effects F(29, 325) 11.94*** Reject
F(29, 330) 13.19*** Reject
Model C vs. Model D No fixed industry effects F(29, 1762) 64.49*** Reject
F(29, 1767) 68.51*** Reject
Model C vs. Model E No fixed firm effects F(360, 1431) 128.97*** Reject
510 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
The estimating sample contains 361 German firms as described in Table 19.1. The split criterion is stock
market quotation which divides the full sample into 183 traded and 178 non-traded firms. The estimating
regression is
ROAit = b1OCit b2Bit b3Sit b4Git b5Kit b6CI,t 1 b7Hjit
d1Dit OCit d2Dit Bit d3Dit Sit d4Dit Git
d5Dit Kit d6Dit Ci,t1 d7Dit Hjit ai t it
in which ROA, OC, B, S, G, K, C, and H are the return on total assets, ownership concentration (unbounded
Herfindahl index), board representation, firm size, firm growth, capital intensity, capital structure, and market
concentration of firm i operating in industry j at time t. See Table 19.2A for definitions of variables. ai is a firm-
specific effect, t a time dummy, and it the stochastic disturbance.
The split sample estimates are based on the full sample regression with the set of explanatory variables
(except for the fixed firm-specific effects) interacted with a dummy variable D which takes on unit value for non-
quoted firms and zero value for quoted firms.The regression coefficients b estimate the impact of the respective
explanatory variables on ROA for the quoted firms, while the coefficients d yield the difference between the
coefficients b and the respective coefficients for the non-quoted firms, to be obtained by adding d’s to matching
b’s, or directly, by estimating the regression a second time but with D defined the other way around. The
“All firms” estimates (column 1) come from the regression excluding the set of interacted variables. For
the governance variables, these latter estimates can be compared to the Within-OLS estimates presented in
Table 19.3A (Model E) which are not robust to first-order serial correlation.
The estimates shown above are robust to heteroskedasticity (White, 1980) and first-order serial correlation
(AR1) (see Hsiao, 1986, pp. 55).The Within-OLS estimator was employed to generate consistent first-step esti-
mates. We used the Within residuals to estimate a firm-specific serial correlation coefficient rho. Alternatively,
we restricted the model to have a serial correlation coefficient common to all firms.The rhos were used to trans-
form the original data. Regressions based on the transformed data were then estimated by OLS (firm-specific
rho) or Maximum Likelihood (common rho). The drawback of the “firm-specific rho” approach is that due to
the larger number of firm-specific coefficients to be estimated the coefficient on board representation was not
estimable for the group of non-quoted firms.Therefore, we only report the estimates based on the “common rho”
approach. However, the coefficients on all other explanatory variables can be estimated in the “firm-specific
rho” approach. As these latter coefficient estimates are qualitatively identical and quantitatively only marginally
different from those shown below, we do not report them separately.The estimates are available from the authors
on request. The first Durbin–Watson statistic refers to the first-step Within-OLS residuals, the second statistic
is based on the residuals from the AR1 regression. The adjusted R squared is given for both the original and, in
parentheses, the transformed data.
*** ** *
/ / Significant at the 0.01/0.05/0.10 error level respectively.
DOES THE GOVERNED CORPORATION PERFORM BETTER? 513
Table 19.4B Robust estimates of the return on total assets equation, 1992–1996. Sample split by
location of control rights
The estimating sample contains 361 German firms as described in Table I. The split criterion is the location of control rights
(identity of voting block owners). The estimating regression is
in which ROA, OC, B, S, G, K, C, and H are the return on total assets, ownership concentration (unbounded Herfindahl index),
board representation, firm size, firm growth, capital intensity, capital structure, and market concentration of firm i operating in
industry j at time t. ai is a firm-specific effect, t a time dummy, and it the stochastic disturbance.
The split sample estimates are based on the full sample regression with the set of explanatory variables x {OC, B, S, G,
K, C, H} interacted with a set of dummy variables Dz which take on unit value for firms from z {FAMILY, FININST, MIX,
FOREIGN, CHANGE} respectively and zero value for other firms. See Tables 19.2A and 19.2B for definitions of variables. The
regression coefficients b estimate the impact of the respective explanatory variables on ROA for firms from the INDFIRM
group, arbitrarily chosen as the base group, while the coefficients d yield the difference between the coefficients b and the coef-
ficients for other groups of firms as represented by z. Adding the respective “coefficient difference” to the base group’s “regres-
sion coefficient”, yields the regression coefficients for the other groups of firms.
The estimates shown above are robust to heteroskedasticity (White, 1980) and first-order serial correlation (AR1) (see
Hsiao, 1986, pp. 55). The Within-OLS estimator was employed to generate consistent first-step estimates. We used the Within
residuals to estimate a firm-specific serial correlation coefficient rho. Alternatively, we restricted the model to have a serial
correlation coefficient common to all firms. The rhos were used to transform the original data. Regressions based on the trans-
formed data were then estimated by OLS (firm-specific rho) or Maximum Likelihood (common rho). The coefficient on board
representation was not estimable for the firms in MIX and CHANGE due to an insufficient number of observations. In line with
Table 19.4A, we report the estimates based on the “common rho” approach. Again, the coefficients of all other explanatory
variables can be estimated in the “firm-specific rho” approach. These latter coefficient estimates are qualitatively identical and
quantitatively only marginally different from those shown below so we do not report them separately.The estimates are available
from the authors on request. The first Durbin–Watson statistic refers to the first-step Within-OLS residuals, the second statis-
tic is based on the residuals from the AR1 regression. The adjusted R squared is given for both the original and, in parentheses,
the transformed data.
– Not estimable due to an insufficient number of observations in the respective group.
*** ** *
/ / Significant at the 0.01/0.05/0.10 error level respectively.
514 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 19.5 OLS estimates of the firm-specific fixed effects equation, 1992–1996
The estimating sample contains 361 German firms as described in Table 19.1. The estimating regression is
âi 1 2FAMILYi 3FININSTi 4MIXi 5FOREIGNi 6CHANGEi
d1 Di INDFIRMi d2Di FAMILYi d3Di FININSTi d4Di MIXi
d5Di FOREIGNi d6Di CHANGEi aj i
with
aˆi ROAi b̂1OCi b̂2Bi b̂3Si b̂4Gi b̂5Ki b̂6Ci b̂7Hji.
The firm-specific effects are denoted by ROAi, OCi, Bi, Si, Gi, Ki, Ci, Hij are the time means (1992–1996) of
the return on total assets, ownership concentration (unbounded Herfindahl index), board representation, firm
size, firm growth, capital intensity, capital structure, and market concentration of firm i operating in industry j.
See Table 19.2A for variable definitions. INDFIRM, FAMILY, FININST, MIX, FOREIGN, and CHANGE are the
mutually excluding groups referring to the location of control rights as defined in Table 19.2B. aj is an industry-
specific effect and ‡i the stochastic disturbance.
The coefficients b̂k (k 1, . . . , 7) are the robust estimates as shown in Table 19.2A, column 1. The
coefficients 2 to 6 estimate the difference between the average fixed firm-specific effect 1 of the chosen base
group INDFIRM and the average fixed firm-specific effect of the other groups. Adding the respective coefficient
difference 2 to 6 to the base group’s regression coefficient 1, yields the regression coefficients for the other
groups of firms.The dummy variable D takes on unit value for non-quoted firms and zero value for quoted firms.
The coefficients d estimate the difference in the coefficients between quoted and non-quoted firms. The OLS
cross-section estimates below are robust to heteroskedasticity (White, 1980).
*** ** *
/ / Significant at the 0.01/0.05/0.10 error level.
DOES THE GOVERNED CORPORATION PERFORM BETTER? 515
of all, for his encouragement. We are also 2 See Short (1994) and Shleifer and Vishny
grateful to an anonymous referee for (1997) for surveys of the field.
substantial advice which greatly improved 3 Allen and Gale (2000, ch. 2 and 3) provide
the paper. Many helpful suggestions on an excellent overview of the financial systems in
the USA, UK, Germany, Japan, and France.
various drafts were received from Rashjree
4 See, e.g., Porter (1992), Jensen (1993),
Agarwal, David Audretsch, Matthew Roe (1994), Bhide (1994), Pound (1995). As
Bennett, Hans Degryse, Paul De Grauwe, Bhide (1994, p. 129, p. 131) argues,“Unwittingly,
Julie Ann Elston, Oliver Fabel, Rainer the system [of U.S. securities regulations
Feuerstack, Günther Franke, Carolin Frohlin, and disclosure rules] nurtures market liquidity
George Gelauff, Sabine Langner, Manfred at the expense of good governance. . . . U.S.
Neumann, Marco Pagano, Hans-Jürgen rules that protect investors don’t just sustain
Ramser, and seminar participants at the market liquidity, they also drive a wedge
1999 Meetings of the European Finance between shareholders and managers. Instead of
Association (EFA, Helsinki), European yielding long-term shareholders who concen-
trate their holdings in a few companies, where
Economic Association (EEA, Santiago de
they provide informed oversight and counsel,
Compostela), European Association for the laws promote diffused, arm’s length stock-
Research in Industrial Economics (EARIE, holding”.The benefits typically attributed to this
Torino), and Western Economic Association so-called market-based system are seen in a
(WEA, San Diego) as well as at the better provision of finance to innovative start-up
University of Central Florida Orlando, firms and higher returns to investors.
Humboldt University Berlin, University of 5 In December 1998, 323 domestic corpora-
Konstanz, and Katholieke Universiteit tions with an aggregate market capitalization of
Leuven. The second author would like to 1,822,103 million DM (equivalent to 48% of
thank Ulrich Hege, Piet Moerland, and the 1998 German GDP) were quoted and offi-
cially traded on the Frankfurt stock exchange.
Luc Renneboog for stimulating discussions
At the same time, 2,399 domestic corporations
on corporate governance. As a matter of (3,829,375 million DM, 182% of GDP) were
fact, any remaining errors are the authors’ listed on the London stock exchange and 2,722
sole responsibility. Generous financial (17,373,835 million DM, 105% of GDP)
support was provided by the Bundesland on the New York stock exchange. (Source:
Mecklenburg-Vorpommern (first author) Deutsche Börse, Factbook, Frankfurt, http://
and the German Science Foundation (DFG, www.exchange.de; own calculations.) Boehmer
second author). Main parts of the paper (1999) provides a thorough analysis of owner-
were written while the second author was ship structures and location of control rights for
visiting Georgia State University (School the 430 stock corporations which were officially
traded on the Frankfurt stock exchange in
of Policy Studies, Atlanta) and Indiana
1996. According to this study, banks, industrial
University (School of Public and firms, holdings, and insurance companies
Environmental Affairs, Bloomington). The controlled as large blockholders almost 80%
hospitality and support of these institutions of the median firm’s voting rights, or roughly
is gratefully acknowledged. This research 50% of the overall market value of firms
is not related to any CPB Netherlands officially listed.
Bureau for Economic Policy Analysis 6 The high tax on the sale of company cross-
projects. The opinions expressed here do holdings (to be abolished soon) and the rigid
not necessarily reflect those of the CPB. German takeover code (under revision) which
requires shareholders to be bought out with cash
have contributed to making takeovers unattractive.
NOTES 7 See, e.g., Edwards and Fischer (1994),
Perlitz and Seeger (1994), Audretsch and
1 See for a discussion and references, e.g., Elston (1997), or Wenger and Kaserer (1998).
Porter (1992), Jensen (1993), Blair (1995), The implementation of the Neuer Markt (new
Berglöf (1997), Shleifer and Vishny (1997), market) on which shares in innovative young
Mayer (1998), Hopt et al. (1998), Barca and firms are traded can be seen as a step in the
Becht (1999), OECD (1999), Allen and Gale direction of a more market-based system of
(2000), and Vives (2000). corporate finance.
518 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
8 See, e.g., Fama and Jensen (1983), Short corporations over the observation period 1965
(1994) and Shleifer and Vishney (1997) for dis- to 1990 but due to incomplete ownership infor-
cussion and references to seminal contributions. mation cannot take advantage of the panel data
9 In Germany, the case of Metallgesellschaft properties. Weigand uses a panel data set which
may serve as an example of supervisory board includes detailed ownership data (identity of
failure. Deutsche Bank, as a major debtholder owners, outstanding shares) for 240 stock
represented on the supervisory board, was corporations during 1965 to 1986.
regarded responsible for the disaster and 16 Franks and Mayer (1997) apply a panel
scolded in the business press for insufficient data set of 171 quoted corporations over the
and bad monitoring. See Frankel and Palmer period 1989 to 1994 and use a fixed effects
(1996) for further discussion. See also Franks panel regression approach. There is only weak
and Mayer (1998) on the dubious role of banks evidence of a relationship between either
in the few German corporate takeovers. management or supervisory board turnover and
10 What insufficient control over a firm’s performance in firms with concentrated owner-
assets means, is nicely depicted by the General ship. This result is consistent with Kaplan’s
Motors example cited in Roe (1994, XII). In (1994) findings. Goergen’s (1999) sample
1990, two ‘large’ (relative to all other share- consists of 86 IPOs during the period 1981 to
holders) institutional investors, being dissatisfied 1988. In both static and dynamic panel data
with the company’s bad performance, wanted to regressions, ownership concentration did not
negotiate the implementation of a new CEO have an impact on profitability (measured by
with GM’s leaders. The management could cash flow over total assets).
calmly decline the request, since each ‘large’ 17 Gedajlovic and Shapiro (1998) studied the
shareholder only accounted for less than one blockholdings-performance link for the largest
per cent of the voting capital. In 1992 GM’s top firms from five major industrialized countries.
management had to take action after all. As Their sample includes 99 publicly traded German
losses piled up to over $6 billion, CEO Robert stock corporations over the period 1986 to
Stempel was eventually fired. 1991. The U-formed relationship between
11 See Hubbard (1998) for a recent survey ownership concentration and ROA was also
on investment and capital market imperfections. found for the US firms.
12 See, e.g., Edwards and Fischer (1994) or 18 Becht (1999) uses the corporations
Hellwig (2000) for discussion and references. represented in the DAX100 stock market index
13 Boehmer (1999, 2000) provides in-depth during 1996 to 1998.
descriptive analyses of governance structures 19 This definition is equivalent to earnings
and stock market performance of traded German before interest, taxes, and depreciation
corporations. As his approach is different from (EBITDA).
the regression studies surveyed here, we will 20 Pension liabilities have been added for two
refer to some of his results below. reasons. First, it is peculiar to the German
14 A firm was defined as owner-controlled if system of accounting that pension assets
individuals or families held at least the blocking and pension liabilities are not netted out in
minority (25% of voting capital plus one vote). companies’ balance sheets. Further, pension
A drawback of the study is that ownership liabilities are not paid into a trust (pension
structures were identifiable only for about 90 fund) but remain within the firm. They are
out of 300 sample firms. Thus in the regression available to the firm as a source of internal
analysis owner-controlled firms had to be long-term finance. Pension liabilities thus can be
compared to a mixture of presumably manager- seen as ‘quasi’ equity. Second, the shareholders’
controlled firms and firms with unknown equity of limited liability companies (GmbH)
governance structures, rendering the results is, by legal construction, extremely low. Adding
questionable. reserves and pension liabilities to shareholders’
15 Cable (1985), Chirinko and Elston (1996) equity helps avoid generating unrealistically
and Gorton and Schmid (2000) apply a cross- high returns on equity for these firms compared
section regression approach. Cable’s analysis to stock corporations.
for the time period 1968 to 1972 is based on 48 21 We referred to the Bundesanzeiger or
out of Germany’s 100 largest corporations in requested annual reports from firms to double-
1970. The Gorton and Schmid study is also check and correct some entries in the
based on a very small sample and only two years Hoppenstedt Bilanzdatenbank (inconsistencies
(82 firms in 1975) and (56 firms in 1986). in reporting, missing values, obviously wrong
Chirinko and Elston investigate 300 stock entries etc.).
DOES THE GOVERNED CORPORATION PERFORM BETTER? 519
22 This exclusion refers to the subsequent 30 With respect to the IO variables in the
empirical analysis of profitability only. Of extended model we find ownership concentra-
course, a holding company can be the owner of tion to be significantly lower for firms which
a sample firm. are larger, use more capital per employee, have
23 In many cases the Hypo-Guide also lists more equity capital, and, hardly surprising,
blocks smaller than 25% and indicates indirect have more large shareholders. By contrast,
ownership (voting rights granted to a large growing firms and firms operating under more
shareholder from other shareholders). Sometimes concentrated market structures have more con-
even very small blockholdings (5% and less) centrated ownership. In an alternative try,
are reported. we endogenized board representation and the
24 As disclosure of smaller blockholdings was number of large shareholders. The results were
not mandatory during the observation period, qualitatively identical. The Wu–Hausman test
the calculated Herfindahl index can only be an did not indicate the endogeneity of board repre-
approximation to the true degree of ownership sentation or the number of large shareholders.
concentration. Two alternative Herfindahl 31 As some of the variables in the regression
indices were constructed for corporations with are expressed in the natural logarithm, the
missing information on this “dispersed” portion means of the firm-specific intercepts are of
of shareholdings. One measure treats the “dis- different magnitude than mean ROA in the
persed” portion as one block. Using the means summary statistics tables.
in Table 19.1 for corporations with only one 32 This model specification gives estimates
large shareholder as an example, the Herfindahl identical to those from separate regressions for
index is (89.07)2 (10.93)2 8,053. The each group of owners but considers the across-
other measure assumes that the dispersed por- groups variation of the regression residuals in
tion is uniformly distributed among an unknown the variance-covariance matrix.
number of shareholders, each holding at most 33 Recall that the Herfindahl index can be
1% of the shares outstanding, which yields a decomposed into n1 nV where, in our case, n
Herfindahl index of (89.07)2 10 (1.00)2 and V denote the number of (large) shareholders
(0.93)2 7,944. The results to be presented and the variance of individual stakes respectively.
below are not affected by the choice of the 34 This is suggested by the Hausman specifi-
calculation method. cation test in Table 19.3A (model E vs. model F)
25 The results are available from the authors from which we concluded that the firm-
on request. specific effects are fixed rather than random.
26 The Herfindahl-Index in its standard The test does indeed check whether these
definition is restricted to take on values between effects are correlated with the right-hand
0 and 10,000. The logit transformation log variables or not.
[H/(10,000 H)] yields an “unbounded” 35 There are only two non-quoted firms with
variable. In the case of Mannesmann, for which bank involvement. These two firms operated
ownership of voting shares is reported to be with mean ROAs of 0.16 and 0.09 respectively
“100% dispersed”, H was set at 1 in the in the notoriously declining and restructuring
transformation. For firms with only one owner mining industry. The mean ROA for the quoted
holding 100% of voting stock, H was set at firms with bank involvement is 0.28.
9,999. In alternative regression runs we used
the bounded Herfindahl-Index as well as the
share of the largest shareholder plus its squared
value as in Gedajlovic and Shapiro (1998). REFERENCES
Further, we replaced ROA by ROE. As the
estimates with respect to governance indicators Admati, A., Pfleiderer, P., and Zechner, J. (1994),
do not differ significantly, we do not report Large shareholder activism, risk sharing,
the results from these alternative specifications and financial market equilibrium, Journal
here. They are available from the authors on of Political Economy 102, 1097–1130.
request.
Allen, F. and Gale, F. (1998), Corporate
27 See for a recent discussion, e.g., Church
and Ware (2000, ch. 12). Governance and Competition, Working paper,
28 See the discussion in Hall and Weiss New York University.
(1967). Allen, F. and Gale, D. (2000), Comparing
29 We discuss the impact of the IO variables Financial Systems. MIT Press, Cambridge
below. and London.
520 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Audretsch, D. and Elston, J. (1997), ‘Financing Bolton, P. and von Thadden, E.-L. (1998b),
the German Mittelstand’, Small Business ‘Liquidity and control’, Journal of Institutional
Economics Journal 9, 97–110. and Theoretical Economics 154, 177–215.
Audretsch, D. and Weigand, J. (1999), Does Breusch,T. and Pagan, A. (1980), ‘A simple test
Science Make a Difference? Investment, for heteroscedasticity and random coefficient
Finance and Corporate Governance in variation’, Econometrica 47, 1287–1294.
German Industries, CEPR working paper Burkhart, M., Gromb, D., and Panunzi, F.
No. 2056, London. (1997), ‘Large shareholders, monitoring, and
Baltagi, B. H. (1995), Econometric Analysis of the value of the firm’, Quarterly Journal of
Panel Data. Wiley, New York. Economics 112, 693–728.
Baumol, W. (1959), Business Behavior, Value, Cable, J. (1985), ‘Capital market information
and Growth. Macmillan, New York. and industrial performance: The role of
Baums, T. and Fraune, C. (1995), ‘Institution- West German banks’, Economic Journal 95,
elle Anleger und Publikumsgesellschaft: 118–132.
Eine empirische Untersuchung’, Die Chirinko, R. and Elston, J. (1996), Finance,
Aktiengesellschaft 40, 97–112. Control, and Profitability: An Evaluation of
Barca, F. and Becht, M. (eds.) (1999), German Bank Influence, American Institute
Ownership and Control in Europe. Oxford for Contemporary German Studies:The Johns
University Press, Oxford. Hopkins University, Working Paper No. 28,
Becht, M. (1999), ‘European corporate gover- Washington, D.C.
nance: Trading off liquidity against control’, Church, J. and Ware, R. (2000), Industrial
European Economic Review 43, 1049–1056. Organization: A Strategic Approach.
Becht, M. and Boehmer, E. (1997), Transparency McGraw-Hill, New York.
of Ownership and Control in Germany, Cubbin, J. and Leech, D. (1983), ‘The effect
Working paper, Humboldt University, Berlin. of shareholding dispersion on the degree of
Bergloef, E. (1997), ‘Reforming corporate control in British companies: Theory and
governance: Redirecting the European evidence’, Economic Journal 93, 351–369.
agenda’, Economic Policy, 93–123. Demsetz, H. (1983),‘The structure of ownership
Berle, A. and Means, G. (1932), The Modern and the theory of the firm’, Journal of Law
Corporation and Private Property. MacMillan, and Economics 26, 375–390.
New York. Demsetz, H. and Lehn, K. (1985), ‘The structure
Bhargava, A., Franzini, L., and Narendanathan, of corporate ownership: Causes and conse-
W. (1982), ‘Serial correlation and the fixed quences’, Journal of Political Economy 93,
effects model’, Review of Economic Studies 1155–1177.
49, 533–549. Diamond, D. (1984), ‘Financial intermediation
Bhide, A. (1994), ‘Efficient Markets, Deficient and delegated monitoring’, Review of
Governance’, Harvard Business Review, Economic Studies 51, 393–414.
November–December, 129–139. Diamond, D. (1991),‘Monitoring and reputation:
Blair, M. (1995), Ownership and Control: The choice between bank loans and directly
Rethinking Corporate Governance for the placed debt’, Journal of Political Economy 99,
Twenty-First Century, The Brookings 689–721.
Institution, Washington, D.C. Edwards, J. and Fischer, K. (1994), Banks,
Boehmer, E. (1999), Who Controls Germany? Finance, and Investment in Germany.
An Exploratory Analysis, Working paper, Cambridge University Press, Cambridge.
Humboldt University, Berlin. Fama, E. and Jensen, M. (1983), ‘Separation of
Boehmer, E. (2000), ‘Business groups, bank ownership and control’, Journal of Law and
control, and large shareholders: An analysis Economics 26, 301–325.
of German takeovers’, Journal of Financial Fluck, Z. (1999), ‘The dynamics of the
Intermediation 9, 117–148. management-shareholder conflict’, Review
Bolton, P. and von Thadden, E.-L. (1998a), of Financial Studies 12, 379–404.
‘Blocks, liquidity, and control’, Journal of Frankel, A. and Palmer, D. (1998),
Finance 53, 1–25. ‘The management of financial risks at
DOES THE GOVERNED CORPORATION PERFORM BETTER? 521
German nonfinancial firms: The case of Huddart, S. (1993), ‘The effects of a large
Metallgesellschaft, in S. Black and M. Moersch, shareholder on corporate value’, Management
eds., Competition and Convergence in Financial Science 39, 1407–1421.
Markets: The German and Anglo-American Jensen, M. C. (1986), ‘Agency costs of free cash
Models, Advances in Finance, Investment flow, corporate finance and takeovers’,
and Banking, vol. 5, Elsevier Science, North- American Economic Review 76, 323–329.
Holland, Amsterdam, pp. 41–78. Jensen, M. C. (1988), ‘Takeovers: Their causes
Franks, J. and Mayer, C. (1997), Ownership, and consequences’, Journal of Economic
Control and the Performance of German Perspectives 2, 21–48.
Corporations, Paper presented at the Jensen, M. C. (1993), ‘The modern industrial
Center for Financial Studies Conference, revolution, exit, and the failure of internal
November 19–20, Frankfurt a.M. control systems’, Journal of Finance 48,
Franks, J. and Mayer, C. (1998), ‘Bank control, 831–880.
takeovers, and corporate governance in Jensen, M. C. and Ruback, R. (1983).‘The market
Germany’, Journal of Banking and Finance for corporate control: The scientific evidence’,
22, 1385–1403. Journal of Financial Economics 11, 5–50.
Gedajlovic, E. and Shapiro, D. (1998), Kaplan, S. (1994), ‘Top executives, turnover,
‘Management and ownership effects: Evidence and firm performance in Germany’, Journal
from five countries’, Strategic Management of Law, Economics, and Organization 10,
Journal 19, 533–553. 142–159.
Goergen, M. (1999), Corporate Governance Lang, L. and Stulz, R. (1994), ‘Tobin’s q, corpo-
and Financial Performance. Edward Elgar, rate diversification, and firm performance’,
Cheltenham. Journal of Political Economy 102,
Gorton, G. and Schmid, F. (2000), ‘Universal 1248–1280.
banking and the performance of German Leibenstein, H. (1966), ‘Allocative efficiency vs.
firms’, Journal of Financial Economics 58, “X-Efficiency”’, American Economic Review
forthcoming. 56, 392–415.
Grossman, S. and Hart, O. (1980), ‘Takeover Manne, H. G. (1965), ‘Mergers and the market
bids, the free-rider problem and the theory of for corporate control’, Journal of Political
the corporation’, Bell Journal of Economics Economy 73, 110–120.
11, 42–64. Mayer, C. (1992), Corporate finance, in
Hausman, J. A. (1978), ‘Specification Tests in P. Newman, M. Milgate and J. Eatwell, eds.,
Econometrics’, Econometrica 46, 1251–1272. The New Palgrave Dictionary of Money &
Hall, M. and Weiss, L. (1967), ‘Firm size and Finance, New York.
profitability’, Review of Economics and Mayer, C. (1996), ‘Corporate governance, com-
Statistics 47, 319–331. petition, and performance’, OECD Economic
Hart, O. (1983), ‘The market mechanism as an Studies 27, 7–34.
incentive scheme’, Bell Journal of Economics Mayer, C. (1998), ‘Financial systems and
14, 366–382. corporate governance: A review of the inter-
Hellwig, M. (2000),‘On the economics and politics national evidence’, Journal of Institutional
of corporate finance and corporate control’, and Theoretical Economics 154, 144–165.
in Xavier Vives, ed., Corporate Governance, Moerland, P. W. (1995), ‘Corporate ownership
Cambridge University Press, Cambridge. and control structures: An international
Hopt, K., Kanda, H., Roe, M. J., Wymersch, E., comparison’, Review of Industrial Organization
and Prigge, S. (1998), Comparative Corporate 10, 443–464.
Governance: The State of the Art and OECD (1999), Principles of Corporate
Emerging Research, Clarendon Press, Oxford. Governance, OECD, Paris.
Hsiao, C. (1986), Analysis of Panel Data, Pagano, M. and Röell, A. (1998), ‘The choice of
Cambridge University Press, Cambridge. stock ownership structure: Agency costs,
Hubbard, R. G. (1998), ‘Capital-market monitoring, and the decision to go public’,
imperfections and investment’, Journal of Quarterly Journal of Economics 113,
Economic Literature 36, 193–225. 187–225.
522 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Perlitz, M. and Seger, F. (1994), ‘The role of Stulz, R. (1988), ‘Managerial control of voting
universal banks in German corporate rights: Financing policies and the market for
governance’, Business and the Contemporary corporate control’, Journal of Financial
World 4. Economics 20, 25–54.
Porter, M. E. (1992), ‘Capital disadvantage: Thompson, J. K. (1998), ‘Shareholder value and
America’s failing capital investment system’, the market for corporate control in OECD
Harvard Business Review, September– countries’, OECD Financial Market Trends
October, 65–82. 69, February, 15–37.
Pound, J. (1995), ‘The promise of the governed Thonet, P. J. and Poengsen, O. (1979),
corporation’, Harvard Business Review, ‘Managerial control and economic perform-
March–April, pp. 89–98. ance in West Germany’, Journal of Industrial
Roe, M. (1994), Strong Managers, Weak Economics 28, 23–37.
Owners: The Political Roots of American Vives, X. (ed.) (2000), Corporate Governance,
Corporate Finance, Princeton: Princeton Cambridge University Press, Cambridge.
University Press. Warner, J., Watts, R., and Wruck, K. (1988),
Schmidt, K. (1997), ‘Managerial incentives ‘Stock prices and top management changes’,
and product market competition’, Review of Journal of Financial Economics 20,
Economic Studies 64, 191–213. 461–492.
Shin, H.-H. and Stulz, R. (1998), ‘Are internal Weigand, J. (1999): Corporate Governance,
capital markets efficient?’, Quarterly Journal Profitability, and Capital Structure, unpublished
of Economics 113, 531–552. post-doctoral thesis (Habilitationsschrift),
Shleifer, A. and Vishny, R. (1986), ‘Large Nuernberg.
shareholders and corporate control’, Journal Wenger, E. and Kaserer, C. (1998), ‘The German
of Political Economy 95, 461–488. system of corporate governance – A model
Shleifer, A. and Vishny, R. (1988), ‘Management that should not be imitated’, in S. Black and
entrenchment: The case of manager-specific M.Moersch, eds., Competition and Convergence
investment’, Journal of Financial Economics in Financial Markets: The German and
25, 123–139. Anglo-American Models, Advances in Finance,
Shleifer, A. and Vishny, R. (1997), ‘A survey of Investment and Banking, vol. 5. Elsevier
corporate governance’, Journal of Finance Science, North-Holland, Amsterdam.
52, 737–783. White, H. (1980),‘A heteroscedasticity-consistent
Short, H. (1994), ‘Ownership, control, finan- covariance matrix estimator and a direct
cial structure, and the performance of test for heteroscedasticity’, Econometrica 48,
firms’, Journal of Economic Surveys 8, 817–838.
203–249. Wu, D. (1972), ‘Alternative tests of inde-
Stiglitz, J. (1985), ‘Credit markets and the pendence between stochastic regressors
control of capital’, Journal Money, Credit and disturbances’, Econometrica 41,
and Banking 17, 133–152. 733–750.
Chapter 20
ABSTRACT
Shareholder rights vary across firms. Using the incidence of 24 governance rules, we construct
a “Governance Index” to proxy for the level of shareholder rights at about 1500 large firms
during the 1990s. An investment strategy that bought firms in the lowest decile of the index
(strongest rights) and sold firms in the highest decile of the index (weakest rights) would have
earned abnormal returns of 8.5 percent per year during the sample period. We find that firms
with stronger shareholder rights had higher firm value, higher profits, higher sales growth, lower
capital expenditures, and made fewer corporate acquisitions.
Most research on the wealth impact of a “Governance Index” as a proxy for the
takeover defenses uses event-study method- balance of power between shareholders
ology, where firms’ stock returns are and managers. Our index construction is
analyzed following the announcement of straightforward: for every firm we add one
a new defense.1 Such studies face the point for every provision that reduces
difficulty that new defenses may be driven shareholder rights. This reduction of rights
by contemporaneous conditions at the firm; is obvious in most cases; the few ambiguous
i.e., adoption of a defense may both change cases are discussed. Firms in the highest
the governance structure and provide a decile of the index are placed in the
signal of managers’ private information “Dictatorship Portfolio” and are referred to
about impending takeover bids. Event as having the “highest management power”
studies of changes in state takeover laws or the “weakest shareholder rights”; firms
are mostly immune from this problem, but in the lowest decile of the index are
it is difficult to identify a single date for an placed in the “Democracy Portfolio” and
event that is preceded by legislative negoti- are described as having the “lowest
ation and followed by judicial uncertainty. management power” or the “strongest
For these and other reasons, some authors shareholder rights.”
argue that event-study methodology cannot In Section III we document the main
identify the impact of governance provisions.2 empirical relationships between gover-
We avoid these difficulties by taking nance and corporate performance. Using
a long-horizon approach. We combine a performance-attribution time-series regres-
large set of governance provisions into an sions from September 1990 to December
index which proxies for the strength of 1999, we find that the Democracy Portfolio
shareholder rights, and then study the outperformed the Dictatorship Portfolio by
empirical relationship between this index a statistically significant 8.5 percent per
and corporate performance. Our analysis year.These return differences induced large
should be thought of as a “longrun event changes in firm value over the sample
study”: we have democracies and dictator- period. By 1999 a one-point difference in
ships, the rules stayed mostly the same the index was negatively associated with
for a decade—how did each type do? Our an 11.4 percentage-point difference in
main results are to demonstrate that, in the Tobin’s Q. After partially controlling for
1990s, democracies earned significantly differences in market expectations by
higher returns, were valued more highly, using the book-to-market ratio, we also
and had better operating performance. Our find evidence that firms with weak share-
analysis is not a test of market efficiency. holder rights were less profitable and
Because theory provides no clear prediction, had lower sales growth than other firms in
there is no reason that investors in 1990 their industry.
should have foreseen the outcome of this The correlation of the Governance Index
novel experiment. Also, because this “exper- with returns, firm value, and operating
iment” did not use random assignment, we performance could be explained in several
cannot make strong claims about causality, ways. Section IV sets out three hypotheses
but we do explore the implications and to explain the results. Hypothesis I is that
assess the supportive evidence for several weak shareholder rights caused additional
causal hypotheses.3 agency costs. If the market underestimated
Our data are derived from publications of these additional costs, then a firm’s stock
the Investor Responsibility Research Center. returns and operating performance would
These publications provide 24 distinct have been worse than expected, and the
corporate-governance provisions for approx- firm’s value at the beginning of the period
imately 1500 firms since 1990.4 In Section would have been too high. Hypothesis II is
II we describe these provisions and data that managers in the 1980s predicted poor
sources in more detail. We divide the rules performance in the 1990s, but investors did
into five thematic groups and then construct not. In this case, the managers could have
CORPORATE GOVERNANCE AND EQUITY PRICES 525
put governance provisions in place to common stock (less than 10 percent of the
protect their jobs. While the provisions total).5 The IRRC universe covers most of
might have real protective power, they would the value-weighted market: even in 1990
not have caused the poor performance. the IRRC tracked more than 93 percent
Hypothesis III is that governance provi- of the total capitalization of the combined
sions did not cause poor performance (and New York Stock Exchange (NYSE),
need not have any protective power) but American Stock Exchange (AMEX), and
rather were correlated with other charac- NASDAQ markets.
teristics that were associated with abnormal The IRRC tracks 22 charter provisions,
returns in the 1990s. While we cannot bylaw provisions, and other firm-level rules
identify any instrument or natural experi- plus coverage under six state takeover laws;
ment to cleanly distinguish among these duplication between firm-level provisions
hypotheses, we do assess some supportive and state laws yields 24 unique provisions.
evidence for each one in Section V. For Table 20.1 lists all of these provisions, and
Hypothesis I we find some evidence of Appendix 1 discusses each one in detail. We
higher agency costs in a positive relation- divide them into five groups: tactics for
ship between the index and both capital delaying hostile bidders (Delay); voting
expenditures and acquisition activity. In rights (Voting); director/officer protection
support of Hypothesis III we find several (Protection); other takeover defenses
observable characteristics that can explain (Other); and state laws (State).
up to one-third of the performance The Delay group includes four provisions
differences. We find no evidence in support designed to slow down a hostile bidder. For
of Hypothesis II. Section VI concludes takeover battles that require a proxy fight
the paper. to either replace a board or dismantle a
takeover defense, these provisions are the
most crucial. Indeed, some legal scholars
II. DATA argue that the dynamics of modern
takeover battles have rendered all other
II.A. Corporate-governance defenses superfluous [Daines and Klausner
2001; Coates 2000]. The Voting group
provisions contains six provisions, all related to share-
Our main data source is the Investor holders’ rights in elections or charter/bylaw
Responsibility Research Center (IRRC), amendments.The Protection group contains
which publishes detailed listings of six provisions designed to insure officers
corporate-governance provisions for indi- and directors against job-related liability
vidual firms in Corporate Takeover Defenses or to compensate them following a termi-
[Rosenbaum 1990, 1993, 1995, 1998]. nation. The Other group includes the six
These data are derived from a variety of remaining firm-level provisions.
public sources including corporate bylaws These provisions tend to cluster within
and charters, proxy statements, annual firms. Out of (22 * 21)/2 231 total
reports, as well as 10-K and 10-Q pairwise correlations for the 22 firm-level
documents filed with the SEC. The IRRC’s provisions, 169 are positive, and 111 of
universe is drawn from the Standard & these positive correlations are significant.6
Poor’s (S&P) 500 as well as the annual In contrast, only 9 of the 62 negative
lists of the largest corporations in the correlations are significant. This clustering
publications of Fortune, Forbes, and suggests that firms may differ significantly
Businessweek. The IRRC’s sample in the balance of power between investors
expanded by several hundred firms in 1998 and management.
through additions of some smaller firms The IRRC firm-level data do not include
and firms with high institutional-ownership provisions that apply automatically under
levels. Our analysis uses all firms in the state law. Thus, we supplement these data
IRRC universe except those with dual-class with state-level data on takeover laws as
526 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 20.1 Governance provisions
Delay
Blank check 76.4 80.0 85.7 87.9
Classified board 59.0 60.4 61.7 59.4
Special meeting 24.5 29.9 31.9 34.5
Written consent 24.4 29.2 32.0 33.1
Protection
Compensation plans 44.7 65.8 72.5 62.4
Contracts 16.4 15.2 12.7 11.7
Golden parachutes 53.1 55.5 55.1 56.6
Indemnification 40.9 39.6 38.7 24.4
Liability 72.3 69.1 65.6 46.8
Severance 13.4 5.5 10.3 11.7
Voting
Bylaws 14.4 16.1 16.0 18.1
Charter 3.2 3.4 3.1 3.0
Cumulative voting 18.5 16.5 14.9 12.2
Secret ballot 2.9 9.5 12.2 9.4
Supermajority 38.8 39.6 38.5 34.1
Unequal voting 2.4 2.0 1.9 1.9
Other
Antigreenmail 6.1 6.9 6.4 5.6
Directors’ duties 6.5 7.4 7.2 6.7
Fair price 33.5 35.2 33.6 27.8
Pension parachutes 3.9 5.2 3.9 2.2
Poison pill 53.9 57.4 56.6 55.3
Silver parachutes 4.1 4.8 3.5 2.3
State
Antigreenmail law 17.2 17.6 17.0 14.1
Business combination law 84.3 88.5 88.9 89.9
Cash-out law 4.2 3.9 3.9 3.5
Directors’ duties law 5.2 5.0 5.0 4.4
Fair price law 35.7 36.9 35.9 31.6
Control share acquisition law 29.6 29.9 29.4 26.4
Number of firms 1357 1343 1373 1708
This table presents the percentage of firms with each provision between 1990 and 1998. The data are drawn
from the IRRC Corporate Takeover Defenses publications [Rosenbaum 1990, 1993, 1995, 1998] and are
supplemented by data on state takeover legislation coded from Pinnell [2000]. See Appendix 1 for detailed
information on each of these provisions. The sample consists of all firms in the IRRC research universe except
those with dual class stock.
given by Pinnell [2000], another IRRC provisions given in other groups. We discuss
publication. From this publication we code these analogues in subsection II.B.
the presence of six types of so-called The IRRC data set is not an exhaustive
“second-generation” state takeover laws listing of all provisions. Although firms can
and place them in the State group.7 Few review their listing and point out mistakes
states have more than three of these laws, before publication, the IRRC does not
and only Pennsylvania has all six.8 Some update every company in each new edition
of these laws are analogues of firm-level of the book, so some changes may be
CORPORATE GOVERNANCE AND EQUITY PRICES 527
missed. Also the charter and bylaws are not require any judgments about the efficacy
available for all companies and thus the or wealth effects of any of these provisions;
IRRC must infer some provisions from we only consider the impact on the balance
proxy statements and other filings. Overall, of power.
the IRRC intends its listings as a starting For example, consider Classified Boards,
point for institutional investors to review a provision that staggers the terms and
governance provisions. Thus, these listings elections of directors and hence can be
are a noisy measure of a firm’s governance used to slow down a hostile takeover. If
provisions, but there is no reason to suspect management uses this power judiciously, it
any systematic bias. Also, all of our analysis could possibly lead to an increase in overall
uses data available at time t to forecast shareholder wealth; if management uses
performance at time t 1 and beyond, so this power to maintain private benefits of
there is no possibility of look-ahead bias control, then this provision would decrease
induced by our statistical procedures. shareholder wealth. In either case, it is
To build the data set, we coded the data clear that Classified Boards increase the
from the individual firm profiles in the power of managers and weaken the control
IRRC books. For each firm we recorded the rights of large shareholders, which is all
identifying information (ticker symbol, that matters for constructing the index.
state of incorporation) and the presence of Most of the provisions can be viewed in
each provision. Although many of the a similar way. Almost every provision gives
provisions can be made stronger or weaker management a tool to resist different types
(e.g., supermajority thresholds can vary of shareholder activism, such as calling
between 51 and 100 percent), we made no special meetings, changing the firm’s
strength distinctions and coded all provi- charter or bylaws, suing the directors, or
sions as simply “present” or “not present.” just replacing them all at once. There are
This methodology sacrifices precision for two exceptions: Secret Ballots and
the simplicity necessary to build an index. Cumulative Voting. A Secret Ballot, also
For most of the analysis of this paper, we called “confidential voting” by some firms,
match the IRRC data to the Center for designates a third party to count proxy
Research in Security Prices (CRSP) and, votes and prevents management from
where necessary, to Standard and Poor’s observing how specific shareholders vote.
Compustat database. CSRP matching was Cumulative Voting allows shareholders to
done by ticker symbol and was supple- concentrate their directors’ votes so that a
mented by handchecking names, exchanges, large minority holder can ensure some
and states of incorporation. These proce- board representation. (See Appendix 1 for
dures enable us to match 100 percent of fuller descriptions.) These two provisions
the IRRC sample to CRSP, with about are usually proposed by shareholders and
90 percent of these matches having opposed by management.9 In contrast, none
complete annual data in Compustat. of the other provisions enjoy consistent
shareholder support or management oppo-
II.B. The Governance index sition; in fact, many of these provisions
receive significant numbers of shareholder
The index construction is straightforward: proposals for their repeal [Ishii 2000].
for every firm we add one point for every Also, both Cumulative Voting and Secret
provision that restricts shareholder rights Ballots tend to be negatively correlated
(increases managerial power). This power with the presence of other firm-level provi-
distinction is straightforward in most sions (19 negative out of 21 for Cumulative
cases, as is discussed below. While this Voting; 11 out of 21 for Secret Ballot).
simple index does not accurately reflect Thus, we consider the presence of Secret
the relative impacts of different provisions, Ballots and Cumulative Voting to be
it has the advantage of being transparent increases in shareholder rights. For each
and easily reproducible. The index does not one we add one point to the Governance
528 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Index when firms do not have it. For all subindex and its Other subindex, but only
other provisions we add one point when one point (not two) would be added to its
firms do have it.10 overall G index. Thus, G has a possible
Thus, the Governance Index (“G”) is just range from 1 to 24 and is not just the sum
the sum of one point for the existence of the five subindices.
(or absence) of each provision. We also Table 20.2 gives summary statistics for
construct subindices for each of the five G and the subindices in 1990, 1993, 1995,
categories: Delay, Protection, Voting, Other, and 1998. Table 20.2 also shows the fre-
and State. Recall that there are 28 total quency of G by year, broken up into groups
provisions listed in the five categories, beginning with G 5, then each value of G
of which 24 are unique. For the state from G 6 through G 13, and finishing
laws with a firm-level analogue, we add one with G 14. These ten “deciles” are simi-
point to the index if the firm is covered lar but not identical in size, with relative
under the firm-level provision, the state sizes that are fairly stable from 1990 to
law, or both.11 For example, a firm that 1995. In the remainder of the paper we
has an Antigreenmail provision and is also pay special attention to the two extreme
covered by the Antigreenmail state law portfolios: the “Dictatorship Portfolio” of
would get one point added to both its State the firms with the weakest shareholder rights
Governance index
Minimum 2 2 2 2
Mean 9.0 9.3 9.4 8.9
Median 9 9 9 9
Mode 10 9 9 10
Maximum 17 17 17 18
Standard deviation 2.9 2.8 2.8 2.8
Number of firms
G 5 (Democracy Portfolio) 158 139 120 215
G6 119 88 108 169
G7 158 140 127 186
G8 165 139 152 201
G9 160 183 183 197
G 10 175 170 178 221
G 11 149 168 166 194
G 12 104 123 142 136
G 13 84 100 110 106
G 14 (Dictatorship portfolio) 85 93 87 83
Total 1357 1343 1373 1708
Subindex means
Delay 1.8 2.0 2.1 2.1
Protection 2.4 2.5 2.5 2.1
Voting 2.2 2.1 2.1 2.2
Other 1.1 1.2 1.1 1.0
State 1.8 1.8 1.8 1.7
This table provides summary statistics on the distribution of G, the Governance Index, and the subindices (Delay,
Protection, Voting, Other, and State) over time. G and the subindices are calculated from the provisions listed
in Table 20.1 as described in Section II. Appendix 1 gives detailed information on each provision. We divide the
sample into ten portfolios based on the level of G and list the number of firms in each portfolio.The Democracy
Portfolio is composed of all firms where G 5, and the Dictatorship Portfolio contains all firms where G 14.
CORPORATE GOVERNANCE AND EQUITY PRICES 529
(G 14), and the “Democracy Portfolio” and 1998. Of the ten largest firms in the
of the firms with the strongest shareholder Democracy Portfolio in 1990, six of them
rights (G 5).These portfolios are updated are still in the Democracy Portfolio in
at the same frequency as G. 1998, three have dropped out of the port-
Most of the changes in the distribution folio and have G 6, and one (Berkshire
of G come from changes in the sample due Hathaway) disappeared from the sample.13
to mergers, bankruptcies, and additions of The Dictatorship Portfolio has a bit more
new firms by the IRRC. In 1998 the sample activity, with only two of the top ten
size increased by about 25 percent, and firms remaining in the portfolio, four firms
these new firms tilted toward lower values dropping out with G 13, and three firms
of G. At the firm level, G is relatively stable. leaving the sample though mergers or the
For individual firms the mean (absolute) addition of another class of stock.14 Thus,
change in G between publication dates 40 percent (eight out of 20) of the largest
(1990, 1993, 1995, 1998) is 0.60, and firms in the extreme portfolios in 1990
the median (absolute) change between were also in these portfolios in 1998. This
publication dates is zero.12 is roughly comparable to the full set of
Table 20.3 shows the correlations firms: among all firms in the Democracy
between pairs of subindices. The Delay, and Dictatorship Portfolios in 1990,
Protection, Voting, and Other subindices 31 percent were still in the same portfolios
all have positive and significant pairwise in 1998.
correlations with each other. State, how- There is no obvious industry concentration
ever, has negative correlations with Delay, among these top firms; the whole portfolios
Protection, and Voting. It could be that are similarly dispersed. Classifying firms
firms view some of the state laws as sub- into 48 industries as in Fama and French
stitutes for the firm-level provisions, but [1997], the portfolios appear to be broadly
then it would be surprising that Other, similar to each other in all years, with a
which contains three provisions that are mix of old-economy and new-economy
direct substitutes for state laws, is the only industries.15 Each portfolio has an impor-
subindex that is positively correlated with tant technology component. “Computers”
State. Overall, it appears that coverage is the largest industry by market value in
under state laws is not highly correlated the Democracy Portfolio in 1990, with
with the adoption of firm-level provisions. 22.4 percent of the portfolio, falling to
This fact has implications for the analysis third place with 12.3 percent of the value
of causality, as is discussed in Section IV. in 1998. “Communications” does not
Table 20.4 lists the ten largest firms (by make the top five in market value for the
market capitalization) in the Democracy Dictatorship Portfolio in 1990, but rises
and Dictatorship Portfolios in 1990 and to first place with 25.3 percent of the
gives the value of G for these firms in 1990 portfolio in 1998.
This table presents pairwise correlations between the subindices, Delay, Protection, Voting, Other, and State
in 1990. The calculation of the subindices is described in Section II. The elements of each subindex are given in
Table 20.1 and are described in detail in Appendix 1. Significance at the 5 percent and 1 percent levels is
indicated by * and **, respectively.
530 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 20.4 The Largest Firms in the Extreme Portfolios
This table presents the firms with the largest market capitalizations at the end of 1990 of all companies within
the Democracy Portfolio (G 5) and the Dictatorship Portfolio (G 14). The calculation of G is described in
Section II. The companies are listed in descending order of market capitalization.
This table gives descriptive statistics for the relationship of G with several financial and accounting measures in
September 1990.The first column gives the correlations for each of these variables with the Governance Index, G.
The second and third columns give means for these same variables within the Democracy Portfolio (G 5) and
the Dictatorship Portfolio (G 14) in 1990. The final column gives the difference of the two means with its
standard error in parentheses. The calculation of G is described in Section II, and definitions of each variable
are given in Appendix 2. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively.
have relatively high levels of all of these multivariate analysis suggesting no robust
characteristics. In addition, the correla- relationship between G and changes in
tion of G with five-year sales growth is institutional ownership.
negative and significant, suggesting that
high-G firms had relatively lower sales III.B. Governance and returns
growth over the second half of the 1980s,
the period when many of the provisions If corporate governance matters for firm
were first adopted. performance and this relationship is fully
Correlations at other times in the sample incorporated by the market, then a stock
period (not shown in the table) are similar. price should quickly adjust to any relevant
Overall, it appears that firms with weaker change in the firm’s governance. This is the
shareholder rights tend to be large S&P logic behind the use of event studies to
firms with relatively high share prices, analyze the impact of takeover defenses. If
institutional ownership and trading volume, such a reaction occurs, then expected
relatively poor sales growth, and poor returns on the stock would be unaffected
stock-market performance.The 1990s were beyond the event window. However, if
a time of rising activism by institutional governance matters but is not incorporated
investors and more attention to governance immediately into stock prices, then real-
provisions; thus, we might expect to see ized returns on the stock would differ
some reduction in the institutional owner- systematically from equivalent securities.
ship of high-G firms. In untabulated tests, In this section we examine the relationship
we find no evidence of such a reduction, between G and subsequent returns. An
with both pairwise correlations and investment of $1 in the (value-weighted)
532 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
We estimate four-factor regressions (equation (1) from the text) of value-weighted monthly returns for portfolios
of firms sorted by G.The calculation of G is described in Section II.The first row contains the results when we use
the portfolio that buys the Democracy Portfolio (G 5) and sells short the Dictatorship Portfolio (G 14). The
portfolios are reset in September 1990, July 1993, July 1995, and February 1998, which are the months after new
data on G became available. The explanatory variables are RMRF, SMB, HML, and Momentum. These variables
are the returns to zero-investment portfolios designed to capture market, size, book-to-market, and momentum
effects, respectively. (Consult Fama and French [1993] and Carhart [1997] on the construction of these factors.)
The sample period is from September 1990 through December 1999. Standard errors are reported in parentheses
and significance at the 5 percent and 1 percent levels is indicated by * and **, respectively.
Table 20.7 reports several variations The first row of Table 20.7 replicates the
of the abnormal-return results. In each baseline portfolio construction used above.
variation we estimate the performance- The remaining rows of the table summarize
attribution regression in equation (1) on tests using industry-adjusted returns (row 2),
the return difference between the Democracy two alternative constructions of the extreme
and Dictatorship Portfolios, while changing portfolios (rows 3 and 4), fixed portfolios
some aspect of the portfolio construction built with 1990 levels of G (row 5), a sub-
or return calculation. We perform all of sample that includes only Delaware firms
these tests using both value-weighted (VW) (row 6), and subsamples split between the
and equal-weighted (EW) portfolios. first half and the second half of the sample
These tests allow us to estimate the fraction period (rows 7 and 8). Details of each of
of the benchmark abnormal returns that these constructions are given in the table
can be attributed to industry composition, note. The main themes of these results are,
choice of cutoffs for the extreme portfolios, first, that the VW returns (Democracy
new provisions during the decade, legal minus Dictatorship) are economically large
variation across states, and different time in all cases and, second, the EW abnormal
periods. returns are usually about two-thirds of the
534 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 20.7 Performance-Attribution Regressions under Alternative Portfolio Constructions
, Value-weighted , Equal-weighted
This table presents the alphas from four-factor regressions for variations on the Democracy (G 5) minus
Dictatorship (G 14) Portfolio. The calculation of G is described in Section II. The portfolios are reset in
September 1990, July 1993, July 1995, and February 1998, which are the months after new data on G became
available. The sample period is September 1990 to December 1999. The first row uses the unadjusted difference
between the monthly returns to the Democracy and Dictatorship Portfolios. The second row contains the results
using industry-adjusted returns, with industry adjustments done relative to the 48 industries of Fama and French
[1997]. The third and fourth rows use alternative definitions of the Democracy and Dictatorship Portfolios. In
the third row, firms are sorted on G and the two portfolios contain the smallest set of firms with extreme values
of G such that each has at least 10 percent of the sample. This implies cutoff values of G for the Democracy
Portfolio of 5, 5, 6, and 5 for September 1990, July 1993, July 1995, and February 1998, respectively. The
cutoffs for the Dictatorship Portfolio are always 13. In the fourth row, the two portfolios contain the largest set
of firms such that each has no more than 10 percent of the sample. The cutoff values of G for the Democracy
Portfolio are 4, 4, 5, and 4 for September 1990, July 1993, July 1995, and February 1998, respectively, and
they are always 14 for the Dictatorship Portfolio. In the fifth row, portfolio returns are calculated maintaining
the 1990 portfolios for the entire sample period. As long as they are listed in CRSP, we neither delete nor add
firms to these portfolios regardless of subsequent changes in G or changes in the IRRC sample in later editions.
The sixth row shows the results of restricting the sample to firms incorporated in Delaware. In the seventh and
eighth rows, the sample period is divided in half at April 30, 1995, and separate regressions are estimated for
the first half and second half of the period (56 months each).The explanatory variables are RMRF, SMB, HML,
Momentum, and a constant. These variables are the returns to zero-investment portfolios designed to capture
market, size, book-to-market, and momentum effects, respectively. (Consult Fama and French [1993] and
Carhart [1997] on the construction of these factors.) All coefficients except for the alpha are omitted in this
table. Standard errors are reported in parentheses, and significance at the 5 percent and 1 percent levels is
indicated by * and **, respectively.
in Delaware have higher valuations than We rejected this alternative mainly because
other U. S. firms. In this section we study there are few changes over time in the
whether variation in firm-specific gover- Governance Index, and the inclusion of
nance is associated with differences in fixed effects would force identification of
firm value. More importantly, we analyze the G coefficient from only these changes.
whether there was a change in the gover- In effect, our chosen method imposes a
nance/value relationship during the 1990s. structure on the fixed effects: they must be
Since there is evidence of differential stock a linear function of G or its components.
returns as a function of G, we would expect Table 20.8 summarizes the results. The
to find relative “mispricing” between 1990 first column gives the results with G as the
and 1999 as a function of G. key regressor. Each row gives the coeffi-
Our valuation measure is Tobin’s Q, cients and standard errors for a different
which has been used for this purpose in year of the sample; the last row gives the
corporate-governance studies since the average coefficient and time-series standard
work of Demsetz and Lehn [1985] and error of these coefficients. The coefficients
Morck, Shleifer, and Vishny [1988]. We on G are negative in every year and signifi-
follow Kaplan and Zingales’ [1997] cantly negative in nine of the ten years.The
method for the computation of Q (details largest absolute value point estimate
are listed in Appendix 2) and also compute occurs in 1999, and the second largest is
the median Q in each year in each of the in 1998.The point estimate in 1999 is eco-
48 industries classified by Fama and nomically large; a one-point increase in G,
French [1997]. We then regress equivalent to adding a single governance
provision, is associated with an 11.4 percent-
Q’it atbtXitctWiteit, (2) age point lower value for Q. If we assume
that the point estimates in 1990 and 1999
where Q’it is industry-adjusted Q (firm Q are independent, then the difference
minus industry-median Q), Xit is a vector of between these two estimates (11.4
governance variables (G, its components, or 2.2 9.2) is statistically significant.
inclusion in one of the extreme portfolios) In the second column of Table 20.8, we
and Wit is a vector of firm characteristics. restrict the sample to include only firms in
As elements of W, we follow Shin and Stulz the Democracy and Dictatorship
[2000] and include the log of the book Portfolios. We then estimate (2) using a
value of assets and the log of firm age as dummy variable for the Democracy
of December of year t.18 Daines [2001] Portfolio. The results are consistent with
found that Q is different for Delaware and the previous regressions on G. The point
non-Delaware firms, so we also include a estimate for 1999 is the largest in the
Delaware dummy in W. Morck and Yang decade, implying that firms in the Democracy
[2001] show that S&P 500 inclusion has Portfolio have a Q that is 56 percentage
a positive impact on Q, and that this points higher, other things being equal, than
impact increased during the 1990s; thus, do firms in the Dictatorship Portfolio. This
we also include a dummy variable for S&P compares with an estimated difference of
500 inclusion in W. 19 percentage points in 1990. While the
Using a variant of the methods of Fama difference in coefficients between 1990
and MacBeth [1973], we estimate annual and 1999 is not statistically significant, it
cross sections of (2) with statistical is similar to the total EW difference in
significance assessed within each year (by abnormal returns estimated in Table 20.7.19
cross-sectional standard errors) and across There is no real pattern for the rest of the
all years (with the time-series standard decade, however, and large standard errors
error of the mean coefficient). This method toward the end of the sample period
of assessing statistical significance deserves prevent any strong inference across years.
some explanation. In particular, one logical The final columns of Table 20.8 give
alternative would be a pooled setup with firm results for a single regression using the
fixed effects and time-varying coefficients. five governance subindices: Delay, Voting,
536 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 20.8 Q Regressions
(2)
(1) Democracy (3) (4) (5) (6) (7)
G Portfolio Delay Protection Voting Other State
The first column of this table presents the coefficients on G, the Governance Index, from regressions of
industryadjusted Tobin’s Q on G and control variables. The second column restricts the sample to firms in the
Democracy (G 5) and Dictatorship (G 14) Portfolios and includes as regressors a dummy variable for the
Democracy Portfolio and the controls.The third through seventh columns show the coefficients on each subindex
from regressions where the explanatory variables are the subindices Delay, Protection, Voting, Other, and State,
and the controls. We include as controls a dummy variable for incorporation in Delaware, the log of assets in the
current fiscal year, the log of firm age measured in months as of December of each year, and a dummy variable
for inclusion in the S&P 500 as of the end of the previous year.The coefficients on the controls and the constant
are omitted from the table. The calculation of G and the subindices is described in Section II. Q is the ratio of
the market value of assets to the book value of assets: the market value is calculated as the sum of the book
value of assets and the market value of common stock less the book value of common stock and deferred taxes.
The market value of equity is measured at the end of the current calendar year, and the accounting variables are
measured in the current fiscal year. Industry adjustments are made by subtracting the industry median, where
medians are calculated by matching the four-digit SIC codes from December of each year to the 48 industries
designated by Fama and French [1997]. The coefficients and standard errors from each annual cross-sectional
regression are reported in each row, and the time-series averages and time-series standard errors are given in the
last row. * and ** indicate significance at the 5 percent and 1 percent levels, respectively.
Protection, Other, and State. The table it difficult to draw strong conclusions.
shows that all subindices except Voting For example, even in 1999 we cannot
have average coefficients that are negative reject the null hypothesis that the coeffi-
and significant (assuming independence cient on Voting is equal to the coefficient
across years). Over the full sample period, on Delay.
Delay and Protection have the most Overall, the results for returns and prices
consistent impact, while the largest tell a consistent story. Firms with the
absolute coefficients are for Voting at the weakest shareholder rights (high values of G)
end of the sample period.The subindices are significantly underperformed firms with the
highly collinear, however, and the resulting strongest shareholder rights (low values of
large standard errors and covariances make G) during the 1990s. Over the course of the
CORPORATE GOVERNANCE AND EQUITY PRICES 537
1990s, these differences have been at least operational measures are the net profit
partially reflected in prices. While high-G margin (income divided by sales), the
firms already sold at a significant discount return on equity (income divided by book
in 1990, this discount became much larger equity), and one-year sales growth. All of
by 1999. these measures are industry-adjusted by
subtracting the median for this measure in
the corresponding Fama-French [1997]
III.D. Governance and operating
industry. This adjustment uses all available
performance Compustat firms.To reduce the influence of
Table 20.9 shows the results of annual large outliers—a common occurrence for
regressions for three operational measures all of these measures—we estimate median
on G (or a Democracy dummy). The three (least-absolute-deviation) regressions in
The first, third, and fifth columns of this table give the results of annual median (least absolute deviation)
regressions for net profit margin, return on equity, and sales growth on the Governance Index, G, measured in
the previous year, and the book-to-market ratio, BM. The second, fourth, and sixth columns restrict the sample
to firms in the Democracy (G 5) and Dictatorship (G 14) portfolios and include as regressors a dummy
variable for the Democracy Portfolio and BM. The coefficients on BM and the constant are omitted from the
table. The calculation of G is described in Section II. Net profit margin is the ratio of income before extraordi-
nary items available for common equity to sales; return on equity is the ratio of income before extraordinary
items available for common equity to the sum of the book value of common equity and deferred taxes; BM is
the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous fiscal
year to size at the close of the previous calendar year. Each dependent variable is net of the industry median,
which is calculated by matching the four-digit SIC codes of all firms in the CRSP-Compustat merged database
in December of each year to the 48 industries designated by Fama and French [1997]. The coefficients and
standard errors from each annual cross-sectional regression are reported in each row, and the time-series
averages and time-series standard errors are given in the last row. Significance at the 5 percent and 1 percent
levels is indicated by * and **, respectively. All coefficients and standard errors are multiplied by 1000.
538 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
each case. While our sample does not HYPOTHESIS III. Governance provisions do
include a natural experiment to identify G not cause higher agency costs, but their
as the cause of operational differences, presence is correlated with other
we attempt to control for “expected” cross- characteristics that earned abnormal
sectional differences by using the log returns in the 1990s.
book-to-market ratio (BM) as an additional
explanatory variable. Most explanations of the Section III
The odd-numbered columns give the results can be fit within these three
results when G is the key regressor. We find hypotheses. Under Hypothesis I, a reduc-
that the average coefficient on G is negative tion in shareholder rights causes an unex-
and significant for both the net-profit-margin pectedly large increase in agency costs
and sales-growth regressions, and is negative through some combination of inefficient
but not significant for the return-on-equity investment, reduced operational efficiency,
regressions. The even-numbered columns or self-dealing. If shareholders find it diffi-
give the results for the subsample of firms cult or costly to replace managers, then
from the extreme deciles, with a dummy managers may be more willing and able to
variable for the Democracy Portfolio as the extract private benefits.This is the standard
key regressor. For all three operating meas- justification for takeover threats as the
ures, the average coefficient on this dummy strongest form of managerial discipline
variable was positive but insignificant. [Jensen 1986]. For Hypothesis I to be cor-
Thus, these results are consistent with the rect, these additional agency costs must
evidence for the full sample but not signifi- have been underestimated in 1990.
cant on their own. In untabulated results, Under Hypothesis II, governance does
we also regressed these same measures not affect performance, but there must be a
on the five subindices. The results show no perception that governance provisions are
clear pattern of differential influence for any protective for management. In this case, the
particular subindex, with most coefficients stock in these companies would have been
having the same sign as G. Overall, we relatively overvalued in 1990, even though
find some significant evidence that more objective measures (e.g., Q regressions)
democratic firms have better operating per- would suggest that it was undervalued
formance and no evidence that they do not. relative to observable characteristics. When
the poor operating performance occurs, the
market is surprised, but the managers are
IV. GOVERNANCE: THREE not. The protective provisions then supply a
HYPOTHESES shield, real or imagined, for managerial
jobs and compensation.
Section III established an empirical Under Hypothesis III, all of the results in
relationship of G with returns, firm value, the previous section would be driven by
and operating performance. Since firms did omitted-variable bias. Since governance
not adopt governance provisions randomly, provisions were certainly not adopted
this evidence does not itself imply a causal randomly, it is plausible that differences in
role by governance provisions. Indeed, industry, S&P 500 inclusion, institutional
there are several plausible explanations for ownership, or other firm characteristics
our results: could be correlated both with G and with
abnormal returns. Under this hypothesis,
HYPOTHESIS I. Governance provisions cause governance provisions could be completely
higher agency costs. These higher costs innocuous, with no influence either on
were underestimated by investors in 1990. managerial power or on agency costs.
HYPOTHESIS II. Governance provisions do Ideally, we would distinguish among these
not cause higher agency costs, but rather three hypotheses by using random variation
were put in place by 1980s managers in some characteristic that was causal for
who forecasted poor performance for G. Unfortunately, we have not been able to
their firms in the 1990s. identify such an instrument. One candidate
CORPORATE GOVERNANCE AND EQUITY PRICES 539
would be the subset of state laws, with the Section IV. Subsection V.A covers
State subindex as a proxy. Though in some Hypothesis I, subsection V.B covers
states these laws were passed at the urging Hypothesis II, and subsection V.C cov-
of large corporations, it seems reasonable ers Hypothesis III. Subsection V.D
to assume that their passage was exoge- summarizes and discusses the evidence.
nous to most firms. But the State subindex
has three flaws as an instrument. First,
firms can choose to reincorporate into V.A. Evidence on Hypothesis I
different states; enough firms have done Increased agency costs at high-G firms can
so that exposure to state laws is not truly directly affect firm performance in several
exogenous [Subramanian 2001]. Second, ways. In the specific case of state takeover
many firms have opted out of the protec- laws, where causality is easier to establish,
tions of some of the most stringent of these researchers have found evidence of increased
laws, so that a firm’s state of incorporation agency costs through a variety of mecha-
is only a noisy measure for its actual legal nisms. Borokhovich, Brunarski, and Parrino
exposure. Third, as shown in Table 20.3, [1997] show that compensation rises for
the State subindex is not positively or CEOs of firms adopting takeover defenses.
consistently correlated with the other Bertrand and Mullainathan [1999a,
components of G. Other potential instru- 1999b, 2000] find a similar result for
ments have different problems. For example, CEOs and other employees in firms newly
if takeover protections were adopted during covered by state takeover laws. They also
industry-specific takeover waves, then we find that these laws cause a decrease in
might be able to use industry as an instru- plant-level efficiency, measured either
ment for G. Unfortunately, this would by total factor productivity or return on
render it impossible to distinguish between capital. Garvey and Hanka [1999] show
G or industry as the cause of poor returns that state takeover laws led to changes in
in the 1990s. leverage consistent with increased corporate
In Section V our tests consist of a search slack. These studies provide the cleanest
for evidence supportive of each hypothesis, evidence in support of Hypothesis I, but,
while acknowledging the impossibility of a of course, do not make use of the full
perfect test to distinguish among them. variation embodied in the G index. We
First, if Hypothesis I is correct, then we supplement these findings by examining the
should observe some “unexpected” differ- empirical relationship of G with two other
ences in agency costs across firms. We possible sources of agency costs: capital
discuss several previous studies on this expenditure and acquisition behavior.
topic and look for such differences in our A substantial literature, dating back at
sample by analyzing capital expenditure least to Baumol [1959], Marris [1964],
and acquisition behavior. Second, for and Williamson [1964], holds that managers
Hypothesis II we analyze insider-trading may undertake inefficient projects in order
activity as a function of G. If governance to extract private benefits. This problem is
provisions were put in place by prescient particularly severe when managers are
managers, these same managers might be entrenched and can resist hostile takeovers
net sellers of the stock in their firms. [Jensen and Ruback 1983; Shleifer and
Finally, for Hypothesis III we test whether a Vishny 1989]. Under this view, if capital
large set of observable firm characteristics expenditure increases following the adoption
can explain the empirical relationship of new takeover defenses, this increase
between returns and G. would be a net negative for firm value.20
To examine the empirical relationship
between capital expenditure and governance,
V. GOVERNANCE: TESTS we estimate annual median regressions
for capital expenditure (CAPEX), scaled
In this section we examine the evidence by either sales or assets, and net of the
for each of the hypotheses described in industry median.To control for the different
540 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Democracy Democracy
G Portfolio G Portfolio
The first and third columns of this table present the results of annual median (least absolute deviation)
regressions of CAPEX/Assets and CAPEX/Sales on the Governance Index, G, measured in the previous year, and
BM. The second and fourth columns restrict the sample to firms in the Democracy (G 5) and Dictatorship
(G 14) portfolios and include as regressors a dummy variable for the Democracy portfolio and BM.The coef-
ficients of BM and the constant are omitted from the table. The calculation of G is described in Section II.
CAPEX is capital expenditures, and BM is the log of the ratio of book value (the sum of book common equity
and deferred taxes) in the previous fiscal year to size at the close of the previous calendar year. Both dependent
variables are net of the industry median, which is calculated by matching the four-digit SIC codes of all firms
in the CRSP-Compustat merged database in December of each year to the 48 industries designated by Fama
and French [1997]. The coefficients and standard errors from each annual cross-sectional regression are
reported in each row, and the time-series averages and time-series standard errors are given in the last row.
Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and
standard errors are multiplied by 1000.
CORPORATE GOVERNANCE AND EQUITY PRICES 541
Democracy Democracy
G Portfolio G Portfolio
The first column of this table presents annual Tobit regressions of the Acquisition ratio on the Governance Index,
G, measured in the previous year, SIZE, BM, and industry dummy variables. The third column presents annual
Poisson regressions of Acquisition count on the same explanatory variables. In the second and fourth columns,
we restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) Portfolios, and we include
as a regressor a dummy variable that equals 1 when the firm is in the Democracy Portfolio and 0 otherwise.The
coefficients on SIZE, BM, and the industry dummy variables are omitted from the table.The calculation of G is
described in Section II. Acquisition ratio is defined as the sum of the value of all corporate acquisitions during
a calendar year scaled by the average of market value at the beginning and end of the year. Acquisition count is
defined as the number of acquisitions during a calendar year. The data on acquisitions are from the SDC data-
base. SIZE is the log of market capitalization at the end of the previous calendar year in millions of dollars, and
BM is the log of the ratio of book value (the sum of book common equity and deferred taxes) in the previous
fiscal year to size at the close of the previous calendar year. Industry dummy variables are created by matching
the four-digit SIC codes of all firms in the CRSP-Compustat merged database in December of each year to the
48 industries designated by Fama and French [1997]. The coefficients and standard errors from each annual
cross-sectional regression are reported in each row, and the time-series averages and time-series standard errors
are given in the last row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively.
All coefficients and standard errors are multiplied by 100.
Thus, we also estimate ordered logit a Democracy dummy as the key regressor.
regressions on the same OLS regressors, Coefficients on all control variables are
in which the dependent variable is equal omitted from the table. We find no signifi-
to one if Net Purchases is positive, zero cant relationships between governance and
if Net Purchases is zero, and negative one if insider trading. Two of four sets of regres-
Net Purchases is negative. sions have positive average coefficients,
Table 20.12 summarizes the results of two have negative average coefficients, and
these regressions. Columns (1) and (3) give none of these average coefficients are
results for the full sample, with G as the key significant. In untabulated results we also
regressor; columns (2) and (4) give results estimated median regressions, replicated all
for the sample restricted to firms in the of the above results using all transactions
Democracy and Dictatorship Portfolios with (the main difference is the inclusion of
Democracy Democracy
G Portfolio G Portfolio
The first and third columns of this table present annual OLS and ordered logit regressions of Net insider
purchases on G measured in the previous year, SIZE, BM, and a constant. In the second and fourth columns, we
restrict the sample to firms in the Democracy (G 5) and Dictatorship (G 14) Portfolios and we include as
a regressor a dummy variable that equals 1 when the firm is in the Democracy Portfolio and 0 otherwise. The
coefficients on SIZE, BM, and the constant are omitted from the table. The calculation of G is described in
Section II. Net insider purchases is the sum of split-adjusted open market purchases less split-adjusted open
market sales during a year scaled by shares outstanding at the end of the previous calendar year. The ordered
logit regressions use a dependent variable that equals 1 if Net insider purchases is positive, 0 if it is zero, and 1
if it is negative. The data on insider sales are from the Thomson database. SIZE is the log of market capital-
ization in millions of dollars measured at the end of the previous calendar year, and BM is the log of the ratio
of book value (the sum of book common equity and deferred taxes) in the previous fiscal year to size at the close
of the previous calendar year. The coefficients and standard errors from each annual cross-sectional regression
are reported in each row, and the time-series averages and time-series standard errors are given in the last
row. Significance at the 5 percent and 1 percent levels is indicated by * and **, respectively. All coefficients and
standard errors are multiplied by 1000.
544 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
option-exercise transactions), and estimated and industry-adjusted, for the full sample
long-horizon regressions using all years of of firms in each month with G as the key
data for each firm. In none of these cases independent variable. In both regressions
did we find a robust relationship between the average coefficient on G is negative but
governance and insider trading. Overall, we not significant. The point estimates are not
find no support for Hypothesis II in the small. For example, the point estimate for
insider-trading data. the coefficient on G in column (3) implies a
lower return of approximately four bp per
month ( 48 bp per year) for each
V.C. Evidence on Hypothesis III additional point of G, but it would require
estimates nearly twice as large before
What other factors might be driving the statistical significance would be reached.
return difference between the Democracy The next two columns give the results
and Dictatorship Portfolios? We saw in when the sample is restricted to stocks
Table II that G is correlated with several in either the Democracy (G 5) or
firm characteristics, including S&P 500 Dictatorship (G 14) Portfolios. In the
membership, institutional ownership, trading first column the dependent variable is
volume, and past sales growth. If returns the raw monthly return for each stock. In
to stocks with these characteristics differed the second column the dependent variable
in the 1990s in a way not captured by the is the industry-adjusted return for each
model in equation (1), then a type of omitted stock, where industry adjustments are
variable bias may drive the abnormal-return relative to the Fama and French [1997]
results. In this section we explore this pos- 48 industries. The key independent variable
sibility using a cross-sectional regression in these regressions is the Democracy
approach. In addition to providing evidence dummy, set equal to one if the stock is in
on Hypothesis III, this method also supple- the Democracy Portfolio and zero if the
ments the analysis of subsection III.B by stock is in the Dictatorship Portfolio. For
allowing a separate regressor for each both the raw and industry-adjusted returns,
component of G. the coefficient on this dummy variable is
For each month in the sample period, positive and significant at the 1 percent
September 1990 to December 1999, we level. The average point estimate can be
estimate interpreted as a monthly abnormal return.
These point estimates, 76 bp per month
rit at btXit ctZit eit (3) raw and 63 bp per month industry-
adjusted, are similar to those found in the
where, for firm i in month t, rit are the factor models, and provide a further
returns (either raw or industry-adjusted), robustness check to the benchmark result.
Xit is a vector of governance variables Here, industry adjustments explain about
(either G, its components, or inclusion in one-sixth of the raw result. In the factor-
one of the extreme portfolios), and Zit is model results of Table 20.7, the industry
a vector of firm characteristics. As adjustment explained about one-third of
elements of Z, we include the full set the raw result.
of regressors used by Brennan, Chordia, Columns (5) and (6) of Table 20.13 give
and Subrahmanyam [1998], plus five-year the results for the full sample of firms
sales growth, S&P 500 inclusion, and insti- when the five subindices are used as the
tutional ownership.23 Variable definitions components of X. In principle, these
are given in Appendix 2. regressions could help us distinguish
We estimate (3) separately for each between Hypotheses I and III. If gover-
month and then calculate the mean and nance provisions cause poor performance,
time-series standard deviation of the then we might expect certain provisions to
112 monthly estimates of the coefficients. play a stronger role. In the absence of such a
Table 20.13 summarizes the results. finding, we should wonder whether the results
The first two columns give the results, raw are driven by some other characteristic.
CORPORATE GOVERNANCE AND EQUITY PRICES 545
G 0.04 0.02
(0.04) (0.03)
Democracy 0.76* 0.63*
Portfolio (0.32) (0.26)
Delay 0.03 0.02
(0.10) (0.07)
Protection 0.07 0.01
(0.08) (0.06)
Voting 0.08 0.08
(0.13) (0.10)
Other 0.01 0.04
(0.08) (0.07)
State 0.02 0.04
(0.08) (0.06)
NASDUM 0.83 0.42 8.23 10.36 2.60 0.29
(6.94) (5.26) (6.45) (5.94) (6.39) (4.98)
SP500 0.19 0.20 0.42 0.21 0.19 0.24
(0.49) (0.42) (0.49) (0.41) (0.45) (0.40)
BM 0.04 0.14 0.06 0.11 0.06 0.15
(0.19) (0.12) (0.38) (0.29) (0.20) (0.11)
SIZE 0.17 0.22 0.47 0.02 0.19 0.24
(0.27) (0.16) (0.38) (0.32) (0.27) (0.17)
PRICE 0.26 0.20 0.28 0.44 0.20 0.16
(0.26) (0.20) (0.31) (0.31) (0.28) (0.22)
IO 0.61 0.10 0.78 0.16 0.59 0.14
(0.47) (0.33) (0.67) (0.60) (0.44) (0.33)
NYDVOL 0.11 0.21 0.49 0.03 0.13 0.21
(0.29) (0.18) (0.36) (0.31) (0.28) (0.18)
NADVOL 0.01 0.13 0.09 0.48 0.06 0.15
(0.43) (0.29) (0.41) (0.39) (0.43) (0.29)
YLD 10.85 10.94 15.74 9.23 6.21 8.76
(10.54) (7.25) (14.62) (11.56) (11.63) (7.70)
RET2–3 0.48 0.93 2.04 1.82 0.57 1.03
(1.40) (1.04) (2.33) (1.73) (1.43) (1.07)
RET4–6 0.68 0.48 2.21 1.12 0.58 0.55
(1.33) (0.92) (1.89) (1.36) (1.33) (0.93)
RET7–12 2.42* 0.89 0.12 1.67 2.69** 1.06
(1.00) (0.65) (1.35) (1.03) (0.99) (0.65)
SGROWTH 0.00 0.03 0.75 0.27 0.01 0.02
(0.26) (0.18) (0.47) (0.40) (0.25) (0.18)
Constant 0.53 0.18 1.17 1.86 0.03 0.16
(2.55) (1.71) (3.43) (2.99) (2.39) (1.69)
This table presents the average coefficients and time-series standard errors for 112 cross-sectional regressions for
each month from September 1990 to December 1999. The dependent variable is the stock return for month t. The
results are presented using both raw and industry-adjusted returns, with industry adjustments done using the 48
industries of Fama and French [1997]. The first and second columns include all firms with data for all right-hand
side variables and use G, the Governance index, as an independent variable. In the third and fourth columns, the
sample is restricted to firms in either the Democracy (G 5) or Dictatorship (G 14) Portfolios, and we use the
independent variable, Democracy Portfolio, a dummy variable that equals 1 when the firm is in the Democracy
Portfolio and 0 otherwise. In the fifth and sixth columns, we again include all firms with data for each explanatory
variable and use the subindices, Delay, Protection, Voting, Other, and State as regressors.The calculation of G and
the subindices is described in Section II. Definitions for all other explanatory variables are provided in Appendix 2.
All regressions are estimated with weighted least squares where all variables are weighted by market value at the
end of month t 1. Significance at the 5 percent and 1 percent levels is indicated by* and **, respectively.
546 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
For example, some legal scholars argue the possibility that some unobserved
that the Delay provisions are the only characteristic is correlated with G and is
defenses with deterrent value [Coates also the main cause of abnormal returns.
2000; Daines and Klausner 2001]. If This type of omitted-variable bias could be
managers also believe this, then the Delay something prosaic, such as imperfect indus-
subindex should also be the most important try adjustments or model misspecification,
driver of the results. or something more difficult to quantify, such
Unfortunately, large standard errors, due as a partially unobservable or immeasurable
in part to the substantial multicollinearity “corporate culture.” Under the latter
between the regressors, makes it difficult explanation, management behavior would
to construct a powerful test. None of the be constrained by cultural norms within the
subindex coefficients are statistically signifi- firm, and democracy and dictatorship
cant in either specification, but many of would be a persistent feature of a corpo-
the point estimates are economically large. rate culture; G would be a symptom, but
In the end, we cannot precisely measure not a cause, of this culture. In this case, all
the relative importance of Delay or any the results of the paper could be explained
other subindex. This is similar to the if investors mispriced culture in 1990, just
problem that occurred in the Q regressions as they appear to have mispriced its
of Table 20.8. For example, in both proxy, G. The policy impact of reducing G
Tables 20.8 and 20.13 the coefficients on would be nonexistent unless it affected the
Voting suggest potentially enormous culture of managerial power that was the
economic significance, but large standard true driver of poor performance.
errors prevent any meaningful statistical In addition to the three hypotheses
inference. considered above, other explanations fall
In untabulated tests, we also included all into the general class of “Type I” error. For
28 provisions from Table 20.1 as separate example, one could argue that investors in
regressors in (3). Regressing raw returns 1990 had rational expectations about the
on these 28 provisions plus the same expected costs and benefits of takeover
controls as in Table 20.13, we find that 16 defenses, where the expected costs are
of the coefficients are negative, and only more severe agency problems and the
one (Unequal Voting) is significant. (With expected benefits are higher takeover pre-
this many regressors, we would expect one miums. Then, when the hostile takeover
to appear “significant” just by chance.) market largely evaporated in the early
Results for industry-adjusted returns are 1990s—perhaps because of macro-
similar. These results highlight and magnify economic conditions unrelated to takeover
the lack of power in the subindex regres- defenses—Dictatorship firms were left
sions. Indeed, many of the point estimates with the costs but none of the benefits of
imply return effects above 20 basis their defenses. Over the subsequent decade,
points per month (2.4 percent per year), the expected takeover premiums eroded as
but are still far from being statistically investors gradually learned about the weak
significant. This result also suggests that takeover market. Simple calculations
the Democracy-minus-Dictatorship return suggest that this explanation cannot be
differences are not driven by the presence that important. Suppose that in 1990
or absence of any one provision. the expected takeover probability for
Dictatorship firms was 30 percent, and the
V.D. Discussion expected takeover premium conditional on
takeover was also 30 percent. Further
The evidence in subsections V.A, V.B, and suppose that both of these numbers were
V.C must be interpreted with caution. zero for Democracy firms. Then, the uncon-
Since this is an experiment without random ditional expected takeover premium for
assignment, no analysis of causality Dictatorship firms would have been only
can be conclusive. The main problem is 9 percent, which is approximately the
CORPORATE GOVERNANCE AND EQUITY PRICES 547
control of the board.This slow replacement above those of other stakeholders [Pinnell
makes a classified board a crucial compo- 2000]. We treat firms in these two states
nent of the Delay group of provisions, and as though they had an expanded directors’
one of the few provisions that clearly retains duty provision unless the firm has explicitly
some deterrent value in modern take-over opted out of coverage under the law.
battles [Daines and Klausner 2001]. Fair-Price provisions limit the range of
Compensation Plans with changes- prices a bidder can pay in two-tier offers.
in-control provisions allow participants in They typically require a bidder to pay to all
incentive bonus plans to cash out options or shareholders the highest price paid to any
accelerate the payout of bonuses if there during a specified period of time before the
should be a change in control. The details commencement of a tender offer, and do
may be a written part of the compensation not apply if the deal is approved by the
agreement, or discretion may be given to board of directors or a supermajority of
the compensation committee. the target’s shareholders. The goal of this
Director indemnification Contracts are provision is to prevent pressure on the
contracts between the company and par- target’s shareholders to tender their shares
ticular officers and directors indemnifying in the front end of a two-tiered tender offer,
them from certain legal expenses and and they have the result of making such an
judgments resulting from lawsuits pertain- acquisition more expensive. Also, 25 states
ing to their conduct. Some firms have had Fair-Price laws in place in 1990, and
both “Indemnification” in their bylaws or two more states passed such laws in 1991.
charter and these additional indemnification The laws work similarly to the firm-level
“Contracts.” provisions.
Control-share Acquisition laws (see Golden Parachutes are severance
Supermajority, below). agreements that provide cash and noncash
Cumulative Voting allows a shareholder compensation to senior executives upon
to allocate his total votes in any manner an event such as termination, demotion, or
desired, where the total number of votes is resignation following a change in control.
the product of the number of shares owned They do not require shareholder approval.
and the number of directors to be elected. While such payments would appear to deter
By allowing them to concentrate their votes, takeovers by increasing their costs, one
this practice helps minority shareholders could argue that these parachutes also ease
to elect directors. Cumulative Voting and the passage of mergers through contractual
Secret Ballot (see below) are the only two compensation to the managers of the target
provisions whose presence is coded as an company [Lambert and Larcker 1985].
increase in shareholder rights, with an While the net impact on managerial
additional point to the Governance Index if entrenchment and shareholder wealth is
the provision is absent. ambiguous, the more important effect is
Directors’ Duties provisions allow direc- the clear decrease in shareholder rights.
tors to consider constituencies other than In this case, the “right” is the ability of a
shareholders when considering a merger. controlling shareholder to fire management
These constituencies may include, for without incurring an additional cost. Golden
example, employees, host communities, or Parachutes are highly correlated with all
suppliers. This provision provides boards the other takeover defenses. Out of 21 pair-
of directors with a legal basis for rejecting wise correlations with the other firm-level
a takeover that would have been beneficial provisions, 15 are positive, 10 of these
to shareholders. Thirty-one states have positive correlations are significant, and only
Directors’ Duties laws allowing similar one of the negative correlations is signifi-
expansions of constituencies, but in cant.Thus, we treat Golden Parachutes as a
only two of these states (Indiana and restriction of shareholder rights.
Pennsylvania) are the laws explicit that the Director Indemnification uses the bylaws,
claims of shareholders should not be held charter, or both to indemnify officers and
550 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
directors from certain legal expenses and are used to count proxy votes, and the
judgments resulting from lawsuits pertain- management usually agrees not to look at
ing to their conduct. Some firms have both individual proxy cards. This can help
this “Indemnification” in their bylaws or eliminate potential conflicts of interest for
charter and additional indemnification fiduciaries voting shares on behalf of
“Contracts.” The cost of such protection others, and can reduce pressure by
can be used as a market measure of the management on shareholder-employees or
quality of corporate governance [Core shareholder-partners. Cumulative Voting
1997, 2000]. (see above) and Secret Ballots are the only
Limitations on director Liability are two provisions whose presence is coded as
charter amendments that limit directors’ an increase in shareholder rights, with an
personal liability to the extent allowed by additional point to the Governance Index if
state law. They often eliminate personal the provision is absent.
liability for breaches of the duty of care, Executive Severance agreements assure
but not for breaches of the duty of loyalty high-level executives of their positions or
or for acts of intentional misconduct or some compensation and are not contingent
knowing violation of the law. upon a change in control (unlike Golden or
Pension Parachutes prevent an acquirer Silver Parachutes).
from using surplus cash in the pension fund Silver Parachutes are similar to Golden
of the target to finance an acquisition. Parachutes in that they provide severance
Surplus funds are required to remain the payments upon a change in corporate
property of the pension fund and to be used control, but differ in that a large number of
for plan participants’ benefits. a firm’s employees are eligible for these
Poison Pills provide their holders with benefits. Since Silver Parachutes do not
special rights in the case of a triggering protect the key decision makers in a
event such as a hostile takeover bid. If a merger, we classified them in the Other
deal is approved by the board of directors, group rather than in the Protection group.
the poison pill can be revoked, but if the Special Meeting limitations either
deal is not approved and the bidder pro- increase the level of shareholder support
ceeds, the pill is triggered. Typical poison required to call a special meeting beyond
pills give the holders of the target’s stock that specified by state law or eliminate the
other than the bidder the right to purchase ability to call one entirely. Such provisions
stock in the target or the bidder’s company add extra time to proxy fights, since bidders
at a steep discount, making the target must wait until the regularly scheduled
unattractive or diluting the acquirer’s annual meeting to replace board members
voting power. Poison pills are a crucial or dismantle takeover defenses. This delay
component of the “delay” strategy at is especially potent when combined with
the core of modern defensive tactics. limitations on actions by written consent
Nevertheless, we do not include poison pills (see below).
in the Delay group of provisions, but Supermajority requirements for approval
include it in the Other group because the of mergers are charter provisions that
pill itself can be passed on less than one- establish voting requirements for mergers or
day’s notice, so it need not be in place for other business combinations that are higher
the other Delay provisions to be effective. than the threshold requirements of state law.
The other provisions in this group require a They are typically 66.7, 75, or 85 percent,
shareholder vote, so they cannot be passed and often exceed attendance at the annual
on short notice. See Coates [2000] and meeting. In practice, these provisions are
Daines and Klausner [2001] for a discussion similar to Control-Share Acquisition
of this point. laws. These laws require a majority of
Under a Secret Ballot (also called disinterested shareholders to vote on
confidential voting), either an independent whether a newly qualifying large share-
third party or employees sworn to secrecy holder has voting rights. They were in place
CORPORATE GOVERNANCE AND EQUITY PRICES 551
suggest that the majority of the provisions were but still allow for some underperformance for
adopted in the 1980s. Danielson and Karpoff acquisitions financed by stock. A related debate
[1998] perform a detailed study on a similar set on whether diversifying acquisitions destroy
of provisions and demonstrate a rapid pace of value has grown too large to survey here. The
change between 1984 and 1989. seminal works are Lang and Stulz [1994] and
13 Berkshire Hathaway disappeared because Berger and Ofek [1995]. Recent work is
it added a second class of stock before 1998. summarized in Holmstrom and Kaplan [2001]
Firms with multiple classes of common stock and Stein [2001].
are not included in our analysis. 22 See Seyhun [1998] for a comprehensive
14 NCR disappeared after a merger. It review of this literature and a discussion of SEC
reappeared in the sample in 1998 as a spinout, rules, filing requirements, and available data.
but since it received a new permanent number 23 All of these additional variables are
from CRSP, we treat the new NCR as a different correlated with G (see Table 20.3) and, in prior
company. studies, with either firm value or abnormal
15 The industry names are from Fama and returns. See Lakonishok, Shleifer, and Vishny
French [1997], but use a slightly updated [1994] (sales growth), Gompers and Metrick
version of the SIC classification of these indus- [2001] (institutional ownership), and Morck
tries that is given on Ken French’s website (June and Yang [2001] (Q).
2001). In Sections III and V we use both this 24 For example, there is evidence that
updated classification and the corresponding earnings surprises [Bernard and Thomas 1989],
industry returns (also from the French website). dividend omissions [Michaely, Thaler, and
16 See Basu [1977] (price-to-earnings ratio), Womack 1995], and stock repurchases
Banz [1981] (size), Fama and French [1993] [Ikenberry, Lakonishok, and Vermaelen 1995]
(size and book-to-market), Lakonishok, Shleifer, have long-term drift following the event, and all
and Vishny [1994] (several value measures), seem to be relatively simple events compared
and Jegadeesh and Titman [1993] (momentum). with changes in governance structure.
17 This model extends the Fama-French
[1993] three-factor model with the addition
of a momentum factor. For details on the REFERENCES
construction of the factors, see Fama and
French [1993] and Carhart [1997]. We are
grateful to Ken French for providing the factor Banz, Rolf, “The Relation between Return and
returns for SMB and HML. Momentum returns Market Value of Stocks,” Journal of
were calculated by the authors using the Financial Economics, XXXVIII (1981),
procedures of Carhart [1997]. 269–296.
18 Unlike Shin and Stulz [2000], we do not Basu, Sanjoy, “The Investment Performance of
trim the sample of observations that have Common Stocks in Relation to Their Price-
extreme independent variables. Results with a to-Earnings: A Test of the Efficient Markets
trimmed sample are nearly identical and are Hypothesis,” Journal of Finance, XXXII
available from the authors.
(1977), 663–682.
19 Table 20.7, first row, second column,
shows an alpha of 45 bp per month for the Baumol, William, Business Behavior, Value, and
EW difference between the Democracy and Growth (New York, NY: Macmillan, 1959).
Dictatorship Portfolios. Over 112 months Baysinger, Barry D., and Henry N. Butler,
this produces a difference of approximately “Corporate Governance and the Board of
50 percent, as compared with the 56 19 Directors: Performance Effects of Changes
37 percent difference estimated for the Q in Board Composition,” Journal of Law,
regressions. We use the EW alpha as a Economics, and Organization, I (1985),
comparison because the Q regressions are also 101–124.
equal-weighted. Berger, Philip G., and Eli Ofek, “Diversification’s
20 For an alternative view, see Stein [1988,
Effect on Firm Value,” Journal of Financial
1989]. Empirical evidence on this issue is given
by Daines and Klausner [2001], Johnson and Economics, XXXVII (1995), 39–65.
Rao [1997], Meulbroek et al. [1990], Pugh, Bernard, Victor, and J. K. Thomas, “Post-
Page, and Jahera [1992], and Titman, Wei, and Earnings Announcement Drift: Delayed Price
Xie [2001]. Response or Risk Premium?” Journal of
21 Mitchell and Stafford [2000] have chal- Accounting Research, Supplement, XXVII
lenged the magnitude of this longrun evidence, (1989), 1–36.
554 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Bertrand, Marianne, and Sendhil Mullainathan, Antitakeover Measures,” Journal of
“Is There Discretion in Wage Setting?” Financial Economics, XXXIX (1995), 3–43.
Rand Journal of Economics, XXX (1999a), Core, John, “On the Corporate Demand for
535–554. Directors’ and Officers’ Insurance,” Journal
Bertrand, Marianne, and Sendhil Mullainathan, of Risk and Insurance, LXIV (1997), 63–87.
“Corporate Governance and Executive Pay: ——, “The Directors’ and Officers’ Insurance
Evidence from Takeover Legislation,”Working Premium: An Outside Assessment of the
Paper, MIT Department of Economics, 1999b. Quality of Corporate Governance,” Journal of
Bertrand, Marianne, and Sendhil Mullainathan, Law, Economics, and Organization, XVI
“Enjoying the Quiet Life? Managerial (2000), 449–477.
Behavior Following Anti-Takeover Core, John E., Robert W. Holthausen, and David
Legislation,” Working Paper, Department of F. Larcker, “Corporate Governance, Chief
Economics, Massachusetts Institute of Executive Officer Compensation, and Firm
Technology, 2000. Performance,” Journal of Financial
Bhagat, Sanjai, and B. S. Black, “The Economics, LI (1999), 371–406.
Relationship between Board Composition and Daines, Robert, “Does Delaware Law Improve
Firm Performance,” in Comparative Corporate Firm Value?” Journal of Financial
Governance: The State of the Art and Economics, LXII (2001), 525–558.
Emerging Research, K. Hopt, M. Roe, and Daines, Robert, and Michael Klausner, “Do IPO
E. Wymeersch, eds. (Oxford, UK: Clarendon Charters Maximize Firm Value? Antitakeover
Press; New York, NY: Oxford University Protection in IPOs,” Journal of Law,
Press, 1998). Economics, and Organization, XVII (2001),
Bhagat, Sanjai, and Roberta Romano, “Event 83–120.
Studies and the Law: Part II—Empirical Danielson, Morris G., and Jonathan M. Karpoff,
Studies of Corporate Law,” Working Paper, “On the Uses of Corporate Governance
Yale University, 2001. Provisions,” Journal of Corporate Finance,
Bittlingmayer, George, “The Market for IV (1998), 347–371.
Corporate Control (Including Take-overs),” in Demsetz, Harold, and Kenneth Lehn, “The
Encyclopedia of Law and Economics, Structure of Corporate Ownership: Causes
Vol. III—The Regulation of Contracts, and Consequences,” Journal of Political
Boudewijn Bouckaert and Gerrit de Geest, eds. Economy, XCIII (1985), 1155–1177.
(Cheltenham, UK, and Northampton, MA: Eckbo, B. Espen,“Valuation Effects of Greenmail
Edward Elgar, 2000). Prohibitions,” Journal of Financial and
Borokhovich, Kenneth A., Kelly R. Brunarski, Quantitative Analysis, XXV (1990), 491–505.
and Robert Parrino, “CEO Contracting and Fama, Eugene F., and Kenneth R. French,
Antitakeover Amendments,” Journal of “Common Risk Factors in the Returns on
Finance, LII (1997), 1495–1518. Bonds and Stocks,” Journal of Financial
Brennan, Michael J., Tarun Chordia, and Economics, XXXIII (1993), 3–53.
Avanidhar Subrahmanyam,“Alternative Factor Fama, Eugene F., and Kenneth R. French,
Specifications, Security Characteristics, and “Industry Costs of Equity,” Journal of
the Cross Section of Expected Stock Financial Economics, XLIII (1997), 153–194.
Returns,” Journal of Financial Economics, Fama, Eugene F., and James D. MacBeth,“Risk,
XLIX (1998), 345–375. Return, and Equilibrium: Empirical Tests,”
Carhart, Mark, “On Persistence in Mutual Fund Journal of Political Economy, LXXXI
Performance,” Journal of Finance, LII (1973), 607–636.
(1997), 57–82. Garvey, Gerald T., and Gordon Hanka, “Capital
Coates, John, “Takeover Defenses in the Shadow Structure and Corporate Control: The Effect
of the Pill: A Critique of the Scientific of Antitakeover Statutes on Firm Leverage,”
Evidence,” Texas Law Review, LXXIX Journal of Finance, LIV (1999), 519–546.
(2000), 271–382. Gillan, Stuart, and Jennifer Bethel,“Managerial
Comment, Robert, and G. William Schwert, Control of the Proxy Process: The Impact on
“Poison or Placebo? Evidence on the Shareholder Voting,” Working Paper, Babson
Deterrence and Wealth Effects of Modern College, 2001.
CORPORATE GOVERNANCE AND EQUITY PRICES 555
Gompers, Paul A., and Andrew Metrick, Lambert, Richard A., and David F. Larcker,
“Institutional Investors and Equity Prices,” “Golden Parachutes, Executive Decision-
Quarterly Journal of Economics, CXIV Making and Shareholder Wealth,” Journal of
(2001), 229–260. Accounting and Economics, VII (1985),
Hermalin, Benjamin, and Michael Weisbach, 179–203.
“The Effects of Board Composition and Lang, Larry H. P., and René M. Stulz,“Tobin’s Q,
Direct Incentives on Firm Performance,” Corporate Diversification, and Firm
Financial Management, XX (1991), Performance,” Journal of Political Economy,
101–112. CII (1994), 1248–1280.
Holmstrom, Bengt, and Steven N. Kaplan, Lang, Larry H. P., René M. Stulz, and Ralph A.
“Corporate Governance and Merger Activity Walkling, “A Test of the Free Cash Flow
in the United States: Making Sense of the Hypothesis: The Case of Bidder Returns,”
1980s and 1990s,” Journal of Economic Journal of Financial Economics, XXIX
Perspectives, XV (2001), 121–144. (1991), 315–335.
Ikenberry, David, Josef Lakonishok, and Theo La Porta, Rafael, Florencio Lopez-de-Silanes,
Vermaelen, “Market Underreaction to Open Andrei Shleifer, and Robert Vishny, “Investor
Market Share Repurchases,” Journal of Protection and Corporate Valuation,”
Financial Economics, XXXIX (1995), Working Paper, Harvard University, 2001.
181–208. Lewellen, Wilbur, Claudio Loderer, and Ahron
Ishii, Joy L., “Corporate Governance in the Rosenfeld, “Merger Decisions and Executive
1990s: New Evidence on the Role of Stock Ownership in Acquiring Firms,”
Shareholder Proposals,” Senior Honors Journal of Accounting and Economics, VII
Thesis, Harvard University, 2000. (1985), 209–231.
Jegadeesh, Narasimhan, and Sheridan Titman, Loughran, Tim, and Anand M. Vijh, “Do Long-
“Returns to Buying Winners and Selling Term Shareholders Benefit from Corporate
Losers: Implications for Stock Market Acquisitions?” Journal of Finance, LII
Efficiency,” Journal of Finance, XLVIII (1997), 1765–1790.
(1993), 65–91. Marris, Robin, The Economic Theory of
Jensen, Michael, “Agency Costs of Free Cash Managerial Capitalism (New York, NY: Free
Flow, Corporate Finance, and Takeovers,” Press of Glencoe, 1964).
American Economic Review, LXXVI (1986), McGurn, Patrick S., Confidential Proxy Voting
323–329. (Washington, DC: Investor Responsibility
Jensen, Michael, and Richard Ruback, “The Research Center Inc., 1989).
Market for Corporate Control: The Scientific Meulbroek, Lisa K., Mark L. Mitchell, J. Harold
Evidence,” Journal of Financial Economics, Mulherin, Jeffrey M. Netter, and Annette B.
XI (1983), 5–50. Poulsen, “Shark Repellents and Managerial
Johnson, Mark S., and Ramesh P. Rao, “The Myopia: An Empirical Test,” Journal of
Impact of Antitakeover Amendments on Political Economy, XCVIII (1990),
Corporate Financial Performance,” Financial 1108–1117.
Review, XXXII (1997), 659–690. Michaely, Roni, Richard Thaler, and Kent
Kaplan, Steven N., and Luigi Zingales, “Do Womack, “Price Reactions to Dividend
Investment-Cash Flow Sensitivities Provide Initiations and Omissions,” Journal of
Useful Measures of Financing Constraints?” Finance, XXXVIII (1995), 1597–1606.
Quarterly Journal of Economics, CXII Mitchell, Mark L., and Erik Stafford,
(1997), 169–216. “Managerial Decisions and Long-Term Stock
Karpoff, Jonathan M., and Paul H. Malatesta, Price Performance,” Journal of Business,
“The Wealth Effects of Second-Generation LXXIII (2000), 287–330.
State Takeover Legislation,” Journal of Morck, Randall, Andrei Shleifer, and Robert
Financial Economics, XXV (1989), 291–322. Vishny, “Management Ownership and Market
Lakonishok, Josef, Andrei Shleifer, and Robert Valuation: An Empirical Analysis,” Journal
Vishny,“Contrarian Investment, Extrapolation, of Financial Economics, XX (1988), 293–315.
and Risk,” Journal of Finance, XLIX (1994), Morck, Randall, Andrei Shleifer, and Robert
1541–1578. Vishny, “Do Managerial Objectives Drive Bad
556 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Acquisitions?” Journal of Finance, XLV Manager-Specific Investments,” Journal
(1990), 31–48. of Financial Economics, XXV (1989),
Morck, Randall, and Fan Yang, “The Mysterious 123–140.
Growing Value of the S&P 500 Shleifer, Andrei, and Robert Vishny, “A Survey
Membership,” Working Paper, University of of Corporate Governance,” Journal of
Alberta, 2001. Finance, LII (1997), 737–783.
Pinnell, Maria Carmen S., State Takeover Laws Stein, Jeremy C., “Takeover Threats and
(Washington, DC: Investor Responsibility Managerial Myopia,” Journal of Political
Research Center Inc., 2000). Economy, XCVI (1988), 61–80.
Pugh, William M., Daniel E. Page, and ——, “Efficient Markets, Inefficient Firms: A
John S. Jahera Jr., “Antitakeover Charter Model of Myopic Firm Behavior,” Quarterly
Amendments: Effects on Corporate Journal of Economics, CIV (1989),
Decisions,” Journal of Financial Research, 655–669.
XV (1992), 57–67. ——, “Agency, Information, and Corporate
Rau, P. Raghavendra, and Theo Vermaelen, Investment,” Handbook of the Economics of
“Glamour, Value and the Post-Acquisition Finance, George Constantinides, Milton
Performance of Acquiring Firms,” Journal of Harris, and René Stulz, eds. (Amsterdam:
Financial Economics, XLIX (1998), Elsevier Science, 2001).
223–253. Subramanian, Guhan, “The Influence of
Rosenbaum, Virginia, Corporate Takeover Antitakeover Statutes on Reincorporation
Defenses (Washington, DC: Investor Choice: Evidence on the ‘Race’ Debate and
Responsibility Research Center Inc., 1990, Antitakeover Overreaching,” Working Paper,
1993, 1995, 1998). Harvard Business School, 2001.
Seyhun, H. Nejat, Investment Intelligence from Titman, Sheridan, K. C. John Wei, and Feixue
Insider Trading (Cambridge, MA: MIT Press, Xie, “Capital Investments and Stock
1998). Returns,” Working Paper, University of Texas
Shin, Hyun-Han, and René M. Stulz, “Firm at Austin, 2001.
Value, Risk, and Growth Opportunities,” Williamson, Oliver, The Economics of
NBER Working Paper No. 7808, 2000. Discretionary Behavior: Managerial
Shleifer, Andrei, and Robert Vishny, “Greenmail, Objectives in a Theory of the Firm
White Knights, and Shareholders’ Interest,” (Englewood Cliffs, NJ: Prentice Hall, 1964).
Rand Journal of Economics, XVII (1986), Yermack, David, “Higher Market Valuation for
293–309. Firms with a Small Board of Directors,”
Shleifer, Andrei, and Robert Vishny, Journal of Financial Economics, XL (1996),
“Management Entrenchment: The Case of 185–211.
Chapter 21
ABSTRACT
We investigate the shareholder wealth effects of 233 joint venture announcements of Dutch
public companies in the period 1987 till 1998.The research shows that, on average, establishing
joint ventures has a positive effect on the market value of Dutch companies. Using the strategic
characteristics of joint ventures it is possible to explain and understand these wealth effects.
Our research shows that the factors of strategic intention, the context in which the strategy is
unfolded and the extent to which the company has control over the implementation strongly
explains the extent to which a joint venture can create value.
Construction
Technology
Food Marketing
Efficiency
Publishers
Chemicals
Industry
Steel/Rubber
Electronics
Transportation
Services
0 10 20 30 40 50 60 70 80
Number of joint ventures
Table 21.1 Location of the Joint Venture and Nationality of the Partners
Nationality Japan &
of Partner Netherlands Europe Nafta S. Korea China Other Total
Dutch 25 1 0 0 1 3 30
European 10 27 0 0 2 5 44
Nafta 11 9 11 0 0 2 33
Japanese &
S. Korean 3 1 1 13 0 2 20
Chinese 0 0 0 0 21 0 21
Other 0 1 1 0 1 32 35
Total 49 39 13 13 25 44 183
Note: When location could not be clearly identified, the joint venture was not included.
Table 21.1 presents the distribution by process of firms; e.g.; Copeland (1995). To
location of joint ventures and nationality explain wealth effects of joint ventures, the
of the partners. Approximately a quarter of dimensions that characterize the strategy
the joint ventures are based in the should be able to explain the observed
Netherlands. Most of the joint ventures variance in CAR.
are established in the country of one of the In the decision-making process involved
partners. This is due to the fact that the in formulating a joint venture, three strat-
Netherlands is a small country with a long egy dimensions are distinguished: Strategy
international trading history. Content, Strategy Context and Strategy
Control. The strategy content of a joint
B. Methodology venture concerns the characteristics of the
actual strategy itself. Two characteristics
The analysis of the value effect of joint of the strategy content are taken into
ventures includes only the abnormal part of account: the functional motive underlying a
the market reaction in respect to the parent joint venture and the degree of diversifica-
company share price.2 The expected returns tion realized by the joint venture. Research
on the relevant days were estimated by into joint venture motivations distinguishes
means of the market model approach. The motives stemming from market and tech-
abnormal return (ARi,t) of a share i at time nology developments; e.g., Koh and
t is calculated as follows:3 Venkatraman (1991) and Das, Sen and
Sengupta (1998). These studies show that
ARi,t Ri,t (ai biRm,t), (1) alliances primarily motivated by technolog-
ical issues have a more positive impact
where Ri,t is the effective return of share i on the market value of companies than
at time t, ai a constant term of share i, bi is marketing-oriented joint ventures. These
the systematic risk of share i, Rm,t is the studies did not examine efficiency improve-
market return at time t and ai biRm,t is ment as a motive. The influence of the
the expected return on share i according to degree of diversification of the joint venture
the estimated market model. is not clear. Depending on the theoretical
To measure the full effect of the concepts used, diversification influences the
announcement, the abnormal return (ARi,t) value effect of the alliance either positively
of the announcement day and the following (Balakrishnan and Koza [1993]) or
day were computed and averaged for all negatively (Koh and Venkatraman [1991]).
sample companies to obtain the cumulative To explain the value effects, the authors
abnormal return (CAR). The standard respectively applied the transaction cost
deviation of the abnormal return of each approach and the strategic behavior
share in the sample was estimated through perspective.
observations made during the estimation The strategy context defines the decision-
period (from 200 to 51 days before the making parameters of the environment.
announcement). A t-test was used to deter- Important elements are partner selection,
mine whether the market reaction on and the nationalities involved, and the associated
around the day of the announcement signif- cultural differences. Mohanram and Nanda
icantly deviates from zero (Brown and (1998) and Koh and Venkatraman (1991)
Warner [1985]). In addition to CAR, the found that companies gain more excess
standardized cumulative abnormal return, returns if they enter into an alliance with a
or SCAR, was included in the analysis.4 larger partner. The relative size of the
To gain further insight into why some partner appears to be a relevant factor in
reactions are negative while others are establishing the value potential of a joint
positive, we accomplish a regression analy- venture. Closely examined in the relation-
sis of the CARs using several different ship between partners is the effect of the
variables. Joint ventures and similar capital relatedness5 of the partners, i.e., the degree
investments come forth out of the strategy of similarity between their activities.
560 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Research conducted by Koh and company can exert on the joint venture.The
Venkatraman (1991) and Balakrishnan distribution of control within joint ventures
and Koza (1993) produced conflicting is discussed in depth in qualitative studies;
conclusions and, as a consequence, have e.g., Bleeke and Ernst (1993). Koh and
been a source of debate. Studies which Venkatraman (1991) found that one party
investigated the nationality of chosen having more control within an alliance had
partners and its impact on joint venture a positive effect. In contrast, Bleeke and
wealth effects yielded few results so far.The Ernst (1993) argue that an equal balance
studies focused on American companies of control makes it easier to manage a joint
(Lee and Wyatt [1990] and Borde, Whyte, venture and consequently increases the
Wiant and Hoffman [1998]) and produced probability of success. None of the studies
contradictory lists of countries and economic in Table 21.2 developed an integral model
regions where joint ventures were either to determine the conditions under which a
successful or unsuccessful in generating joint venture creates value.
shareholder value.
Following Datta and Puia (1995) and
Kogut and Singh (1988), for the purposes III. RESULTS
of this study cultural difference, measured
by the variable individualism, was selected A. Results of the event study
as the distinguishing indicator of national-
ity, since it was also used in qualitative Average CAR is positive but there is
studies by Bleeke and Ernst (1993) and substantial variance, with 57% of joint
others. As a rule, these studies show that ventures associated with positive share price
cultural differences can lead to manage- impacts, and 43% negative. Table 21.3
ment problems. We therefore expected this provides the results obtained by testing the
dimension to have a negative effect. CAR and SCAR of the 233 joint venture
Finally, Strategic control is the extent to announcements. Over the two-day testing
which an enterprise can exert influence period positive market reactions were
on the development of the joint venture. found with a significance level of 0.01 for
The ownership structure is a clear manifes- both performance criteria (CAR and
tation of the degree of influence that a SCAR). The CAR is equal to 0.40%.
Note: The table shows the CAR and SCAR for various event windows for a sample of 233 joint venture
announcements by Dutch public companies in the period 1987 to 1998. ***p 0.01 and **0.01 p 0.05.
According to CAR, the reactions to joint analysis, the CAR is the dependent variable
ventures were both strongly positive for the following variables:
(1.77% on average) and negative
(1.34% on average).
To gain further insight into why some
Technology
Carj b0 b1
Dummy j
reactions are negative while others are
b
Dummy
Efficiency
positive, we accomplish a regression 2
analysis of the CARs using a strategic j
explanation model. b3(Diversification)j
b4(Individualism)j
B. Impact of the joint venture b5(RelSize)j
strategy on CAR
None of the American studies attempted to
b6 Majority
Dummy j
develop an integrated model that reveals
the dynamics among the strategic factors
and the value effects of the joint ventures. b7 Minority
Dummy j
The distribution of the factors according to
strategic dimensions is addressed above in
section II. Investor’s interpretation of b8 Risk-free
Rate j
the strategy content, the context in which
Industry
Dummy
this strategy must function, and the extent
of the company’s control over the imple- b9 e j (2)
j
mentation of the strategic option determine
how much shareholder value is generated. where ej N(0,2). An explanation of
The variables investment climate (risk-free the independent variables of the model
rate6) and industry were included in the follows below:
analysis as control variables to ensure that
they did not have any impact on the
research results. Strategy content
Three dichotomous variables are used to
test the impact on the parent company
Joint venture strategy of the underlying motive of the joint venture.
This section describes the variables used The value of the dichotomous variable is 1
to identify factors that influence the share- if there is a distinguishable motive,
holder wealth effects. Each strategic dimen- otherwise it is 0. The dummy variables are
sion is discussed in terms of what factors technology, in the case of a technology
were examined and how the variables were development joint venture; marketing, for
examined and calculated. In the regression a market development joint venture; and
562 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
efficiency, for an efficiency-driven joint and thus economic growth. In their long
venture. term study they find empirical support for
The diversification variable represents their thesis. From the proxies used to char-
the relatedness between the activities of the acterize cultural distance, it became clear
parent company and those of the joint that the cultural dimension Individualism had
venture. The relatedness is the difference the most explanatory power in explaining
between the primary three-digit US-SIC economic growth. The dimension of individ-
number for the parent company7 and the ualism is a criterion for the way in which an
joint venture.8 The relatedness is divided by individual views his or her relationship with
899, the maximum possible distance used the rest of the collective10. Frank, Hofstede
by Balakrishnan and Koza (1993) in their and Bond (1991) constructed scores on
research. this dimension based on most major
countries. Datta and Puia (1995), Kogut
and Singh (1988), Erramilli (1991) and
Strategy context Shane (1992) used these bipolar scores
The relative size of the partner is calculated to investigate cultural distance. This study
by dividing the number of staff the partner used these scores in calculating cultural
employs by the number of staff working for distance measured by the individualism
the company whose shares are analyzed.9 dimension. The variable Individualism is
In discussing the success and failure of calculated by:
cooperative agreements, cultural differences
are often used as a reason for failure (Ij INL)2
(Bleeke and Ernst; 1993). Franke, Hofstede Individualism , (3)
V
and Bond (1991) argue that differences in
cultural values, rather than in material and where Ij is an index-score on the individu-
structural conditions, are ultimate determi- alism index of the country of origin of
nants of human organization and behavior, the partner, INL is the index-score of the
Europe
Technology
driven Minority
Nafta-countries interest
China
SCAR CAR
R2 0.492 0.492
Adj.R2 0.399 0.399
F-value 5.530*** 5.303***
N 110 110
Note: This table shows the results of the estimated functions of the cumulative abnormal returns of a sample
of 110 joint venture announcements of Dutch public companies in the period 1987 till 1998. ***p 0.01,
**
0.01 p 0.05 and *0.05 p 0.10.
individualism index of the Netherlands and return is equal to the yield of ten-year
V is the variance of individualism scores of government bonds.
all countries. The study also takes industry effects into
account. Based on SIC number,11 the com-
panies in the sample are classified into the
Strategy control following sectors of industry: construction,
food, publishing, (petro) chemical, steel and
The ownership structure within the joint rubber, electronics, logistics, and services.
venture is represented by three dummy Industry effects are measured using
variables.Three structures are possible: the dummy variables which take the value of
company can have a majority, a minority or one if the firm is in that particular industry
an equal distribution of shares in the joint otherwise zero.
venture. It is assumed that the ownership
structure correlates with the degree of con- Autonomous relationships
trol in the joint venture. The dummy vari-
ables take the value of one if the firm has Figure 21.2 shows the autonomous
that particular ownership structure and the relationships uncovered by the study.
value of zero, otherwise. Technology joint ventures are mostly
initiated between non-horizontal partners
in NAFTA territory. Efficiency driven joint
Control variables ventures are often between horizontal
partners within the E.U. region. Marketing
In order to prevent investment climate driven joint ventures are frequently setup in
and industry specific factors to influence emerging markets where for China the
the overall results of this study we control Dutch firms mostly have a majority stake.
for them by using control variables in In China it is still obligatory to form joint
the regression. To measure the effect of ventures; the Chinese government prevents
investment climate the variable risk-free foreign firms to have full control of Chinese
rate (Rf) is added to the model. This subsidiaries.
564 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
of the economic power of the partner’s venture is especially positive for companies
country of origin and business location. that have a majority interest. A majority
Research shows that the relatedness of interest gives a company a dominant posi-
the partner’s activities is of less impor- tion in the collaboration, ensuring that the
tance in determining the value generated by company has more control over the
a joint venture. Investors focus on the achievement of the objectives of the joint
activities of the joint venture. However, the venture. The results contradict the prevail-
analysis reveals that horizontal partners are ing paradigms that assert the importance
usually selected for horizontal joint of equality in cooperative alliances. Bleeke
ventures and vice versa. Neither conven- and Ernst (1993), Harrigan (1988) and
tional testing of differences in means nor Copeland et al. (1995) underlined the
regression analysis uncovered any impact importance of equality in the ownership
on value creation by joint ventures. The structure. If shareholder value is the tar-
overlap in the partners’ activities is the get, then equality within an alliance is not
result, rather than the cause, of the joint essential. It is more important that the
venture and its objective. The debate con- ratios should be fair and bear a connection
ducted in the literature on the impact of to the value of the resources (knowledge,
overlapping strategic activities appears to capital and other assets) in which each
be irrelevant in our research. party invests in the joint venture. It appears
Partner size has a strong effect on the that the equal distribution of share capital
performance of a joint venture.The variable at any cost is not appreciated by investors
is highly significant and positive for all per- and has a negative impact on market value.
formance criteria. In their qualitative Furthermore, investors have more confi-
study, Bleeke and Ernst (1993) concluded dence in joint ventures with one captain
that a stronger partner is a prerequisite for than those with two.
a successful joint venture. The regression The synthesis of the aggregate strategic
results and the univariate tests support this dimensions explains a great deal of the
theory. The results of this study show that share price reaction. Clearly, investors
venturing with relative large partners respond consistently enough to joint ven-
result in higher wealth effects. For smaller ture announcements to justify the develop-
firms the joint venture can underline the ment of an analytical model based on these
value of its business proposition through factors.
the acceptance of larger partners. The ini-
tial impact of venturing with smaller part-
ners doesn’t seem to be very high. When NOTES
there is much excitement surrounding a
joint venture with a smaller partner, this * We acknowledge comments on prior drafts
can be attributed to the promise of tech- by the anonymous referee and the editor Peter
Theodossiou.
nology/innovation. 1 Financial institutions were not included in
the sample because they are not comparable to
other sectors in that they are regulated by the
C. Strategy control Dutch Central Bank. Moreover their annual
Joint venture announcements with an accounts differ strongly from those of other
unequal partner shares gain higher wealth companies.
effects then with an equal (50/50) partner 2 In order to verify the stability of the esti-
mated market model, Cannella and Hambrick
share. From the data observed, it seems (1993), three alternative performance criteria
that joint ventures with an equal share dis- have been calculated. They are based on the
tribution receive a negative premium. following: (i) Returns corrected for the market.
Minority partner shares receive on average This method comprises the effective return
positive wealth effects, but report less minus the market return, i.e. without estimating
strong significance in the multivariate the parameters alpha and beta as in performance
regression tests. The valuation of a joint criteria (1). (ii) Returns corrected for the
566 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
average. The return is calculated by subtracting Borde, S.F.; Whyte, A.M.; Wiant, K.J; and
the average return of the share in the estimation Hoffman L.L. 1998. New evidence on factors
period from the effective return. (iii) Effective that influence the wealth of international joint
return. The uncorrected return during the ventures. Journal of Multinational Financial
announcement. In addition a verification Management 8: 63–77.
analysis was also performed based on the
Brown, S.J. and Warner, J.B. 1985. Using daily
‘buy and hold’ investment strategy. The extra-
ordinary buy and hold return is also based on stock returns: The case of event studies.
the market model; however a composite return Journal of Financial Economics 14: 3–31.
is the starting point.The results of the alternative Cannella Jr, A.A. and Hambrick, D.C. 1993.
methodologies are similar to those reported in Effects of executive departures on the per-
the paper. formance of acquired firms. Strategic
3 The parameters a and b were estimated Management Journal 14: 137–152.
using the least square method, based on obser- Chan, S.H.; Kensinger, J.W.; Keown, A.J; and
vations made during the estimation period. The Martin J.D. 1997. Do strategic alliances cre-
estimation period runs from 200 to 51 days ate value? Journal of Financial Economics
before the announcement day.
46: 199–221.
4 Unlike CAR, SCAR is normally distributed
for each hypothesis (Strong [1992]). Copeland, T.; Koller, T.; and Murrin J. 1995.
5 To avoid confusion the term relatedness is Valuation: Measuring and managing the
used to characterize similarities between value of companies. John Wiley & Sons, Inc:
parents and the term diversification is reserved New York.
to characterize similarities between the activi- Das, S.; Sen, P. K.; and Sengupta, S. 1998.
ties of the joint venture and the parent that is Impact of strategic alliances on firm valua-
under investigation. tion. Academy of Management Journal 41:
6 In accordance to economic theory, when 27–41.
the risk free rate lowers, so does the cost of Datta, D.K. and Puia G. 1995. Cross-border
capital. With the cost of capital lower it is more
acquisitions: an examination of the influence
attractive for companies to invest. So the risk
free rate is a proxy of the investment climate. of relatedness and cultural fit on shareholder
This control variable helps in identifying over-or value creation in U.S. acquiring firms.
undervaluation of the venture due to the Management International Review 35:
economic situation at that particular time. 337–359.
7 Source: Worldscope. Erramilli, M.K. 1991. The experience factor in
8 Source: KPMG Dealwatch. foreign market entry behavior of service
9 Das el al. (1998) also used this proxy for firms. Journal of International Business
the partners in the joint venture. Studies 22: 479–501.
10 Also used in Hofstede (1980). Fama, E.F; Fisher, L.; Jensen, M.C.; and Roll, R.
11 Standard industry classifications.
1969. The adjustment of stock prices to new
information. International Economic Review
10: 1–21.
REFERENCES Franke, R.H; Hofstede, G.; and Bond, H. 1991.
Cultural roots of economic performance: A
Balakrishnan, S. and Koza, M.P. 1993. research note. Strategic Management
Information asymmetry, adverse selection Journal 12: 165–173.
and joint-ventures. Journal of Economic Harrigan, K.R. 1988. Joint ventures and com-
Behavior and Organization 20: 99–117. petitive strategy. Strategic Management
Berg, S. and Friedman P. 1977. Joint ventures, Journal 9: 141–158.
competition and technological complemen- Kogut, B. and Singh, H. 1988. The effect of
taries. Southern Economic Journal 43: national culture on the choice of entry mode.
1330–1337. Journal of International Business Studies 19:
Bleeke, J. and Ernst D. 1993. Collaborating 411–432.
to Compete. John Wiley & Sons, Inc: Koh, J. and Venkatraman, N. 1991. Joint
New York. venture formations and stock market
SHAREHOLDERS WEALTH EFFECTS OF JOINT VENTURE STRATEGIES 567
ABSTRACT
Firms in performance decline may choose a variety of restructuring strategies for recovery with
conflicting welfare implications for different stakeholders such as shareholders, lenders and
managers. Choice of recovery strategies is therefore determined by the complex interplay of
ownership structure, corporate governance and lender monitoring of such firms. For a sample
of 297 U.K. firms experiencing relative stock return decline during 1987–93, we examine the
impact of these factors as well as other control factors on their turnaround strategies. Strategy
choices during the decline year and two post-decline years are modelled with logit regressions.
Our results show that turnaround strategy choices are significantly influenced by both agency
and control variables. While there is agreement among stakeholders on certain strategies
there is also evidence of conflict of interests among them. There is further evidence of shifting
coalitions of stakeholders for or against certain strategies.
2.1. Impact of agency conflicts such debt restructuring. On the other hand,
on turnaround strategy choice turnaround based on fresh infusion of
equity is likely to be preferred by lenders
The restraints on managerial choice of but frowned upon by shareholders. Dividend
turnaround strategies may be examined cuts may be loathed by shareholders but
within the context of the agency conflicts supported by lenders.
among shareholders, managers and Strategic or asset restructuring in the
lenders. The motivations of these players form of divestments may be favoured by
also provide the impetus to the pursuit of shareholders provided the divestment
turnaround strategies so that firm value proceeds are not used to pay down debt.
is enhanced and its ability to meet its Lenders may support divestment provided
financial commitments is restored. their debt is paid off. New investments of
While both lenders and shareholders a high risk nature may be preferred by
have a common interest in restoring firm shareholders but not necessarily by lenders
viability and its ability to generate (Myers, 1977).
adequate returns to their investment in the
firm, in the turnaround process either group
2.2. Impact of lender monitoring
may gain at the expense of the other.
Shareholders may benefit from a transfer on managerial choice
of wealth from creditors when managers In the agency model of the firm posited by
undertake risky investments (Myers, 1977). Jensen (1989a) a highly leveraged firm
Likewise, lenders may benefit from a will react faster to poor performance than
wealth transfer from shareholders when less leveraged ones due to a desire to avoid
sale of assets to pay off debts eliminates the breaching debt covenants. This early
option value of those assets (Lang et al., response preserves the going-concern value
1995). Similarly, managers may pursue of highly leveraged firms as compared to
turnaround strategies which least harm less-leveraged firms.
them while the burden of turnaround is Ofek (1993) examines the role of lender
borne by shareholders or lenders or both. monitoring within the agency paradigm in
Managers’ pursuit of self-serving objectives influencing the choice of restucturing
may manifest itself in their choice of strategies of poorly performing companies
strategies. However, managerial discretion in the U.S.A. He finds that high gearing
in choice of strategy may be tempered significantly increases the probability of
by the agency control mechanism in place financial and operational restructuring.
in the firm. Gilson, John and Lang (1990) find no
Broadly, the turnaround options available relation between gearing and financial
to declining firms include: operational, restructuring.1
business strategic, managerial and financial High leverage is also found by Storey et
restructuring. Not all of these actions will al. (1987) to be more positively associated
appeal equally to shareholders, managers with equity rights issues (i.e. financial
and lenders since they demand different restructuring) in failing firms than in
degrees of sacrifice from these stakeholders non-failing firms. They attribute this to the
during the turnaround process. monitoring pressure from bank creditors
Managerial restructuring, e.g. replacement who are only willing to continue financial
of the top managers, is obviously unlikely to support conditional upon shareholders
be favoured by managers and where the sharing a part of the burden of turnaround.
governance structure is weak and the man- Impact of debt on managerial choice of
agement is entrenched such replacement turnaround strategies may depend on the
may not happen. Similarly, where financial characteristics of debt such as ownership,
restructuring involves additional borrowing maturity structure and security available.
or dilution of the covenants protecting Debt ownership by informed bank creditors
existing lenders they are likely to resist promotes more effective monitoring than
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 571
institutional block shareholding. Agrawal issue (e.g. Schipper and Smith, 1986) and
and Mandelker (1990) provide evidence of asset sales where proceeds are utilised to
a positive relation between institutional pay down debts (Lang et al., 1995).
ownership and stockholder wealth effects of
various types of anti-takeover amendments
2.4. Impact of corporate
in target companies. Jarrel and Poulsen
(1987) show that firms that adopt the
governance structure
most value-reducing forms of anti-takeover Corporate governance structure as a
charter amendments also have lower insti- monitoring mechanism to reduce the agency
tutional shareholding than do other firms. problem between shareholders and man-
Brickley et al. (1988) find evidence that agers has recently received much attention
institutional investors who do not have (Cadbury Report, 1992). Composition of
business dealings with corporate manage- the board of directors is an important part
ment are more likely to vote against of this structure, influencing its policing
anti-takeover amendments. All these effectiveness. Board composition is there-
U.S.-based results are consistent with fore likely to impact significantly on the
the reduction of agency costs due to large choice of turnaround strategies.
shareholders monitoring.2 Boards of directors differ in a number of
The positive valuation impact of large ways: the relative importance of executive
share acquisitions has been evidenced in a versus non-executive directors; the leader-
number of studies. Barclay and Holderness ship of the board by an executive or
(1991), Mikkelson and Ruback (1985), non-executive chairman and the separation
Holderness and Sheehan (1985) and Choi of the roles of the chairman of the board
(1991) report, for the U.S., that block and the CEO. In the context of declining
acquisitions in excess of 5% generate firms, their performance decline may have
significant wealth gains for target share- been caused by managerial entrenchment
holders. For the U.K., Sudarsanam (1996) and weak governance structure may have
reports similar results. contributed to this entrenchment. Turn-
Shivdasani (1993) emphasises the need to around may, therefore, demand managerial
differentiate associated from non-associated restructuring with the top management
non-institutional block holders. Shareholders being replaced. Whether managerial
associated with incumbent management, restructuring can be carried out depends
e.g. family trusts or company pension funds upon the independence and strength of the
are less likely to provide effective monitoring board as well as the power of block share-
of managers. Shivdasani (1993) finds holders and lenders.
evidence that unassociated shareholders According to Fama and Jensen (1983),
increase the probability of hostile takeovers, the separation of decision management and
invoking their disciplinary motives. decision control in the decision-making
Bethel and Liebeskind (1993) report process can alleviate the agency problem.
that block share ownership is associated Whilst inside directors are responsible for
with corporate restructuring. Sudarsanam decision management, decision control
(1995) finds substantial asset, financial should be left with outside directors.
and managerial restructuring following Outside directors has an incentive to
large block acquisitions and value monitor management actions since they have
increases attendant upon such acquisitions staked their reputation as professional
are maintained or enhanced over the corporate referees. Consequently, the higher
following three years. the proportion of non-executive to executive
In contrast, shareholders are also known directors, the more effective would be the
to frown upon certain strategies which are board monitoring of management.
painful to themselves such as dividend Empirically, Weisbach (1988) finds
cut/omission (Asquith and Mullins, 1986; that CEO turnover is highly correlated with
DeAngelo and DeAngelo, 1990), rights proportion of outside directors to inside
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 573
Restructuring strategies selected by firms experiencing stock return performance decline are
identified and defined. Information on strategies is from press releases to the London Stock
Exchange which are documented by Extel Financial News Summary from 1987, with the exception
of capital expenditure. Capital expenditure is based on annual reports and accounts. Supplementary
information is also collected from Andersen Corporate Register and Hambro Company Guide,
Datastream International, and Company Reports and Accounts. These alternative sources are also
used for cross-checking information reported in the Extel Financial News Summary.
Strategy Definition
Operational restructuring
Operational restructuring Cost rationalisation, layoffs, closures and integration of
business units.
Asset restructuring
Asset sales Divestment of subsidiaries, management buy-outs, spin-offs,
sale-and-leaseback, and other asset sales.
Acquisitions Full and partial acquisitions of businesses.
Capital expenditure Internal capital expenditure on fixed assets such as plant and
machinery.
Managerial restructuring
Managerial restructuring Removal of Chairman or Chief Executive Officer
Officer/Managing Director (retirement under the age of 65
is included).
Financial restructuring
Dividend cut or omission Omission or reduction of dividends from previous year.
Equity issue Issue of equity for cash.
Debt restructuring Debt refinancing involving extending, converting or forgiving
of debt and interest.
Combination strategies
Cash generative actions Asset sales and cash equity issue.
The table defines three groups of variables representing firms’ agency monitoring mechanism which
are expected to influence the choice of restructuring strategies by poor performing firms. Debt
structure is based on accounting information provided by Datastream International and Extel
Company Research. Ownership and governance data are extracted from Andersen Corporate
Register, Hambro Company Guide and Annual Reports and Accounts. Block shareholding is holding
of 5% or more (3% or more since 31st May 1990) as disclosed in the company annual reports.
Industry is the Financial Times-Actuaries Index industry group to which the sample firm belongs.
Variable Definition
Debt structure
Leverage Total book debt over total book debt and equity
Short term leverage Short term debt over total book debt and equity
Bank leverage Bank debt over total book debt and equity
Unsecured leverage Unsecured debt over total book debt and equity
Ownership structure
Managerial shareholding Shareholding by members of the board of directors
Associated block shareholding Shareholding by family members or trusts of members
of the board and company pension plans
Institutional block shareholding Shareholding by institutional investors
Non-institutional unassociated Shareholding by non-institutional blockholders
block shareholding unassociated with management
Governance structure
Chairman cum CEO Combined role of Chief Executive and Chairman
(CEO duality)
Non-executive Chairman Chairman in a non-executive capacity
Proportion of outside directors Non-executive directors as a percentage of total
number of directors
Control variables
Economic condition GDP growth in the restructuring year
Industry condition FTA industry return in the restructuring year
Internal cause of decline Reported internal problems such as project failures,
bad acquisitions or poor financial control
Severity of decline Stock returns ranking in the year of decline
Size Log of market value of equity at the end of pre-decline year
Moreover, the decline in performance for unassociated block holding has a median
the sample is also very steep. This pattern of 0% (but a mean of 7%).
of steep decline is repeated for each of the Comparison with the only other study by
sample decline years 1987 to 1993. The Ofek (1993) that examines the determinants
sample median returns in the base years for of restructuring strategy choice during
the sample range from 6% in 1990, a performance decline reveals interesting
recession period, to 53% in 1986. In the differences in agency mechanism between
decline years the median return ranges U.S. and U.K. firms. Ofek reports median
from 12% in 1993 to 112% in 1990. leverage, managerial shareholding and
Panel B of Table 22.3 gives the per- outside (non-managerial) shareholding of
formance statistics based on accounting 31%, 22% and 6% respectively. This
variables. Both profitability and cash flows compares with our sample’s 27%, 9.5%
deteriorate significantly in the decline and 15.7% (not shown in Table 22.4).
year. The median fall in the profitability U.S. firms have apparently higher levels of
measures – operating margin, earnings managerial shareholding but lower levels
per share, return on equity and return on of outside or non-managerial sharehold-
asset – ranges from 18% to 22%, all ings. The difference may lie in difference
significant at 1%. Median fall in operating between sampling criteria as firms studied
cash flows – profit before interest, tax and by Ofek are in the bottom 10% in the
depreciation deflated by total assets – is decline year. When we test for this differ-
15%, significant at 1%. The accounting- ence by selecting only firms that fall to the
based performance measures thus reflect bottom 10% ie. 50 : 50 : 10 firms the sta-
the stock return decline. Our sample thus tistics are respectively 27%, 11% and
captures both operating performance and 16%. Therefore, the differences between
stock return performance decline.7 Excess the two studies are more likely to stem
volatility of stock returns does not appear from market differences than sampling
to cause the performance decline in our criterion differences.
sample firms. Indeed, the sample betas are As regards board composition, in 44%
not unusual (mean and median values of sample firms one person plays the dual
are both less than 1) and likely to cause the roles of Chairman and CEO. Non-executive
stock return decline. Chairmen preside over the board in 24% of
the companies. The median proportion
of outside directors in the sample boards
3.4.2. Agency monitoring mechanisms is 22%.
Table 22.4 provides descriptive statistics
on the leverage, share ownership and 3.4.3. Frequency of turnaround
governance structure in base year 1. The strategies
median leverage, total book debt to book
debt and equity, is 27% and median short- Panel A of Table 22.5 reports the frequen-
term debt, bank debt and unsecured debt as cies of sample firms undertaking different
proportions of book debt and equity are in turnaround strategies in the decline year
the range 12% to 16%. The leverage and in the two post-decline years. We find
variables are not entirely mutually exclu- that the most frequent form of restructur-
sive. For example, there may be an overlap ing is operational with 59% of the sample
between short term and bank leverage.This firms undertaking it in the decline year and
implies that when all the leverage variables 47% and 52% of the firms in the two fol-
are included in a regression, the empirical lowing years. This is comparable to Ofek’s
result has to be interpreted cautiously. 53% for the decline year.
Median directors’ shareholding is 9.5% Asset sales are carried out by between
and associated block holding is negligible. 27% and 38% of the firms in those years.
Institutional ownership amounts to a median This is much higher than the 15% rate
value of 6.9%. Non-institutional but reported by Ofek (1993) for the decline year.
Table 22.3 Descriptive statistics for stock returns and accounting performance
Panel A shows the sample firms’ stock log returns in the two years prior to and including the year of decline. Return on the Financial Times All Share
Index (FT-All), a value-weighted index based on around 800 firms covering in excess of 90% of stock market capitalisation in the London Stock
Exchange, is provided for comparison. Panel B shows changes in accounting based performance in the year of decline. PBIT profit before interest and
tax. EPS profit after tax, minority interests and preference dividends but before extraordinary items/the average number of shares in issue in the year.
Return on equity profit attributable to shareholders/share capital and reserves less intangibles. Return on assets profit before interest and tax/total
assets less current liabilities. PBITD profit before interest and tax plus depreciation (proxy for cash flows).Total debt is the total of all interest-bearing
debt, i.e. short, long and subordinated debt. Capital employed is the sum of book debt plus book equity. Beta is based on figures computed by the Risk
Measurement Service of London Business School, at the beginning of the year prior to decline. Size is market capitalisation at the beginning of the year
of decline. in Panel B the mean and median are tested using the t-test and the Mann-Whitney Wilcoxon test.
Panel A: Annual stock returns in the year of, one year prior to, and two years prior to, decline
Mean Median*
(t statistics) (Z statistics)
The table shows the descriptive statistics for the explanatory variables in the pre-decline year. For
definitions of the variables see Tables 22.1 and 22.2. Chairman cum CEO, Non-executive Chairman
and internal cause of decline are dummy variables coded as 1 when these are present and 0 if oth-
erwise. Source: Hambro Company Guide and Corporate Register, Datastream International and
Company Reports and Accounts
Mean Median
Capital structure 0.29 0.27
Leverage 0.15 0.12
Short leverage 0.19 0.16
Bank leverage 0.17 0.12
Unsecured leverage
Ownership structure (%)
Managerial shareholding 19.90 9.5
Associated block shareholding 0.71 0.0
Institutional block shareholding 12.2 6.9
Non-institutional unassociated 7.1 0.0
block shareholding
Governance structure
Chairman cum CEO 44.1% –
Non-executive Chairman 24.2% –
Proportion of outside directors 0.20 0.22
Control variables
Internal cause of decline 0.30 –
Severity of decline 10.8 10.9
Size (£M) 344.2 54.8
Surprisingly, acquisitions do not cease the decline year and 3% and 7% in the
when firms hit trouble and they are carried following years respectively taking recourse
out by nearly 50% of the sample firms in to it. In contrast, Ofek (1993) finds 11%
the decline year and by 36% and 27% of of his sample firms restructure their debt in
the firms in the post-decline years. Internal the year of decline. Debt restructuring
capital expenditure, again surprisingly, appears to be more common among U.S.
does not cease but is incurred by 62% of than U.K. firms (only 4% of firms declining
firms in the decline year and by 50% and to the bottom 10%, i.e. 50 : 50 : 10 firms,
48% in the following years. adopt it). Equity issues are made by 20%
Removal of top management is observed of sample firms in the decline year but by
in 20% (in decline year) to 26% (in decline only about 10% in the following years. The
year 1) of the sample firms. Again, Ofek most frequently employed financial
(1993) reports a similar 21% of top restructuring device is dividend cut or
management replacement in the year of omission.The proportions of firms adopting
decline. Debt restructuring is quite infre- this strategy in the three years are: 24%,
quent with only 2% of sample firms in 27% and 34%.
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 583
Table 22.5 Descriptive statistics for restructuring strategies and frequency of restructuring strategies
pursued by U.K. listed firms during 1989–1994
This table shows the distribution of corporate restructuring actions in the year of decline, one year,
and two years post-decline. Frequency is the proportion of sample firms adopting the strategy.
Sample size declines in post-decline years due to failure of firm, takeover or where no data is
available, i.e. firms declining in 1993 (68 firms) are excluded from decline year 2 analysis.
Source: Company press releases and Company Reports and Accounts.
Panel A: Descriptive statistics for restructuring strategies in the year of, one and two years after
decline
Operational restructuring
Cost rationalisation, closures and
integration of business units 58.6 46.7 51.6
Asset restructuring
Asset reduction 26.6 37.8 35.6
Acquisition 50.2 35.9 27.1
Capital expenditure 61.6 50.4 47.9
Managerial restructuring
Remove top management 19.5 25.9 21.8
Financial restructuring
Dividend cut or omission 23.6 27.0 34.0
Equity issue 20.2 10.4 13.3
Debt restructuring 2.4 3.3 7.4
Cash generative actions 40.1 44.1 43.6
Sample size 297 270 188
Panel B: Frequency of restructuring strategies pursued by U.K. listed firms during 1989–1994
Asset sales cover divestments, management buy-outs and other asset sales. Acquisitions represent
full and partial acquisitions. Dividend cut/omission refers to cut/omission in dividends per share over
the previous year. Rights issues encompass rights issue, rights offer, offer for sale, open offer and
placing of firm shares with institutions and financial intermediaries. Source: Financial Times Extel
Company Research and Company Analysis.
Average Asset sales Acquisitions Dividend cut/ Rights issue Cash generative
no. of firms % % omission % % actions %
than in the population (20%),8 from the between dominant and non-dominant
year of decline to two years thereafter. In groups are significant. Sample firms are
the case of acquisitions, sample firms lenders dominated when their leverage is in
clearly overtake the population with 50% the top quartile of all the sample firms and
versus 35% of firms in the population they are in severe decline (bottom 50% in
making acquisitions in the year of decline. sample stock ranking in the year of
However, with the onset of decline, sample decline). Lenders under such circumstances
firms reduce their rate of acquisitions to are likely to have high stakes in recovery
the population rate one year after decline. and to exercise their priority rights. Sample
Two years post-decline, far fewer sample firms are manager-owner dominated when
firms seek acquisitions compared to the they are not lender dominated according to
population. A similar pattern is observed the above definition and the managerial
with regard to equity issue. More sample and associated shareholdings are in the top
firms tap the market than the population at quartile of all sample firms.
large, 20% in the decline year, versus 15%. Where neither lenders nor manager-
However, sample firms are less likely to be owners are dominant according to the
successful in raising finance from equity above definitions and the unassociated
investors subsequent to performance block shareholding is in the top quartile of
decline than the firm in the population. all sample firms, the firms are deemed
Finally, more sample firms resort to cash block shareholders dominated. Finally, the
generative actions than the population with sample firms not dominated by lenders,
an annual average of 43% of sample firms manager-owners and block shareholders,
taking it compared to only 32% of firms they are deemed to be under the control of
in the population. Overall, our sample of the board of directors. In turn, the board
declining firms carries out various restruc- may be dominated by a dual CEO or
turing activities more intensively than the collectively by the board members.
firm population at large. Panel A of Table 22.6 shows the effect of
lender dominance on strategy choice. To
summarise, lender-dominated firms are
4. RESULTS more likely to opt for operational restruc-
turing, cash generative actions, dividend cut/
4.1. Impact of stakeholder omission and debt restructuring. They are
dominance on turnaround strategy less likely to approve of cash-consuming
choice strategy such as capital expenditure. They
have little influence on management
We divide our sample into two groups – one changes.
stakeholder dominated and the other non- In Panel B, the effects of manager-
dominated by that stakeholder. For each owner dominance are shown. Manager-owner
stakeholder group – lenders, manager- dominated firms are more likely to under-
owners, block shareholders, CEO and take capital expenditure and less inclined
collective board of directors – we examine to pursue operational restructuring and
the likelihood of a given strategy being acquisitions. They are also less likely to
chosen. The difference in the proportions sack their top management!
of sample firms in the dominated and When firms are dominated by block-
non-dominated groups choosing a strategy holders, their influence is less pronounced
is tested for statistical significance. Any and limited to two strategies as shown in
significant difference reflects the influence Panel C. These shareholders make opera-
of the dominant stakeholder. tional restructuring and capital expenditure
Table 22.6 shows the proportions of less likely. CEO dominant board influence
sample firms pursuing a given strategy in is also limited but stronger than with
the decline and two post-decline years blockholder dominance. Declining firms
when the differences in these proportions dominated by their CEOs prefer capital
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 585
The choice of restructuring strategies is determined by the relative bargaining powers of various stakeholders in the firm. The table compares the choice
of strategy between stakeholder dominated and non-dominated groups of declining firms. Four stakeholder groups are identified: lenders, managers, block-
holders and board of directors. Where a firm is highly leveraged (i.e. top quartile of sample firms ranked on leverage) and has suffered a severe decline
(bottom 50% of sample stock ranking in year of decline), leaders are deemed dominant in influencing the type of actions the firm takes. If the firm’s
decision-making process is not dominated by lenders, and managerial and associated shareholding are high (i.e. top quartile of sample firms ranked on
such shareholding), manager-owners are deemed to be entrenched and possess dominant influence. Where neither lenders nor manager-owners are dom-
inant, and unassociated blockholding are high (i.e. top quartile of sample firms) relative to the total of all significant shareholders, blockholders are
deemed dominant.The remaining firms, which are not lenders, manager-owners or blockholders dominated, are deemed to be controlled by the board.The
board is deemed to be CEO dominated if Chairman is also the CEO and collectively dominated, otherwise.The table shows the proportion of sample firms
in each group choosing the strategy. Only variables for which there is a significant difference between dominant and non-dominant groups are shown. Z is
the absolute value of test statistic for the Mann-Whitney Wilcoxon test.
Operational restructuring 0.73 0.56 1.94** 0.69 0.43 2.78*** 0.47 0.53 0.59
Cash generative actions 0.49 0.39 1.27 0.69 0.40 3.13*** 0.53 0.42 1.17
Capital expenditure 0.53 0.63 1.18 0.31 0.53 2.39** 0.33 0.51 1.73*
Dividend cut/omission 0.44 0.20 3.44*** 0.51 0.23 3.48*** 0.37 0.34 0.33
Debt restructuring 0.09 0.01 3.24*** 0.17 0.01 4.87*** 0.10 0.07 0.58
Sample size 43 254 35 235 30 158
Panel B: Manager-owner dominance
Operational restructuring 0.41 0.63 3.13*** 0.36 0.49 1.80* 0.50 0.52 0.24
Capital expenditure 0.20 0.45 3.64*** 0.22 0.50 3.74*** 0.25 0.49 2.84***
Acquisitions 0.36 0.54 2.47** 0.22 0.39 2.42** 0.23 0.28 0.75
Managerial restructuring 0.15 0.21 1.05 0.16 0.29 2.04** 0.23 0.22 0.17
Sample size 61 236 58 212 44 144
Panel C: Blockholder dominance
Operational restructuring 0.55 0.51 0.59 0.33 0.50 2.34** 0.43 0.50 1.00
Capital expenditure 0.51 0.65 2.03** 0.40 0.53 1.81* 0.53 0.48 0.24
Sample size 65 232 60 210 28 160
Panel D: CEO board dominance
Capital expenditure 0.81 0.57 3.29*** 0.67 0.46 2.58*** 0.56 0.46 1.19
Managerial restructuring 0.18 0.20 0.42 0.16 0.28 1.85* 0.26 0.21 0.65
Dividend cut/omission 0.11 0.27 2.58** 0.19 0.29 1.32 0.38 0.33 0.65
Sample size 57 240 51 219 39 149
Panel D: Non-CEO board dominance
Operational restructuring 0.73 0.54 2.89*** 0.50 0.46 0.62 0.64 0.48 1.93*
Cash generative actions 0.52 0.36 2.37** 0.53 0.42 1.68* 0.51 0.41 1.18
Acquisitions 0.61 0.47 2.00** 0.47 0.32 2.15** 0.26 0.28 0.28
Sample size 71 226 66 204 47 141
***, **, *
indicate significance at 1%, 5% and 10% respectively.
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 587
Table 22.7 Logistic regression of restructuring strategies on agency and control variables (year of decline)
Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. The dependent variable equals one if a
strategy is taken, and zero if otherwise. The sample consists of 297 declining firms over the period 1987 to 1993. The decline year refers to the year in
which a firm declines to the bottom 20% ranking in stock returns, in the market, after having been in the top 50% for two previous consecutive years.
Coefficients are tested for significance using the Wald test statistic.
Model: Restructuring strategy f (Debt, ownership, governance and control variables)
Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. The dependent variable equals one if a
strategy is taken, and zero if otherwise. The sample consists of 270 declining firms which are neither insolvent (bankrupt) nor acquired, one year post
decline, in the period 1987 to 1993. Coefficients are tested for significance using the Wald test statistic.
Model: Restructuring strategy f (Debt, ownership, governance and control variables)
Coefficients of the logistic regressions of restructuring strategies on agency and control variables are presented. The dependent variable equals one if a
strategy is taken, and zero if otherwise. The sample consists of 188 declining firms which are neither insolvent (bankrupt) nor acquired, two years post-
decline. The sample covers only firms in decline in the period 1987 to 1992, as firms declining in 1993 have only one year post-decline strategies to the
end of the analysis period i.e. December 1994. Sample size is therefore reduced. Coefficients are tested for significance using the Wald test statistic.
Model: Restructuring strategy f (Debt, ownership, governance and control variables)
suggesting delayed reaction to the onset The logit models of strategic choices
of decline. made in the third year of decline (decline
Unsurprisingly, lenders continue to press year 2) are shown in Table 22.9. In
for cash generative actions. They are also contrast to the model for the previous year
more likely to resort to debt restructuring. in table 8, only five of the third year models
Ownership continues to influence strategy are significant at least at the 10% level.
choices in the second year. Associated McFadden’s R2 ranges from 5% to 27%.
shareholding decreases the probability of The third year models thus generally have
operational restructuring and cash genera- less explanatory power than the models for
tive actions. Thus, management-associated the first two decline years. It appears that
shareholders’ resistance to these strategies the influence of the agency and control
in the decline year is reinforced in the second variables on strategy choices is waning.
year. Interestingly, associated shareholders Nevertheless, some of these variables
resist debt restructuring. Unassociated continue to exert significant impact.
shareholding begins to influence adoption Lenders continue to restrict capital
of strategies in the second year. They expenditure. Debt restructuring is again
are negatively related to the choice of made more likely by lenders. Associated
operational restructuring. CEO- duality shareholders stubbornly resist cash genera-
continues to increase the chances of capital tive strategies. Interestingly, unassociated
expenditure but, unsurprisingly, reduces the shareholders join associated shareholders
probability of managerial restructuring. in resisting cash generative strategies.
Non-executive Chairmen make cash As regards governance structure, dual
generative actions and debt restructuring CEOs are still more likely to go for capital
less likely but capital expenditure more expenditure. Non-executive Chairmen,
likely. More outside directors now means however, are quite inactive. With more
greater chances of cash generative actions outside directors a poorly performing firm
and managerial restructuring. Outside is now more likely to undertake not only
directors’ activism in the second year is more cash generative actions but acquisi-
in stark contrast to their passivity in the tions and capital expenditure too, perhaps
decline year. to expand the firm after two years of
The effects of economic downturn are restructuring.
equally significant in the second year. It The impact of external environment is
continues to increase the probability of still important in the third year. Economic
operational restructuring, dividend cut/ downturn still increases the probability of
omission and debt restructuring, but operational restructuring and dividend
reduces the probability of investments both cut/omission, and reduces the probability of
acquisitions and capital expenditure. acquisitions and capital expenditure. If the
Industry downturn makes operational industry is depressed in the third year,
and debt restructuring less likely and sample firms would be inclined to remove
dividend cut/omission more likely. Severity top management, cut/omit dividends and
of decline impacts further in the second restructure their debts. Operational
year. It continues to make operational restructuring, however, is less likely to be
restructuring, dividend cut/omission and needed. Sample firms are still constrained
debt restructuring more likely. In addition, in their strategy by the existence of an
the more severely declining firms are also internal cause of decline and severity of
less likely to undertake acquisitions. the initial decline. Firms with an internal
Internal cause of decline has little influence cause of decline are still more likely to
on strategy choice in the second year of restructure their operations. Severe decline
decline. Finally, large firms increase the firms are still more likely to cut/omit their
probability of investments in the form dividends. Firm size, however, ceases to
acquisitions and capital expenditure. have any impact in the third year.
592 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 22.10 Summary of the effect of each explanatory variable on the choice of restructuring
strategies
This table summarises the results in Tables 22.7–22.9 The multiple influences of each explanatory
variable on the probability of various restructuring actions occurring are highlighted. Variables that
are significantly positively/negatively related to particular strategies (i.e. increasing/decreasing the
probability of those actions occurring) in the logistic regression models in Tables 22.7–22.9, are
separately listed.
As explanatory variables collectively influence the choice of restructuring strategies, the combined
impact of explanatory variables on the choice of a specific restructuring strategy is summarised from
the results reported in Tables 22.7–22.9. Explanatory variables that are significantly positively/
negatively related to a specific strategy, in the logistic regression models in Tables 22.7–22.9,
increase/decrease the probability of that action occurring.
Explanatory variables
broadly confirm these tendencies. The The results show that short term and
question then is which type of lenders and unsecured lenders are the parties behind
owners favour which type of strategy. To lenders’ demand for operational restructur-
test for these individual impacts we rerun ing.10 Bank and short term lenders jointly
all the regressions in Tables 22.7–22.9 press for cash generative actions. However,
with three types of lenders instead of one, only bank lenders have the clout to restrict
short term lenders, bank lenders and firms’ capital expenditure.
unsecured lenders, and four types of Manager shareholders alone resist
shareholders instead of two, manager operational restructuring, cash generative
shareholders, associated block shareholders, and acquisition strategies. Conversely,
institutional block shareholders and associated block shareholders prefer
non-institutional unassociated block acquisitions and surprisingly, in a show
shareholders. of independence, managerial restructuring.
CORPORATE RESTRUCTURING IN RESPONSE TO PERFORMANCE DECLINE 595
Barclay, M. and Holderness, C. (1991) corporate ownership and control when firms
Negotiated block trades and corporate default, J. Financ. Econom. 27, 355–387.
control, J. Finance 46, 861–878. Gilson, S. C., John, K., and Lang, L. H. P. (1990)
Baysinger, B. D. and Hoskisson, R. E. (1990) The Troubled debt restructuring: An empirical
composition of boards of directors and strate- study of private reorganization of firms in
gic control: Effects on corporate strategy, default, J. Financ. Econom. 27, 315–353.
Academy of Managt Rev. 15, 72–87. Hermalin, B. and Weisbach, M. (1992) The
Bethel, J. E. and Liebeskind, J. (1993) The effects effects of board composition and direct
of ownership structure on corporate restructur- incentives on firm performance, Financ.
ing, Strategic Managt J. 14, 15–31. Managt 20, 101–112.
Boeker,W. and Goodstein, J. (1993) Performance Hill, C. W. L. and Snell, S. A. (1989) Effects of
and successor choice: The moderating effects ownership structure on corporate productivity,
of governance and ownership, Academy of Academy Managt J. 32, 25–46.
Managt J. 36, 172–186. Hirschey, M., Slovin, M. B., and Zaima, J. K.
Brickley, J. A., Lease, R. C., and Smith, C. W. (1990) Bank debt, insider trading and the
(1988) Ownership structure and voting return to corporate sell-offs, J. Banking and
on anti-takeover amendments, J. Financ. Finance 14, 85–98.
Econom. 20, 267–292. Holderness, C. G. and Sheehan, D. P. (1985)
Brown, D.T., James, C. M., and Mooradian, R. M. Raiders or saviours?: The Evidence on six
(1994) Asset sales by financially distressed controversial investors, J. Financ. Econom.
firms, J. Corporate Finance 1, 233–257. 14, 555–579.
The Cadbury report on the Financial Aspects of James, C. (1987) Some evidence of the unique-
Corporate Governance, Institute of Chartered ness of bank loans, J. Financ. Econom. 19,
Accountants, England and Wales, 1992. 217–235.
Choi, D. Toehold acquisitions, shareholder Jarrell, G. A. and Poulsen, A. B. (1987) Shark
wealth and the market for corporate control, repellents and stock prices: The effects of
J. Financ. Quantitative Anal. 26(3), Sep. anti-takeover amendments since 1980,
De Angelo, H. and De Angelo, L. (1990) J. Financ. Econom. 18, 127–168.
Dividend policy and financial distress: An Jensen, M. C. and Meckling, W. H. (1976)
empirical investigation of troubled NYSE Theory of the firm: Managerial behaviour,
firms, Journal of Finance 45, 1425–1431. agency costs and ownership structure,
De Bondt, W. F. M. and Thaler, R. H. (1985) J. Financ. Econom. October, 305–360.
Does the stock market overreact?, Journal of Jensen, M. (1989a) Active investors, LBO’s and
Finance 40, 793–805. privatisation of bankruptcy, J. Appl.
Demsetz, H. and Lehn, K. (1985) The structure Corporate Finance 2, 35–44.
of corporate ownership: Causes and conse- Jensen, M. (1989b) Eclipse of the public
quences, J. Polit. Econ. 93, 1155–1177. corporation, Harvard Bus. Rev. 61–74,
Diamond, D. (1983) Seniority and maturity Sep./Oct.
of debt contracts, J. Financ. Econom. 33, Lang, L., Poulsen, A., and Stulz, R. (1995)
341–368. Asset sales, firm performance, and the agency
Fama, E. F. and Jensen, M. C. (1993) costs of managerial discretion, J. Financ.
Separation of ownership and control, J. Law Econom. 37, 3–37.
Econom. 26, 301–325. Mallette, P. and Fowler, K. L. (1992) Effects of
Gertner, R. and Scharfstein, D. (1991) A theory board composition and stock ownership on
of workouts and the effects of reorganization the adoption of poison pills, Academy Managt
law, J. Finance XLVI(4), September, J. 35(5), 1010–1035.
1189–1222. Marris, R. (1964) The Economic Theory of
Gilson, S. C. (1989) Management turnover and Managerial Capitalism, Free Press, New York.
financial distress, J. Financ. Econom. 25, McConnell, J. J. and Servaes, H. (1990)
241–262. Additional evidence on equity ownership and
Gilson, S. C. (1990) Bankruptcy, boards, banks, corporate value, J. Financ. Econom. 27,
and bondholders – Evidence on changes in 595–612.
598 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Mikkleson, W. H. Ruback, R. S. (1985) An Schleifer, A. and Vishny, R. W. (1986) Large
empirical analysis of the interfirm equity shareholders and corporate control, J. Polit.
investment process, J. Financ. Econom. 14, Economy 94, 461–488.
523–553. Schipper, K. and Smith, A. (1986) A comparison
Murphy, K. J. and Zimmerman, J. L. (1993) of equity carve-outs and seasoned equity
‘Financial performance surrounding CEO offerings: Share price effects and corporate
turnover’, Journal of Accounting and restructuring, J. Financ. Econom. 15,
Economics 16, 273–315. 153–186.
Myers, S. C. (1977) Determinants of corporate Shivdasani, A. (1993) Board composition,
borrowing, J. Financ. Econom. November, ownership structure, and hostile takeovers,
147–176. J. Account. Econom. 16, 167–198.
Ofek, E. (1993) Capital structure and firm Slatter, S. (1984) Corporate Recovery:
response to poor performance: An empirical Successful turnaround strategies and their
analysis, J. Financ. Econom. 34, 3–30. implementation, Penguin.
Pound, J. (1988) Proxy contests and the Storey, in Keasey and Watson (1987) J. Banking
efficiency of shareholder oversight, J. Financ. and Finance 14(3), Autumn, 335–350.
Econom. 20, 237–265. Sudarsanam, S. (1995) Large shareholders,
Rajan, R. (1992) Insiders and outsiders: The corporate restructuring and target valua-
choice between informed and arm’s-length tion, Paper presented to Financial
debt, J. Finance 47, 1367–1400. Management Association Annual Meeting,
Rechner, P. L. and Dalton, D. R. (1989) The New York.
impact of CEO as board chairperson on Sudarsanam, S. (1996) Large shareholders,
corporate performance: Evidence vs. rhetoric, takeovers and target valuation, J. Busin.
Academy Managt Exec. 3, 141–143. Finance Account. 23(2), March.
Rosenstein, S. and Wyatt, J. H. (1990) Outside Weisbach, M. (1989) Outside directors and
directors, board independence, and shareholder CEO turnover, J. Financ. Econom. 20,
wealth, J. Financ. Econom. 26, 175–191. 431–460.
Chapter 23
Jorge Farinha*
DIVIDEND POLICY, CORPORATE
GOVERNANCE AND THE
MANAGERIAL ENTRENCHMENT
HYPOTHESIS: AN EMPIRICAL
ANALYSIS
Source: Journal of Business Finance & Accounting, 30(9–10) (2003):
1173–1210.
market monitoring of the firm’s activities assets to obtain personal benefits, such as
and performance. The reason is that higher shirking and perquisite consumption. As
dividend payouts increase the likelihood insider ownership increases, agency costs
that the firm will have to sell common may be reduced since managers bear a
stock in primary capital markets. This in larger share of these costs. However, as
turn leads to an investigation of manage- Demsetz (1983) and Fama and Jensen
ment by investment banks, securities (1983) point out, managers holding a
exchanges and capital suppliers. Studies by substantial portion of a firm’s equity may
Baghat (1986), Smith (1986), Hansen and have enough voting power to ensure that
Torregrosa (1992) and Jain and Kini their position inside the company is secure.
(1999) have recognised the importance of As a result, they may become to a great
monitoring by investment bankers in new extent insulated from external disciplining
equity issues. Recent theoretical work by forces such as the takeover threat or the
Fluck (1998), and Myers (2000) also managerial labour market. Stulz (1988)
presents agency-theoretic models of presents a model where high ownership by
dividend behaviour where managers pay managers can effectively preclude the
dividends in order to avoid disciplining possibility of a takeover, in accordance with
action by shareholders. an entrenchment hypothesis. Consistent
Additionally, Jensen (1986) sees expected, with this, Weston (1979) reports that firms
continuing dividend payments as helping where insiders held more than 30% have
to dissipate cash which might otherwise never been acquired in hostile takeovers.
have been wasted in non-value maximising Morck et al (1988) and McConnel and
projects, therefore reducing the extent of Servaes (1990) find an inverted U-shaped
overinvestment by managers. relationship between insider ownership and
In Rozeff’s (1982) model, an optimal firm performance in accordance with the
dividend policy is the outcome of a trade-off existence of managerial entrenchment
between equity agency costs and transac- above a critical level of ownership.
tion costs. Consistent with such trade-off This entrenchment hypothesis taken
model, Rozeff reports evidence of a strong from the agency literature is particularly
relationship between dividend payouts and interesting since it has consequences for
a set of variables proxying for agency dividend policy which are distinct from
and transaction costs in a large sample other competing theories of dividend
composed of one thousand US firms for the behaviour. Specifically, the prediction is
period 1974 to 1980. that below an entrenchment level insider
A cross-sectional analysis of dividend ownership and dividend policies can be seen
policy by Crutchley and Hansen (1989) as substitute corporate governance devices,
also shows results consistent with dividend therefore leading to a negative relationship
policy acting as a corporate monitoring between these two variables. After such
vehicle and with substitution effects critical entrenchment level, however,
between dividend payments and two other when insider ownership increases are
control mechanisms, managerial ownership associated with additional, entrenchment-
and leverage. related, agency costs, dividend policy
may become a compensating monitoring
(ii) The managerial entrenchment force and, accordingly, a positive relation-
hypothesis ship with insider ownership would be
observed. This prediction is a distinctive
Following Jensen and Meckling (1976), one given that signalling, tax clienteles or
when managers hold little equity and other competing theories for dividend
shareholders are too dispersed to take behaviour do not predict such U-shaped
action against non-value maximisation relationship between insider ownership
behaviour, insiders may deploy corporate and dividend policy.
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 601
Schooley and Barney (1994), using US is the notion that dividend policy is set up
data, document a U-shaped relationship optimally so as to minimise total agency
between dividend yield and CEO ownership. and transaction costs.1
That study suffers from several limitations. Additionally, given that the presence of
CEO ownership is not always the best other managerial monitoring devices is
measure of insider ownership as in frequent likely to affect the usage of any particular
cases board members other than the CEO mechanism for reducing agency costs
hold significant amounts of a firm’s equity. (Hart, 1995), allowance is made for the
Apart from using a relatively small sample presence of such variables when modelling
size (235 firms against our sample in excess dividend policy. In line with previous
of 600 firms) Schooley and Barney’s data research, it is possible that not just inverse
is confined to large firms with a small relationships, or substitution effects, exist
number of cases where the CEO holds between different governance mechanisms.
substantial holdings (the average CEO In fact, Rediker and Seth (1995) argue
ownership in their sample is 2.5% against that different control mechanisms may
our mean insider ownership of 16% (in be complementary, and not just alternative,
1991) and 13% (in 1996). Finally, they do instruments for corporate governance.
not control for alternative monitoring The other monitoring mechanisms consid-
mechanisms on managers which have been ered in the analysis are debt, analysts
recognised in the literature. following, institutional ownership, the
Also, no studies have analysed, in the presence of outsiders in the board and
context of dividend policy, the possibility compliance with Cadbury (1992) Code of
that beneficial and non-beneficial insider Best Practice.
holdings may be conducive to entrenchment. The formal basic model used, along with
Such possibility has, however, been suggested the variables employed, is described in
by previous research from Gordon and sub-section 3 (iii).
Pound (1990), Chang and Mayers (1992)
and Cole and Mehran (1998), who find (ii) Testable hypotheses
evidence that manager’s voting control
over Employees Stock Ownership Plans A particular concern of this paper is an
(ESOPs) can contribute to managerial attempt to distinguish the agency per-
entrenchment. spective of dividend policy from other
competing explanations by focusing on the
relationship between insider ownership and
3. RESEARCH DESIGN dividend payouts. Following the discussion
above, it is likely that managers controlling
(i) General model specification large holdings in the firm can to a signifi-
cant extent insulate themselves from other
The general model specified for this analysis disciplining mechanisms. If dividend policy
is represented by a single-equation cross- and insider ownership both perform a
sectional regression between dividend monitoring role, one might expect that
payouts and a set of variables related to before a critical level of entrenchment,
Rozeff’s (1982) trade-off argument. This dividend policy and insider ownership could
trade-off is between the marginal benefits be substitute monitoring mechanisms. As
of dividend payouts (a reduction in agency such, increasing insider ownership would
costs) and respective marginal costs (an be accompanied by decreasing dividend
increase in the so-called ‘transaction payouts. After a critical level of ownership,
costs’). ‘Transaction costs’ here relate to however, larger dividend payouts may be
the direct or indirect costs of external needed to compensate for entrenchment-
equity financing and potential tax costs related agency costs arising from larger
associated with dividend payouts. Implicit (above critical level) insider holdings.
602 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
This leads to the following testable for the U-shaped relationship, but also as a
hypothesis: test for the proposition that non-beneficial
holdings can be conducive to entrenchment
H1: The cross-sectional distribution of by insiders.
dividend payouts, all else constant, A final test is to analyse whether
is negatively related to beneficial compliance with Cadbury (1992) Code of
insider ownership below an entren- Best Practice affects dividend policy. Since
chment level of ownership, and the Code was published in 1992, this is
positively related above that level. done in the 1996 regression only. That
document reviewed the role of corporate
Hypothesis H1 can be modelled by using boards in corporate governance and pro-
a second-degree polynomial specification vided a set of recommendations of best
for the insider ownership variable and test- practice to enhance the accountability and
ing the expectation of a negative sign for monitoring function of the directors of UK
the first term and a positive one for the firms. After publication of the report, the
second. However, the positive sign predicted London Stock Exchange required listed
after the critical level of insider ownership firms to state their compliance, and rea-
could alternatively be explained by a man- sons for not complying, with the Code’s
agerial desire to diversify their wealth by prescriptions. The analysis of the relation-
increasing liquidity (Beck and Zorn, 1982) ship between dividend policy and Cadbury
when their shareholdings in the firm are (1992) compliance is a novel way of testing
large, rather than by an entrenchment the agency explanation for dividend policy
explanation.To control for this possibility, it given the Cadbury (1992) recognised role
is hypothesised that in a setting where in corporate governance in the UK.2 If
insiders control shares on non-beneficial dividend policy is a substitute (complemen-
terms at the same time that their beneficial tary) disciplining mechanism to compliance
stakes are too small to enable entrench- with the Cadbury (1992) Code, then a
ment, a similar U-shaped relationship will negative (positive) association between
still be observed between dividend payouts dividend payouts and Cadbury (1992)
and total (beneficial or not) holdings compliance should be apparent. Since none
with the turning point above the level of of these two alternative possibilities can be
beneficial holdings. This is because both ruled out, the null hypothesis is that there
categories of holdings can bring control is no association between Cadbury (1992)
and, therefore, potential entrenchment. In compliance and dividend payouts. The
other words, this hypothesis looks at the corresponding testable hypothesis is thus:
possibility that managers may use not just
beneficial but also non-beneficial holdings H3: Compliance with the Cadbury
to entrench themselves. This leads to the (1992) Code of Best Practice has
following testable hypothesis: a zero impact on dividend payouts,
all else constant.
H2: For low (below critical entrench-
ment point) levels of beneficial
insider ownership, dividend payouts The cross-sectional analyses in this
have a U-shaped relationship with paper are reported for 1991 and 1996.
total (beneficial and non-beneficial) Using two periods provides a robustness
insider ownership, all else constant, check on the results obtained for a single
as predicted in Hypothesis H1 for year. It also enables a control for possible
beneficial ownership, with a turning structural changes, particularly those poten-
point above the level of beneficial tially arising from the intense academic
holdings. and public debate on corporate governance
in the UK since the early nineties. A five-year
Hypothesis H2 can thus also be seen not interval between the two cross-sections
just as a test for an alternative explanation was considered because the construction
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 603
of some variables, notably the dividend choice of a five-year period balanced the
payout measure, required computations trade-off between the advantage of using a
over five years. longer period to provide a more accurate
measure of the long term dividend payout
(iii) Formal model and ratio, and the costs associated to the
survivorship bias problems arising from the
variables employed
requirement of longer series of data for
The empirical mode model used in the each firm in the sample.
analysis can be described as follows: Beneficial insider ownership, INSBEN, is
defined as the percentage of the company’s
MNPAYi 0 1INSBENi shares directly or indirectly controlled by
2
2INSBEN i the firm’s managers, their families or
3GROW1i family trusts (as disclosed in firm’s annual
4GROW2i 5DEBTi reports). Jensen and Meckling (1976)
6 VARIAB i 7CASHi
posit a negative relationship between
DISPERS insider equity ownership and agency costs
8 i
while Morck et al. (1988) and McConnel
9INSTITi
and Servaes (1990) present evidence
10NONEXPCTi consistent with that assertion. Within
11IACTi 12SIZEi certain ownership ranges, higher insider
13LANALYSTi ownership can reduce expected agency
14ROAi costs and hence dividend policy may
15ROAxDUMNEGi becomes less important as a monitoring
16CADBURYi vehicle. Therefore, the expected sign for the
n coefficient of INSBEN in the regression is
jINDUMMYj,i 1 negative. A square term for insider owner-
j1
ship is included, as discussed above, to
where: regression coefficients,
account for the possibility of managerial
i index of the ith firm, entrenchment, which translates into the
j sector index, expectation of a positive sign. This follows,
n number of sector dummies in particular, arguments and evidence by
(2-digit AIC codes), Morck et al (1988) and McConnel and
1 error term.
Servaes (1990) that the effect of insider
ownership in the reduction of agency costs
The dependent variable, MNPAY is the may change its sign after a certain critical
dividend payout ratio, constructed as a five- level of ownership.
year mean ratio of total ordinary annual The remaining variables are control
dividends declared (interim plus final) to factors that either (i) have been observed
after-tax earnings (before extraordinary in the literature to influence dividend
items).3 Similar to Rozeff (1982), a mean payments, (ii) can be seen as alternative
payout ratio is preferred to annual payout or complementary managerial monitoring
figures, to reduce the effects of transitory vehicles or (iii) can proxy for the presence
and noisy components in short-term earn- of potential agency problems.
ings. Thus, the focus is on a measure Past growth (GROW1), defined as the
of long-term dividend payout, given the geometric mean rate of growth of the
evidence, from a series of studies dating firm’s total assets for the last five years, is
back to Lintner (1956), that firms typically included on the grounds that higher historic
stabilise dividends around a long-term growth may render dividend policy less
payout objective.4 Observations with mean relevant for inducing primary market
dividend payout ratios in excess of one or monitoring vehicle given the likelihood that
negative are excluded due to the lack of growth may already be inducing external
economic significance of these values. The fund raising (and associated monitoring).
604 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Hence, a negative sign is expected.5 A expected sign for its regression coefficient
similar argument applies to GROW2, a is positive.
variable proxying for future growth oppor- Shareholder dispersion (DISPERS),
tunities, measured as the ratio of market represents a measure of stock ownership
to book value of equity). Consistent with diffusion. This variable is defined as 100%
these assertions, Rozeff (1982) reports a minus the accumulated sum of the owner-
negative association between dividend ship by individual entities with more than
payouts and variables proxying for past or 3% of the firm’s stock in 1991 or 1996.
future growth opportunities. The existence of a large number of (small)
The inclusion of DEBT, the book value of shareholders (or a low level of ownership
total debt deflated by the market value concentration) increases the potential
of equity is mainly motivated by its poten- agency costs given the free-rider problem
tial monitoring role on managers. In associated with higher ownership diffusion.
particular, financial leverage has been The predicted sign for the coefficient of
argued by Jensen and Meckling (1976), DISPERS is thus positive.
Jensen (1986) and Stulz (1988), among INSTIT measures total institutional
others, to play a role in reducing agency blockholder ownership of the firm’s shares.
costs arising from the shareholder-manager Institutional blockholders may act as a
conflict. Debt may also have an impact on monitoring device on the firm’s managers,
dividends because of debt covenants and as argued by Demsetz and Lehn (1985)
related restrictions imposed by debtholders.6 and Schleifer and Vishny (1986), thus
Thus a negative sign is expected. dampening in principle the need for high
The total variance of a firm’s stock dividend payouts. However, it is possible
returns, VARIAB, is also included in the that institutions may influence higher divi-
analysis. High fixed operating costs or dend payouts by a company to enhance
business risk may affect the firm’s dividend managerial monitoring by external capital
payout, all else constant, to the extent that markets, namely if they believe their own
these will increase the frequency of costly direct monitoring efforts to be insufficient
additional external financing. This is due to or too costly. In this case a complemen-
the greater variability in earnings and tarity between these alternative governance
funding needs that high operating leverage mechanisms could be apparent. Thus, the
or business risk may induce in a firm. The expected sign for this coefficient may
same reasoning applies to interest charges, be positive or negative.
which are characterised by Rozeff (1982) Following Winter (1977), Fama (1980)
as ‘quasi-fixed costs’. Both these operating and Weisbach (1988), the percentage of
and financial risks should translate into a non-executives on the firm’s board, NON-
high total risk (or variance) of the firm’s EXPCT, is also included to account for
stock returns. In addition, as observed by the possibility that such outside directors
Holder et al. (1998), transaction costs of may act as management monitors.Thus, the
new issues in the form of underwriting fees expected sign for this coefficient is negative,
are usually larger for riskier firms. The unless the same observations referred
expected sign of the coefficient of VARIAB about INSTIT apply, in which case a
in the regression is thus negative. positive relationship might emerge.
To the extent that high figures of CASH, IACT is a control variable defined as the
defined as a five-year average of cash and cumulative 5-year sum (1987–91 or
cash equivalents as a percentage of a firm’s 1992–96) of the amounts shown in a
assets represent, or are correlated with, firm’s accounts as irrecoverable Advance
a firm’s free cash-flow in the Jensen Corporate Taxation (ACT), deflated by
(1986) sense and associated agency costs, total assets.The usage of a cumulative sum
expected dividend payouts will be higher. instead of single-year figures is mainly to
Thus, greater payouts might be associated account for the often observed situation of
with higher figures of CASH and so the a firm declaring in one year surplus ACT
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 605
which eventually is written off in following Jensen et al. (1992) find evidence of a
years. If a company has had significant positive association between return on
irrecoverable ACT in the past then it is assets and dividend payouts. To the extent,
likely that this should translate into a however, that there are links between past
higher perceived cost of paying dividends7 profitability and current or expected
(or dividend ‘transaction costs’). A negative growth, such measures of profitability may
sign is thus expected for this variable’s have a different impact on payouts. For
coefficient. instance, past profitability may capture
The control variable firm size (SIZE) is information on growth prospects missed by
defined as the log of market capitalisation. other variables (namely GROW2), possibly
Size may be an important factor not just as because more profitable firms may be more
a proxy for agency costs (which can be (or less) likely to grow in the future. In
expected to be higher in larger firms) but addition, higher profitability may be evi-
also because transaction costs associated dence that agency problems are not very
with the issue of securities are also (nega- relevant so that monitoring mechanisms
tively) related to firm size as documented, such as dividend policy are less needed.
among others, by Smith (1977). However, Therefore, the sign for this control variable
as Smith and Watts (1992) point out, the can either be negative or positive. A
theoretical basis for an impact of size on problem arises, however, because firms may
dividend policy is not strong, and indeed face constraints to pay dividends when
some negative relationships have been their earnings are negative. DeAngelo and
observed (Allen and Michaely, 1995; and DeAngelo (1990) and DeAngelo et al.
Keim, 1985). Therefore, the inclusion of (1992) document that a significant pro-
size may be best regarded as a simple portion of firms having losses over a 5 year
control variable, with no particular sign period tend to omit their dividends entirely.
expectation. Similarly, Baker (1989) finds that an
LANALYST is the log of the number important reason cited by firms for not
of analysts (ANALYSTS) following a partic- paying dividends is ‘poor earnings’.
ular firm (as taken from I/B/E/S). Former Therefore, small or zero dividend payouts
research suggests that financial analysts could reflect not high levels of alternative
may constitute a source of managerial monitoring mechanisms included in
monitoring. Specifically, Moyer, Chatfield equation (1) but simply be the result of
and Sisneros (1989) and Chung and Jo negative earnings. Given such possible non-
(1996) present evidence consistent with linearities, ROA is included in the analysis
the number of financial analysts following along with a dummy (DUMNEG) account-
a firm having a negative impact on agency ing for the existence of any negative
costs. The expected sign for the impact of earnings during the period used for the
analysts following can either be positive or calculation of ROA, as well as an interac-
negative, in accordance with the Rediker tive term between such dummy and the
and Seth (1995) argument. A logarithmic ROA measure.
transformation is used because it is likely The final independent variable is
that the impact of an additional analyst CADBURY, which consists of a dummy
may become smaller as the number of term taking the value of 1 if the firm states
analysts following a firm increases. its full compliance, in the 1996 regression,
Return on assets, ROA, is defined as the with the Cadbury (1992) Code of Best
mean ratio between after-tax earnings Practice. Since the purpose for the intro-
before extraordinary items and total assets duction of this Code was the improvement
calculated over a 5-year period (1987–91 of firm’s corporate governance practices,
or 1992–96). In general accordance with a an impact may be expected on firm’s
signalling perspective (Miller and Rock, dividend payouts. The sign of this impact
1985), dividend payouts may be positively is, however, unclear, as discussed in the
related with measures of profitability. statement of Hypothesis H3.
606 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
1991 1996
payouts, either for 1991 or 1996. In should be mentioned that modelling the
economic terms, an increase in dispersion impact of profitability with a ROA variable,
of 10 p.p. increases on average the dividend a dummy for negative earnings and an
payout ratio in about 1 p.p. interactive term provides a significantly
Other variables included in the regres- better fit, either for 1991 or 1996, but
sions are, in general terms, either signed as particularly so in 1991 (where adjusted
expected or insignificant. An exception to R-square increases from 20.64% in model
this is the Irrecoverable Advance Corporate 3 to 32.93% in model 4).
Taxes variable, where its transaction cost Consistent with Allen and Michaely
role for dividends generated the expecta- (1995) and Keim (1985), a significant
tion of a positive sign but the results yield negative relationship between firm size and
a positive coefficient (in the 1991 regres- dividend payouts is observed for 1991 and
sion only). Such result is, however, consis- 1996. An interesting result also is that full
tent with Adedeji (1998), who finds a compliance with the Cadbury (1992)
similar association between irrecoverable Report has a positive impact on dividend
ACT and dividend payouts and interprets payouts (with a p-value close to 1%) as
that evidence as related to firms seeing well as economic significance (compliance
irrecoverable ACT as a tax allowance that with Cadbury increases payouts in around
can enhance distributable earnings. It also 4% of earnings).8 This result is consistent
Table 23.2 Summary Descriptive Statistics
Panel A: 1996
Sample (N 609)
MNPAY 0.373 0.228 0.999 0.517 0.379 0.205 0.000 0.295 0.305
INSBEN 10.804 15.747 75.243 14.000 3.585 0.347 0.001 1.848 2.692
INSNONB 2.333 7.163 67.489 0.813 0.000 0.000 0.000 5.050 30.115
INSIDER 13.137 17.861 82.861 18.916 4.410 0.417 0.001 1.574 1.565
GROW1 0.108 0.188 1.921 0.147 0.078 0.018 0.374 3.575 25.633
GROW2 1.820 1.217 12.664 2.050 1.515 1.123 0.555 3.742 21.642
DEBT 0.819 11.233 276.412 0.354 0.161 0.057 0.000 24.376 598.737
VARIAB 37.036 16.394 110.000 46.000 32.000 25.000 16.000 1.264 1.577
CASH 11.102 10.981 70.209 15.294 8.025 3.155 0.000 1.886 4.557
DISPERS 55.273 21.348 99.985 71.247 54.175 38.840 5.838 0.062 0.765
INSTIT 23.665 16.767 78.770 35.890 21.300 9.990 0.000 0.483 0.567
DIR 7.433 2.531 25.000 9.000 7.000 6.000 2.000 1.447 4.585
NONEX 3.174 1.550 10.000 4.000 3.000 2.000 0.000 0.972 1.510
NONEXPCT 0.424 0.144 0.857 0.500 0.429 0.333 0.000 0.048 0.260
IACT 0.120 1.119 22.722 0.000 0.000 0.000 3.661 14.081 276.736
MKCAP 720.350 3,055.271 39,544.000 336.000 87.000 24.000 0.310 9.679 105.307
SEQ 228.586 953.304 14,887.000 125.400 33.852 11.411 1,034.400 10.744 143.768
AT 559.758 1,961.070 32,572.000 298.800 81.517 31.730 0.562 9.635 129.445
ANALYSTS 5.814 5.945 27.000 9.000 3.000 1.000 0.000 1.086 0.167
608 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
LANALYST 1.509 0.948 3.332 2.303 1.386 0.693 0.000 0.059 1.085
SIZE1 4.560 1.933 10.585 5.817 4.466 3.178 1.171 0.268 0.072
SIZE2 4.646 1.707 10.391 5.700 4.401 3.457 0.576 0.411 0.246
ROA 3.844 12.960 31.414 8.424 5.643 2.199 217.692 9.650 146.947
DUMNEG 0.388 0.488 1.000 1.000 0.000 0.000 0.000 0.463 1.792
CLOSE 0.084 0.277 1.000 0.000 0.000 0.000 0.000 3.013 7.101
CADBURY 0.486 0.500 1.000 1.000 0.000 0.000 0.000 0.056 2.003
Panel B: 1991
Sample (N 693)
MNPAY 0.375 0.194 0.997 0.481 0.355 0.247 0.000 0.537 0.440
INSBEN 14.412 19.487 79.570 21.370 5.290 0.410 0.000 1.511 1.179
INSNONB 2.166 5.848 50.910 1.150 0.000 0.000 0.000 4.464 24.944
INSIDER 16.578 20.656 79.570 26.640 6.760 0.530 0.000 1.287 0.504
GROW1 0.201 0.218 2.202 0.251 0.152 0.083 0.260 3.220 18.224
GROW2 1.421 0.790 9.123 1.622 1.252 0.957 0.420 3.792 24.175
DEBT 0.987 8.247 209.378 0.542 0.248 0.091 0.000 23.611 592.204
VARIAB 40.329 12.973 92.000 46.000 38.000 31.000 13.000 1.111 1.457
CASH 10.022 10.373 70.994 13.282 6.973 2.825 0.000 2.073 5.704
DISPERS 50.361 22.333 99.900 65.810 49.160 33.000 0.470 0.163 0.772
INSTIT 22.667 16.415 90.430 32.900 20.360 9.600 0.000 0.746 0.350
DIR 7.339 2.613 21.000 9.000 7.000 6.000 3.000 1.188 2.594
NONEX 2.657 1.716 9.000 4.000 2.000 1.000 0.000 0.803 0.710
NONEXPCT 0.349 0.170 0.800 0.462 0.375 0.250 0.000 0.176 0.275
IACT 0.093 0.826 6.926 0.000 0.000 0.000 5.135 1.519 23.079
MKCAP 399.560 1,626.134 25,638.000 176.000 46.000 15.000 0.540 9.487 114.211
SEQ 199.929 757.814 11,561.000 96.790 30.442 10.886 455.883 9.968 128.868
AT 470.385 1,740.742 31,792.000 218.600 68.556 25.141 1.356 11.042 167.806
ANALYSTS 5.023 5.501 21.000 8.000 3.000 1.000 0.000 1.019 0.192
LANALYST 1.336 0.999 3.091 2.197 1.386 0.693 0.000 0.055 1.327
SIZE1 3.988 1.882 10.152 5.170 3.829 2.708 0.616 0.374 0.007
SIZE2 4.423 1.714 10.367 5.387 4.228 3.225 0.305 0.510 0.112
ROA 6.472 6.949 24.885 9.520 6.749 4.362 49.127 3.282 21.119
DUMNEG 0.227 0.419 1.000 0.000 0.000 0.000 0.000 1.309 -0.286
CLOSE 0.121 0.327 1.000 0.000 0.000 0.000 0.000 2.326 3.421
Notes: MNPAY 5-year mean of the ratio of interim plus final ordinary dividends divided by after-tax earnings before extraordinary items; INSBEN Percentage of the
firms’s shares controlled beneficially by board directors; INSNONB Percentage of the firm’s shares controlled on non-beneficial terms by board directors; INSIDER Sum
of INSBEN and INSNONB; GROW1 5-year geometric mean rate of growth in total assets; GROW2 market to book value, defined as market capitalisation of equity
plus book value of assets minus book value of equity, divided by book value of total assets; DEBT Total debt deflated by market capitalisation; VARIAB 5-year volatil-
ity of stock returns; CASH: 5-year mean of the ratio of cash plus cash equivalents deflated by total assets; DISPERS Percentage of the firm’s shares owned collectively
by entities (non-insiders) with less than 3% individual stakes; INSTIT Percentage of firm’s shares owned collectively by institutions with 3% or more of the firm’s stock;
DIR Number of directors in the board; NONEX Number of external directors in the Board; NONEXPCT Percentage of external directors on the board; IACT Sum
of consecutive five years of irrecoverable advance tax deflated by total assets; MKCAP Market capitalisation of a firms’s equity as of 31 December; SEQ Book value of
equity; AT Book value of total assets; ANALYSTS; Number of one year ahead earnings forecasts by analysts; LANALYST Natural log of ANALYSTS; SIZE1 Natural
log of the firms’s market capitalisation; SIZE2: Natural log of the book value of total assets; ROA: 5-year mean return on assets; DUMNEG: Dummy of 1 if at least some
of the earnings are negative in the 5-year period and 0 otherwise; CLOSE: Dummy taking the value of 1 if the company has a close company status, and zero otherwise;
CADBURY: Dummy taking the value of 1 of the company complies in full with the Cadbury (1992) Code of Best Practice, and zero otherwise.
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 609
Table 23.3 Beneficial and Non-beneficial Insider Ownership Statistics
with the idea that firms with better internal disciplinary force could eventually be
corporate governance rules are also those possible with a smaller share ownership by
that use dividend payouts more intensely, insiders. Accordingly, and using a proce-
suggesting that these two monitoring forces dure employed by Peasnell et al. (1998), a
act as complements rather than substi- dummy variable was created taking the
tutes. This is similar in spirit to findings by value of 1 if a company’s measure of size is
Laporta et al. (2000), who observe that in above the sample median and zero other-
countries where investor protection is wise.This binary variable was then made to
greater, dividend payouts tend to be higher interact with the insider ownership vari-
as well, suggesting that the legal environ- ables as for 1991 and 1996. Results are
ment and dividend policy may complement reported in Table 23.5 (Panel A). Contrary
each other in terms of their disciplining to the hypothesis that critical entrench-
effects on managers. Thus, Hypothesis H3 ment levels vary according to firm size,
of no impact of compliance with Cadbury small and large firms have virtually identi-
on dividend policy can be rejected. cal estimated critical entrenchment levels
in 1991 (around 32%). In 1996, however,
it can be observed that the U-shaped rela-
(iii) Critical entrenchment levels tionship between dividend payouts and
From the results in Table 23.4, critical insider ownership is confined to larger
entrenchment levels can be derived as the firms (above the size median), where it is
turning points in the U-shaped relationship significant at the 1% level, with an esti-
between dividend payouts and beneficial mated entrenchment level almost identical
insider ownership. The estimated critical to that of 1991. The number of firms in
entrenchment levels for beneficial insider 1996 where insider ownership levels are
ownership are approximately 32% in 1991 higher than 32% is 66 (about 11% of the
(model 4) and 25% in 1996 (model 5). sample). Thus, results suggest that an
These numbers are intuitively plausible and entrenchment level slightly above 30% is a
in line with Weston’s (1979) observation of consistent feature for all firms in 1991 and
no hostile takeovers occurring in firms large firms in 1996. From the findings
where insiders hold 30% or more of the above it can be inferred that either some
equity. structural change affected the U-shaped
An important question that could be relationship between insider ownership and
asked is whether the number of firms above dividend policy for small firms or that some
the estimated critical entrenchment level is empirical problems might be affecting the
sufficiently significant to make the results significance of the estimated coefficients
reliable. In 1991, 120 firms have beneficial for small firms in 1996.
insider ownership in excess of the estimated To analyse this issue in a somewhat dif-
critical level of 32%, which corresponds to ferent way, the sample was split in two
about 17% of the firms in the sample. In according to size. Regressions were then
1996, 92 firms (15% of the sample) have re-run separately for firms below and above
ownership above the 25% threshold. the size median in 1991 and 1996. Results
Overall, this suggests that the estimated are reported in Panel B of Table 23.5 and
turning points are driven by a non-negligible it can be seen they are very close to those
number of observations. presented in Panel A.
The hypothesis that size may affect the
critical level of entrenchment was investi- (iv) Entrenchment versus liquidity
gated as one might expect that in larger
firms less ownership would be needed to
hypothesis
achieve entrenchment. This could happen, A possible alternative explanation to the
for instance, because larger firms may be entrenchment hypothesis for the U-shaped
more difficult to acquire by means of hos- relationship between dividend payouts is
tile takeovers, so that insulation from such the possibility of liquidity motivations.
Table 23.4 OLS Regression Results: All Firms Dependent Variable: Dividend Payout Ratio (MNPAY)
1991 1996
(1) (2) (3) (4) (1) (2) (3) (4) (5)
Expected Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient Coefficient
Variable Sign p-value p-value p-value p-value p-value p-value p-value p-value p-value
*** *** *** *** *** *** *** *** ***
INTERCEPT 0.5485 0.5893 0.5362 0.6052 0.6433 0.6942 0.6747 0.7033 0.7129
0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
*** *** *** * * ** **
INSBEN () 0.0003 0.0041 0.0044 0.0049 0.0004 0.0031 0.0031 0.0042 0.0039
0.5726 0.0019 0.0009 0.0001 0.5445 0.0926 0.0930 0.0107 0.0157
*** *** *** ** ** *** ***
INSBEN2 () 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001 0.0001
0.0018 0.0012 0.0001 0.0368 0.0404 0.0028 0.0045
*** *** *** *** ** **
GROW1 () 0.1318 0.1182 0.1271 0.1443 0.0316 0.0190 0.0229 0.0859 0.0780
0.0001 0.0005 0.0002 0.0001 0.4792 0.6729 0.6092 0.0354 0.0560
** ** ***
GROW2 ()/() 0.0252 0.0237 0.0273 0.0046 0.0027 0.0036 0.0024 0.0097 0.0104
0.0169 0.0240 0.0093 0.6801 0.7366 0.6545 0.7638 0.2237 0.1923
DEBT () 0.0010 -0.0011 0.0010 0.0006 0.0001 0.0000 0.0001 0.0003 0.0002
0.2571 0.1878 0.2347 0.4532 0.8703 0.9629 0.9069 0.6390 0.7156
*** *** *** * *** *** *** *** ***
VARIAB ()
612 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
F test for signif. of ind.dummies 2.7495 2.7519 2.2381 2.2657 1.6349 1.7009 1.5779 2.5958 2.6132
(p-value) 0.0001 0.0001 0.0010 0.0009 0.0344 0.0244 0.0460 0.0001 0.0001
Estimated critical
entrenchment level – 33.14% 34.26% 31.99% – 23.93% 24.34% 25.37% 25.25%
(No. Firms above
critical level) (111) (107) (120) (98) (98) (90) (92)
Notes: MNPAY 5-year mean of the ratio of interim plus final ordinary dividends divided by after-tax earnings before extraordinary items; INTERCEP: Intercept term; INSBEN Percentage of
the firms’s shares controlled on beneficial terms by board directors; GROW1 5-year geometric mean rate of growth in total assets; GROW2 market to book value, defined as market capitalisa-
tion of equity plus book value of assets minus book value of equity, divided by book value of total assets; DEBT Total debt deflated by market capitalisation; VARIAB 5-year volatility of stock
returns; CASH: 5-year mean of the ratio of cash plus cash equivalents deflated by total assets; DISPERS Percentage of the firm’s shares owned collectively by entities (non-insiders) with less than
3% individual stakes; INSTIT Percentage of firm’s shares owned collectively by institutions with 3% or more of the firm’s stock; NONEXPCT Percentage of external directors on the board;
IACT Sum of consecutive 5-years of irrecoverable advance tax deflated by total assets; LANALYST Natural log of ANALYSTS; SIZE Natural log of the firms’s market capitalisation; ROA:
5-year mean return on assets; DUMNEG: Dummy of 1 if at least some of the earnings are negative in the 5-year period and 0 otherwise; CADBURY: Dummy taking the value of 1 of the company
complies in full with the Cadbury (1992) Code of Best Practice, and zero otherwise. The critical entrenchment level (last line) is computed as the turning pont where the relationship between the
dividend payout ratio (MNPAY) and beneficial insider ownership changes from negative to positive, as implied by the estimated coefficients for INSBEN and INSBEN2. *, ** and *** indicate
two-tailed significance at the 10%, 5% and 1% levels, respectively.
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 613
Table 23.5 Critical Entrenchment Levels: Small (below median size) versus Large (above median size) Firms
Panel A: estimates from the inclusion of an interactive term between a size dummy and beneficial insider ownership (INSBEN) variables
Panel B: estimates from restricting the regression to either large or small firms
Panel A
All firms 31.99% 0.0001 0.0001 25.25% 0.0157 0.0045
Small 32.03% 0.0017 0.0011 16.48% 0.4688 0.1989
Large 31.45% 0.0014 0.0044 32.08% 0.0003 0.0181
Panel B
Small 30.92% 0.0142 0.0062 20.44% 0.1707 0.0406
Large 31.07% 0.0030 0.0037 33.40% 0.0072 0.0251
where STD standard deviation, and STD(i j) VAR(i) VAR(j) 2 COV(ij)
VAR variance
COV covariance
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 615
In firms where insider holdings are relatively without the contribution of non-beneficial
high, managers could be tempted to increase holdings. The number of firms for which, in
dividend payouts to obtain liquidity in order regressions 3 and 4, total insider ownership
to diversify their personal wealth by invest- exceeds the critical turning is around 3%
ing elsewhere the cash received without of the sample in either 1991 (16 firms) or
reducing their share of the firm. However, 1996 (18 firms). Although the number
this liquidity argument would lose most of observations above the turning points is
of its power if one could demonstrate a not large, the consistent results across the
similar upward swing driven by shareholdings two cross-sections and the significance
that were controlled (in terms of voting of the coefficients lend some support to
power), but not beneficially owned, by Hypothesis H2, i.e., the proposition that
insiders. This is the basis for the statement liquidity is not behind the upward
of Hypothesis H2. movement in dividend policy after the
To test for this hypothesis, models 4 in turning points.
1991 and 5 in 1996 were re-run with the The results above also offer an interesting
restriction that beneficial insider ownership insight on the role of non-beneficial insider
is below the estimated entrenchment level ownership that has been little addressed in
and by substituting beneficial insider the literature. Specifically, they show that
ownership with total (beneficial and non- non-beneficial holdings where insiders can
beneficial) holdings by directors. If again a control voting rights (but not cash-flow
U-shaped relationship is observed with the rights) can be used as an entrenchment
turning point above the maximum level tool, along with their beneficial holdings.
allowed for beneficial holdings, one would Such findings are consistent with Gordon
conclude that the upward swing in the and Pound’s (1990) and Cole and Mehran’s
dividend/insider ownership curve can only (1998) results that managers can use their
be driven with the contribution of non- voting control over ESOPs (Employee
beneficial holdings. This would then Stock Ownership Plans) as a management
contradict the liquidity hypothesis. entrenchment device against takeovers.
Table 23.6 reports the results. They are also in accordance with Chang
Regressions from Table 23.4 are re-run and Mayers’ (1992) finding that the usage
with the substitution of beneficial insider of ESOPs is especially prone to provoke
ownership for total (beneficial and non- entrenchment when insiders already have
beneficial) holdings by insiders. The results substantial voting rights. However, our
in model 1 (1991) and 2 (1996) show that results are more general than these given
the U-shaped relationship between dividend that our data on non-beneficial holdings
payouts and insider ownership is still includes holdings owned not just by ESOPs
strong for the new definition of insider but also by a number of other entities (like
ownership. The new critical entrenchment charity trusts, founder trusts and company
points are now 36.43% and 30.47% for pension funds).
1991 and 1996, respectively. Next, in regres- Since our results on the liquidity hypoth-
sions 3 and 4, the calculations are repeated esis, although consistent along the two
with the exclusion of firms for which benefi- cross-sections, rely on a relatively small
cial ownership by insiders (INSBEN) is number of observations above the turning
below estimated entrenchment levels (given points, an additional test was made to test
that these levels are estimated with a the robustness of the findings. Specifically,
degree of error, the restriction on INSBEN if diversification driven-liquidity needs are
was arbitrarily set at 1.5 p.p. below the deemed to increase dividend payouts when
estimated entrenchment levels).9 One can holdings by insiders are large, then one
see that the U-shaped relation is still should expect that the larger the market
apparent, with the crucial difference that value of insider holdings, the larger dividend
the upward movement in the dividend-insider payouts should be, all else constant.
ownership curve cannot now be driven Therefore, a positive relationship between
Table 23.6 OLS Regression Results with Insider Ownership Defined as Total (beneficial and non-beneficial) (models 1 and 2) and with the Restriction
that INSBEN Turning Points (models 3 and 4) Dependent Variable: MNPAY
Estimated critical
entrenchment level 36.43% 30.47% 37.90% 33.76%
(No. Firms above
critical level) (124) (99) (16) (18)
Notes: MNPAY 5-year mean of the ratio of interim plus final ordinary dividends divided by after-tax earnings before extraordinary items; INTERCEP: Intercept term;
INSIDER Percentage of the firms’s shares controlled on beneficial and non-beneficial terms by board directors; GROW1 5-year geometric mean rate of growth in total
assets; GROW2 market to book value, defined as market capitalisation of equity plus book value of assets minus book value of equity, divided by book value of total assets;
DEBT Total debt deflated by market capitalisation; VARIAB 5-year volatility of stock returns; CASH: 5-year mean of the ratio of cash plus cash equivalents deflated
by total assets; DISPERS Percentage of the firm’s shares owned collectively by entities (non-insiders) with less than 3% individual stakes; INSTIT Percentage of firm’s
shares owned collectively by institutions with 3% or more of the firm’s stock; NONEXPCT Percentage of external directors on the board; IACT Sum of consecutive five
years of irrecoverable advance tax deflated by total assets; LANALYST Natural log of ANALYSTS; SIZE Natural log of the firms’s market capitalisation; ROA: 5-year
mean return on assets; DUMNEG: Dummy of 1 if at least some of the earnings are negative in the 5-year period and 0 otherwise; CADBURY: Dummy taking the value of 1
of the company complies in full with the Cadbury (1992) Code of Best Practice, and zero otherwise. The critical entrenchment level (last line) is computed as the turning
point where the relationship between the dividend payout ratio (MNPAY) and insider ownership changes from negative to positive, as implied by the estimated coefficients for
INSIDER and INSIDER2.
* **
, and *** indicate two-tailed significance at the 10%, 5% and 1% levels, respectively.
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 617
618 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
dividend payouts and the market value of entered in the regressions and this
insider holdings should emerge. variable, when significant, is negatively,
Accordingly, regressions were re-run not positively, signed.
with the inclusion of MKVINS, a variable
constructed as the product between SIZE
(v) Alternative specifications
(the market value of the company) and
INSBEN, while keeping in the regression The general reasoning behind the way the
the ownership variables INSBEN and variables were defined above was to con-
INSBEN2. The null hypothesis to test is sider contemporaneous variables (at 1991
whether MKVINS enters the regression or 1996) in both sides of equation (1).
with a significant positive slope. Also, if However, in the definition of the dividend
the documented positive coefficient of payout and cash variables, the annual
INSBEN2 is due to liquidity rather than figures were observed to be remarkably
entrenchment, one would expect that the unstable so in those cases a mean was
inclusion of MKVINS would alter the judged, as referred above, to be the best
significance of INSBEN2. Unreported estimate for the value of each of those
results show that MKVINS has either an variables in 1991 or 1996. Given that the
insignificantly different from zero coefficient dividend payout variable was thus defined
(in 1991) or a significantly negative one (in as a mean over a five-year period (1987–91
1996), and in this last case the slope of or 1992–96), while some of the variables
INSBEN is no longer significant (most (e.g., debt, the number of analysts or own-
likely as a result of the correlation between ership variables) relate to either 1991 or
MKVINS and INSBEN). As for INSBEN2, 1996, a degree of look-ahead bias could
its slope remains positive and significant occur, although the direction of such poten-
throughout. Also, an attempt was made to tial bias is undetermined. However, when
see if results would change with the alternative definitions of dividend payout
omission of INSBEN2, so as to analyse if were attempted (defining it, in the 1991
potential correlation problems between cross-section, as either the mean over the
INSBEN2 and MKVINS are affecting the 1989–93 or the 1991–95 period), the
significance of MKVINS. Results show, results were similar so any possible look-
however, that MKVINS remains insignifi- ahead bias was dismissed as not serious.
cantly different from zero in 1991 and Several other robustness checks were
significantly negative in 1996.The slope on made to see if the conclusions above were
INSBEN in 1996 becomes significantly sensitive to the usage of other specifica-
positive but this is most likely biased due to tions or when considering other potentially
the misspecification arising from the omis- relevant factors. Regressions were thus
sion of INSBEN2. One can thus conclude repeated by excluding firms having close
that, once again, the liquidity hypothesis company tax status10 which could affect
stated above is contradicted by the data. the relative transaction costs of dividends.
To summarise, our tests offer some Results were, however, unchanged.
support for the notion that liquidity is A piecewise linear regression was also
not behind the U-shaped relationship performed allowing for one turning point in
documented between dividend payouts and the vicinity of the estimated critical
insider ownership. Results show in fact that entrenchment levels in Table 23.4. The
the same relationship can be observed for a hypothesis of changing slopes under this
restricted sample where non-beneficial new specification was confirmed with
holdings are essential to achieve total significance levels and R-squares very close
insider holdings above the critical turning to those under the quadratic specification.
points. Secondly, the quadratic term for The usage of switching regimes to analyse
INSBEN2 remains significantly positive alternative critical entrenchment points
when a variable controlling for the market under the piecewise linear regression did
value of insider holdings (MKVINS) is not produce results that could contradict
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 619
the existence of a negative slope followed year periods (1987–91 and 1992–96) and
by a positive one after a critical level. Such a considerably large sample (in excess of
alternative specifications did not yield, 600 firms), it tests the hypothesis that
however, higher R-squares than the insider ownership affects dividend policies
quadratic specification used before. Since in a manner consistent with a managerial
it is reasonable to think that in general entrenchment perspective, drawn from the
terms a piecewise linear regression imposes agency literature.
a much stricter structure than the quadratic In line with predictions, and controlling
specification used above, our preference for other factors, strong evidence is found
goes to this one. that after a critical entrenchment level of
Collinearity diagnostics prescribed by insider ownership estimated in the region of
Belsley et al. (1980) and VIF (variance 30%, the coefficient on insider ownership
inflation factors) analysis were used to changes from negative to positive.
observe if multicollinearity problems The hypothesis that liquidity needs on
could be obscuring some of the results. the part of insiders are responsible for the
The conclusion was that the statistically positive association between dividend pay-
insignificant variables in Table 23.4 were outs and insider ownership after the critical
not significantly affected by collinearity turning point was also investigated. The
problems. conclusion was the rejection of the liquidity
In addition, log transformations was explanation given that a similar relation-
used in all variables for which skewness ship is also observed when insiders hold
was seen as relatively high but again no non-beneficial holdings in addition to
relevant departures from former results beneficial holdings that alone are below
were observed. Also, since some evidence of the critical turning point. This point was
non-normality in the residuals was observed also reinforced when no positive association
that could be a symptom of misspecification, was observed between dividend payouts
a robust estimation analysis was performed. and the market value of beneficial insider
Specifically, a rank regression procedure by holdings. The analysis suggested that hold-
which all variables (except dummies) were ings over which insiders have control, but
converted into their respective ranks (and not cash-flow rights, can be conducive to
INSBEN2 was redefined as the square of entrenchment.
the rank of INSBEN). The results of using Consistent with the existence of links
this procedure revealed that the insider between corporate governance and dividend
ownership variables were still highly signif- policy, compliance with the Cadbury (1992)
icant at the 1% level or very close to it. Code of Best Practice was observed to have
Finally, no evidence of significant a statistically and economically significant
heteroskedasticity was found when using a impact on dividend payouts. Also in accor-
White (1980) test, which is also a test for dance with an agency perspective, strong
misspecification. In spite of this, White evidence was produced that shareholder
(1980) adjusted t-statistics were used but, dispersion has a significant positive impact
as expected, did not reveal any significant on dividend policy.
departures from the significance levels The main results presented in this paper
observed before. vindicate the agency explanation for cross-
sectional dividend policy. Some limitations
of the analysis should, however, be kept in
7. SUMMARY AND DISCUSSION mind. First, the non-insider ownership data
OF FINDINGS does not include ownership levels below
3%. In principle, it is possible that allowing
This paper provides an empirical examina- for a finer partition could alter the signifi-
tion of the agency theory explanation for cance of some of the ownership variables.
the cross-sectional distribution of dividend In addition, the results may not be easily
policies in the UK. Using data for two five extrapolated to the smallest firms. Indeed,
620 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
the analysis of interactions between insider 1 In the UK, dividends are proposed by
ownership and size for the 1996 sample directors and no dividends can be approved by
suggested that in smaller firms a U-shaped shareholders if they exceed the amount proposed
relationship between dividend policy and (see Companies Act 1985,Table A, article 102).
2 For an analysis of the Cadbury (1992)
insider holdings might not hold.11 Finally,
report in the context of UK corporate gover-
the results on the usage by managers of nance see Sheikh and Rees (1995).
non-beneficial holdings as an entrenchment 3 DeAngelo et al. (1992) point out that, con-
vehicle rely on a relatively small number sistent with arguments by Modigliani and Miller
of observations above the critical levels of (1959), discarding unusual income items pro-
insider ownership, suggesting therefore a vides a better explanation for firm’s dividend
degree of caution in the interpretation of decisions.
these particular findings. 4 An attempt was also made to use annual
A final issue regards the existing literature cross-sections as an alternative to the specifica-
on the simultaneous determination, or tion above. As predicted, the noisiness of divi-
dend payout ratios in the short term reduced
endogeneity, of several alternative or
dramatically the overall significance of the
complementary corporate governance equation. Results became, however, closer to
mechanisms (see for instance Agrawal and those presented in the text (albeit with much
Knoeber, 1996). Specifically, and in the lower significance levels), when annual regres-
spirit of Jensen et al. (1992), a simultane- sions were restricted so as to exclude firms
ous specification for the joint determination whose short-term earnings were more volatile.
of dividend policy and other monitoring 5 In a questionnaire survey of companies’
devices with allowance for the entrenchment reasons for not paying dividends, Baker (1989)
effects suggested in this paper, might yield observes that growth and expansion through
some incremental explanatory power. investment is a reason listed by 76% of the
respondents.
6 Baker (1989) documents that 22% of
companies inquired on the reasons for paying no
NOTES dividends cite debt covenants and restrictions.
7 Surplus ACT can arise when dividends are
* The author is from CETE-Centro de Estudos paid in excess of the maximum amount of tax-
de Economia Industrial, do Trabalho e da able profits that the UK tax system allowed
Empresa, Faculdade de Economia, Universidade ACT to be set against. This surplus can be the
do Porto, Portugal. This paper is based on part result of a variety of situations, namely when the
of his PhD dissertation at Lancaster University, company pays dividends out of reserves, when
UK. He is particularly grateful to his supervi- the tax system allows capital to be written off at
sors, Ken Peasnell and Peter Pope, and also to a different rate than that used in the accounts,
Sudi Sudarsanam, Steve Young, Sanjay Unni, or when dividends are paid out of foreign
Massoud Mussavian, Robin Limmack, Abe de income. Although this surplus can, with some
Jong, participants at the 1998 EFA-European limitations, be relieved by carrying it back or
Finance Association Meeting, the 1998 forward, permanent differences between divi-
BAAICAEW Doctoral Colloquium, the dends paid and taxable profits can occur that
Accounting and Finance Seminar Series at lead to structural irrecoverable ACT surplus. See
Lancaster University, the 2002 PFN-Portuguese Freeman and Griffith (1993) for a description
Finance Network Meeting, the 2002 EFMA- of the mechanics of ACT.
European Financial Management Association 8 It should be noted, however, that one cannot
Conference, and the 2002 1st International reject the possibility of inverse causality.
Conference on Corporate Governance at the 9 Using or benchmarks set at 0.5, 1, or 2
University of Birmingham, for all the helpful p.p. below estimated entrenchment levels
comments received. He is thankful to I/B/E/S yielded similar results.
International Inc. for allowing the usage of the 10 For a description of this tax condition
I/B/E/S database and also gratefully acknowl- see, for instance, Whitehouse et al. (1993,
edges the generous financial support received pp. 514–18).
from Fundação para a Ciência e Tecnologia/ 11 Also, in a (unreported) more detailed
Praxis XXI Programme and from Faculdade de analysis of the sample, comparison between
Economia, Universidade do Porto. firms in the LSPD-London Share Price
DIVIDEND POLICY AND MANAGERIAL ENTRENCHMENT 621
ABSTRACT
This paper examines systematic differences in earnings management across 31 countries. We
propose an explanation for these differences based on the notion that insiders, in an attempt
to protect their private control benefits, use earnings management to conceal firm performance
from outsiders. Thus, earnings management is expected to decrease in investor protection
because strong protection limits insiders’ ability to acquire private control benefits, which
reduces their incentives to mask firm performance. Our findings are consistent with this
prediction and suggest an endogenous link between corporate governance and the quality of
reported earnings.
of interest between firms’ insiders and enjoy greater private control benefits and
outsiders. Insiders, such as controlling own- hence have stronger incentives to obfuscate
ers or managers, can use their control over firm performance.
the firm to benefit themselves at the Our analysis is based on financial
expense of other stakeholders. Examples of accounting data from 1990 to 1999 for
such private control benefits range from over 8,000 firms from 31 countries. To
perquisite consumption to the transfer of measure the pervasiveness of earnings
firm assets to other firms owned by insiders management in a country, we create four
or their families. The common theme, proxies that capture the extent to which
however, is that some value is enjoyed corporate insiders use their accounting
exclusively by insiders and thus not shared discretion to mask their firm’s economic
with non-controlling outsiders. performance. As it is difficult to specify ex
Insiders have incentives to conceal their ante which techniques firms use to obfus-
private control benefits from outsiders cate firm performance, our earnings
because, if these benefits are detected, management proxies are designed to
outsiders will likely take disciplinary action capture a variety of earnings management
against them (see, e.g., Zingales, 1994; practices such as earnings smoothing and
Shleifer and Vishny, 1997). Accordingly, we accrual manipulations.
argue that managers and controlling own- We begin with a descriptive country
ers have incentives to manage reported cluster analysis, which groups countries
earnings in order to mask true firm with similar legal and institutional charac-
performance and to conceal their private teristics.Three distinct country clusters are
control benefits from outsiders. For exam- identified: (1) outsider economies with
ple, insiders can use their financial report- large stock markets, dispersed ownership,
ing discretion to overstate earnings and strong investor rights, and strong legal
conceal unfavorable earnings realizations enforcement (e.g., United Kingdom and
(i.e., losses) that would prompt outsider United States); (2) insider economies with
interference. Insiders can also use their less-developed stock markets, concentrated
accounting discretion to create reserves for ownership, weak investor rights, but strong
future periods by understating earnings in legal enforcement (e.g. Germany and
years of good performance, effectively Sweden); and, (3) insider economies with
making reported earnings less variable weak legal enforcement (e.g., Italy and
than the firm’s true economic perform- India). These clusters closely parallel
ance. In essence, insiders mask their private simple code-law and common-law as well
control benefits and hence reduce the as regional characterizations used in prior
likelihood of outside intervention by work (e.g., La Porta et al., 1997; Ball et al.,
managing the level and variability of 2000). We find significant differences in
reported earnings. earnings management across these three
Legal systems protect investors by institutional clusters. Outsider economies
conferring on them rights to discipline with strong enforcement display the lowest
insiders (e.g., to replace managers), as well level of earnings management and insider
as by enforcing contracts designed to limit economies with weak enforcement the
insiders’ private control benefits (e.g., La highest level of earnings management.That
Porta et al., 1998; Nenova, 2000; is, earnings management appears to be
Claessens et al., 2002; Dyck and Zingales, lower in economies with large stock mar-
2002).2 As a result, legal systems that kets, dispersed ownership, strong investor
effectively protect outside investors reduce rights, and strong legal enforcement.
insiders’ need to conceal their activities.We To examine more explicitly whether
therefore propose that earnings manage- differences in earnings management are
ment is more pervasive in countries where related to private control benefits and
the legal protection of outside investors is investor protection, we undertake a multi-
weak, because in these countries insiders ple regression analysis.We measure outside
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 625
benefits when faced with higher penalties. median ratio of the firm-level standard
While we acknowledge the potential deviation of operating earnings divided by
existence of such a penalty effect, the the firm-level standard deviation of cash
empirical evidence suggests that it is flow from operations. Scaling by the cash
dominated by international differences in flow from operations controls for differ-
private control benefits, and thus the ences in the variability of economic per-
negative relation between investor protection formance across firms. Low values of this
and earnings management prevails. measure indicate that, ceteris paribus,
The remainder of the paper is organized insiders exercise accounting discretion to
as follows. Section 2 describes the con- smooth reported earnings.
struction of our earnings management Cash flow from operations is computed
measures. In Section 3, we describe the indirectly by subtracting the accrual
sample and provide descriptive statistics. component from earnings because direct
Empirical tests and results are presented in information on firms’ cash flows is not
Section 4. Section 5 concludes. widely available in many countries.
Following Dechow et al. (1995), we com-
pute the accrual component of earnings as
2. EARNINGS MANAGEMENT
MEASURES Accrualsit (CAit Cashit)
(CLit STDit
This section describes the earnings man- TPit) Depit, (1)
agement measures used in our empirical
analysis. Drawing on the existing earnings where CAit change in total current
management literature (see Healy and assets, Cashit change in cash/cash
Wahlen, 1999; Dechow and Skinner, equivalents, CLit change in total cur-
2000), we develop four different country- rent liabilities, STDit change in short-
level measures of earnings management term debt included in current liabilities,
that capture various dimensions along TPit change in income taxes payable,
which insiders can exercise their discretion and Depit depreciation and amortization
to manage reported earnings. The four expense for firm i in year t. Changes in
measures capture outcomes of insiders’ short-term debt are excluded from accruals
earnings management activities and avoid because they relate to financing transac-
the problem that stated accounting rules tions as opposed to operating activities. If
can be (and often are) circumvented by a firm does not report information on taxes
insiders and hence do not reflect firms’ payable or short-term debt, then the change
actual reporting practices (see also Ball in both variables is assumed to be zero.
et al., 2003).
2.2. Smoothing and the correlation
2.1. Smoothing reported operating between changes in accounting
earnings using accruals accruals and operating cash flows
Insiders can conceal changes in their firm’s Insiders can also use their accounting dis-
economic performance using both real cretion to conceal economic shocks to the
operating decisions and financial reporting firm’s operating cash flow. For example,
choices. Focusing on insiders’ reporting they may accelerate the reporting of future
choices, our first earnings management revenues or delay the reporting of current
measure captures the degree to which costs to hide poor current performance.
insiders “smooth”, i.e., reduce the variabil- Conversely, insiders underreport strong
ity of reported earnings by altering the current performance to create reserves for
accounting component of earnings, namely the future. In either case, accounting accru-
accruals. The measure is a country’s als buffer cash flow shocks and result in a
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 627
around the world. Banks and financial earnings management measures as well as
institutions are excluded from the empiri- the aggregate earnings management score.
cal analysis.To be included in the sample, a The countries are sorted in descending
country must have at least 300 firm-year order based on their aggregate score.
observations for a number of accounting The four individual earnings management
variables, including total assets, sales, net measures exhibit striking differences across
income, and operating income. Each firm countries, but similar patterns in terms of
must have income statement and balance their relative magnitudes. The statistics
sheet information for at least three consec- of the first measure (EM1) show that
utive years. Finally, Argentina, Brazil, and earnings are smoother in Continental
Mexico experienced hyperinflation over the Europe and Asia than in Anglo-American
sample period and are excluded from the countries, after controlling for the volatility
main sample because high inflation may of cash flows. Similarly, large negative
unduly affect our earnings management correlations between changes in firms’
measures. However, the results are qualita- accruals and cash flows (EM2) indicate
tively unchanged if these countries remain that earnings smoothing is more pervasive
in the sample. The final sample consists in, for instance, Greece and Japan than in
of 70,955 firm-year observations, across Canada and the U.S. With regard to
31 countries and 8,616 non-financial firms accounting discretion, the third measure
for the fiscal years 1990 to 1999. (EM3) shows that the magnitude of firms’
Table 24.1 presents the number of firm- accruals, relative to the magnitude of their
year observations per country as well as operating cash flows, is small in the U.K.
descriptive statistics for the sample firms and the U.S. compared to Austria,
and countries. There is significant variation Germany, and South Korea. Similarly, the
in the number of firm-year observations fourth measure (EM4) reveals that
across countries due to differences in European and Asian firms exhibit a greater
capital market development, country size, degree of loss avoidance than Anglo-
and the availability of complete financial American firms.5
accounting data. Note that the U.S. version The earnings management measures
of the Worldscope Database includes only are highly correlated and the rankings
U.S. firms belonging to the S&P 500 index. corresponding to the four individual
However, our results are not sensitive to the measures and the aggregate earnings
inclusion of the U.S. (or any particular management score are similar. Factor
country). To allow for direct firm size analysis suggests that a single factor
comparisons across countries, the median represents the four individual measures.
firm’s sales in US$ is reported for each Thus, it seems appropriate to combine
country. Based on the large differences in the four measures into a single summary
the median firm size across countries, measure of earnings management. Our
we scale all financial variables by the results hold for the smoothing and discre-
lagged value of total assets. Scaling by tion measures separately, as well as for the
other variables such as lagged sales or single factor identified by factor analysis.
market value of equity does not affect the The last column of Table 24.2 Panel A
results.Table 24.1 also shows a substantial presents a country ranking based on this
cross-country variation in capital intensity, aggregate earnings management score,
the fraction of manufacturing firms, per showing high ranks for countries such as
capita GDP, inflation and volatility Austria, Italy, and South Korea, and low
of growth. We address the potentially ranks for countries such as Australia, the
confounding effects of cross-country U.K. and the U.S.
differences in these variables in subsequent Panel B of Table 24.2 provides descriptive
multiple regressions. statistics on the institutional characteris-
Panel A of Table 24.2 provides tics of each country in the sample and is
descriptive statistics for the four individual sorted based on countries’ aggregate
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 629
The full sample consists of 70,955 firm-year observations for the fiscal years 1990 to 1999 across 31
countries and 8,616 non-financial firms. Financial accounting information is obtained from the
November 2000 version of the Worldscope Database. To be included in our sample, countries must have
at least 300 firm-year observations for a number of accounting variables, including total assets, sales,
net income, and operating income. For each firm, we require income statement and balance sheet
information for at least three consecutive years. We discard three countries (Chile, New Zealand,Turkey)
because of an insufficient number of observations to compute the loss avoidance measure, and three
countries (Argentina, Brazil, Mexico) due to hyperinflation. Firm size is measured as total US$ sales (in
thousands). Capital intensity is measured as the ratio of long-term assets over total assets. The fraction
of manufacturing firms is the percentage of firm-year observations with SIC 2000 to 3999. Average
per capita GDP in constant 1995 US$ is computed from 1990 to 1999. Inflation is measured as the
average percentage change in consumer prices from 1990 to 1998. Volatility of GDP growth is measured
as the standard deviation of the growth rate in real per capita GDP from 1990 to 1998.
Panel A: Country scores for earnings management measures (Sorted by aggregate earnings management)
The table presents results from a k-means cluster analysis using three distinct clusters and nine
institutional variables from La Porta et al., (1997, 1998). See Panel B of Table 24.2 for details.
The variables are standardized to z-scores. Panel A reports the means of the institutional variables
by cluster. Panel B reports the cluster membership for the 31 sample countries based on the clus-
ter analysis performed on the variables in panel A. Countries in each cluster are sorted by the aggre-
gate earnings management score from Panel A in Table 24.2. CD (CM) indicates a code-law
(common-law) tradition. This variable is not used in the cluster analysis. Panel C reports the mean
aggregate earnings management score for each cluster. The last row reports one-sided p-values for
differences in the means of the aggregate earnings management across clusters using a t-test.
Countries Sorted by Aggregate Earnings Singapore (CM) Austria (CD) Greece (CD)
Management Score Hong Kong (CM) Taiwan (CD) Korea (CD)
Malaysia (CM) Switzerland (CD) Portugal (CD)
UK (CM) Germany (CD) Italy (CD)
Norway (CD) Japan (CD) India (CM)
Canada (CM) Belgium (CD) Spain (CD)
Australia (CM) Netherlands (CD) Indonesia (CD)
USA (CM) Denmark (CD) Thailand (CM)
France (CD) Pakistan (CM)
Finland (CD) Philippines (CD)
Sweden (CD)
South Africa (CM)
Ireland (CM)
between the cluster centers supports our protection, large stock markets and dispersed
interpretation that clusters 2 and 3 are ownership. The third cluster exhibits signif-
closer to each other than clusters 1 and 2. icantly higher earnings management than
Overall, the results in Table 24.3, Panel A the second cluster, highlighting the salient
are consistent with the existence of institu- importance of legal enforcement.
tional complementarities.
Table 24.3, Panel B shows the cluster
membership of the sample countries.
4.2. The role of investor protection:
Groupings are consistent with the common- multiple regression analysis
and code-law as well as regional distinc- The previous analyses suggest that the
tions used in prior research to classify pervasiveness of earnings management is
countries (see, e.g., Ball et al., 2000; Ball systematically related to a country’s insti-
et al., 2003). As indicated in Panel B, all tutional characteristics. A key question,
countries in the first cluster with the excep- however, is: Which institutional factors are
tion of Norway have a common-law tradi- primary determinants of earnings manage-
tion. The three Southeast Asian countries ment and which are correlated outcomes?
(Hong Kong, Malaysia, and Singapore) in We posit that better investor protection
this cluster were formerly under British results in less earnings management
rule and have inherited parts of the Anglo- because insiders enjoy fewer private control
Saxon institutional framework. The fact benefits and hence have lower incentives to
that the three East Asian countries have by conceal firm performance from outside
far the worst earnings management ratings investors. This hypothesis ties in closely
in this group is consistent with Ball et al. with findings in Nenova (2000) and Dyck
(2003) who argue that, despite the com- and Zingales (2002), suggesting that
mon-law influence, reported earnings do private control benefits decrease in the
not exhibit common-law properties (i.e., level of investor protection. The notion of
asymmetric timeliness). Fan and Wong investor protection as a key primitive is
(2001) present similar findings. In the sec- also reinforced by recent work relating to
ond cluster, all countries except Ireland and capital market development (e.g., Beck et
South Africa have a code-law tradition. al., 2003), corporate policy choices around
This cluster contains most of the Northern the world (e.g., La Porta et al., 2000), and
European and Scandinavian countries. The cross-listing in the U.S. (e.g., Doidge et al.,
third cluster consists of several Asian and 2003; Lang et al., 2003). Consistent with
Southern European countries with both this literature, we view low earnings
common- and code-law traditions.Thus, the management, large equity markets, and
cluster approach suggests that the com- dispersed ownership patterns as comple-
mon- and code-law distinction matters only ments and joint outcomes of strong investor
when legal enforcement is relatively high, protection. This view is in contrast to La
as in the first and second clusters. In the Porta et al. (1997, 1999) who treat the
third cluster, for which the quality of legal level of disclosure as an exogenous factor
enforcement is low, legal tradition seems in explaining financing and ownership
unrelated to cluster membership. patterns. Our results suggest, however, that
Panel C of Table 24.3 shows that differ- the quality of reported earnings and finan-
ences between the clusters’ average earn- cial disclosure is endogenous and hence a
ings management scores are statistically joint outcome.
significant. Outsider economies (cluster 1) Our multiple regressions examine the
exhibit lower levels of earnings manage- relation between earnings management and
ment than insider economies (clusters 2 investor protection. Column 1 of Table 24. 4
and 3). Thus, even after controlling for reports a rank regression using the aggre-
interactions among various institutional gate earnings management measure as
factors, earnings management appears to the dependent variable. Results show that
be lower in economies with strong investor outside investor protection explains a
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 637
substantial portion (39%) of the variation is to directly estimate the relation between
in earnings management. Outsider rights and earnings management and private control
legal enforcement both exhibit a significant benefits, explicitly accounting for the effect
negative association with earnings of investor protection on the level of private
management. Ordinary least squares (OLS) control benefits. We use a country’s average
regressions of the aggregate earnings block premium estimated by Dyck and
management score on the unranked Zingales (2002) as a proxy for the level
variables yield similar results in this and in of private control benefits. We estimate a
subsequent regressions. 2SLS regression of the aggregate earnings
The multiple regressions assume, however, management score on the control benefits
that outside investor rights and legal proxy using the level of outsider rights and
enforcement are exogenous variables. If, on legal enforcement as instruments. The
the other hand, outsider protection and results presented in column 3 of Table 24.4
earnings management are simultaneously show that earnings management and private
determined, our results suffer from an control benefits exhibit a significantly
endogeneity bias. We address this concern positive association as predicted by our
by using countries’ legal origins and wealth hypothesis. Similar results are obtained if
as instruments for the investor protection the legal origins and per capita GDP are
variables as suggested by Levine (1999). used as instruments (as in column 2).
While related to the level of investor
protection (see La Porta et al., 1998), a
country’s legal origin can be considered 4.3. Robustness checks
as predetermined and exogenous to our Prior work shows that per capita GDP
analysis because the origins of most legal explains differences in financing, ownership,
systems are several centuries old and many and payout policies across countries.
countries obtained their legal system Consequently, we re-estimate our primary
through occupation and colonization. We regressions using contemporaneous per
use three dummy variables, indicating capita GDP as an additional explanatory
English, French, German, and Scandinavian variable (not reported). While GDP is
legal origins, as instrumental variables. In marginally significant in this regression
addition, we use a country’s average per (p 0.140), the negative relation between
capita GDP — measured prior to our investor protection and earnings manage-
sample period, 1980 to 1989 — as an ment is robust to the inclusion of this proxy.
instrument because an effective legal infra- Another potential concern is that our
structure is costly to create and maintain, results are driven by economic heterogeneity
and hence a country’s wealth potentially across countries. Although we control for
influences the level of legal enforcement. economic differences across firms by scaling
Column 2 of Table 24.4 reports results our earnings management measures by
of a 2SLS regression using ranked firms’ operating cash flows, variation in
variables. The regression results support industry composition and firm size across
our hypothesis that the pervasiveness countries can potentially affect our results.
of earnings management decreases in the Since Table 24.1 shows that the fraction of
level of investor protection, and suggest manufacturing firms and median firm size
that this relation is not driven by the poten- vary considerably across countries, the
tial endogeneity of investor protection. regressions are re-estimated using two
Finally, we attempt to provide more subsamples comprised exclusively of
direct evidence on the hypothesis that manufacturing firms (SIC 2000–3999)
insiders’ private control benefits are and medium-size firms from each country,
positively related to earnings management. respectively.The medium-size firm subsample
In the previous regressions, we employ an also eliminates many multinationals
indirect approach by using the investor operating in several institutional settings.
protection variables. An alternative approach The regression results for these subsamples
638 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
Table 24.4 Earnings management, outside investor protection and private control benefits
The table presents coefficients and two-sided p-values (in parentheses) from rank regressions with
the Aggregate Earnings Management Measure as the dependent variable, which is created by
averaging the ranks of all four earnings management measures, EM1–EM4 (see Table 24.2).
Outside Investor Rights are measured by the anti-director rights index from La Porta et al., (1998),
which ranges from zero to five. Legal Enforcement is measured as the average score across three
legal variables used in La Porta et al., (1998): (1) the efficiency of the judicial system, (2) an
assessment of rule of law, and (3) the corruption index. All three variables range from zero to ten.
Private Control Benefits are measured at the country level as the average block premium estimated
by Dyck and Zingales (2002) based on transfers of controlling blocks of shares. The first column
presents a simple rank regression.The second regression is estimated using two-stage least squares.
Instrumental variables are the rank of the country’s real per capita GDP averaged from 1980 to
1989, and three binary variables indicating an English, German, French, or Scandinavian legal
origin based on the classification in La Porta et al., (1998). The third regression is also estimated
using two-stage least squares. The instrumental variables are the Outsider Rights Index and the
Legal Enforcement.
(not reported) are essentially the same investor rights and legal enforcement
as those presented in Table 24.4, alleviat- continue to have a significantly negative
ing concerns that international differences relation with earnings management.
in firm size and industry composition drive
our findings. 4.4. The role of other
Finally, we are concerned that differences institutional factors
in firm characteristics and macroeconomic
stability affect our inferences. For instance, While the robustness checks in the previous
larger firms have smoother earnings, and section suggest that our findings are not
operating leverage is positively related to driven by economic heterogeneity across
earnings volatility. Similarly, inflation rates countries, we must still address the concern
and growth rate volatility influence the vari- that other institutional variables, which
ability of accounting earnings. Consequently, are correlated with investor protection,
we re-estimate the regressions using median are responsible for our main findings.
firm size, median capital intensity, a coun- In particular, we are concerned about
try’s average yearly inflation rate, and the the influence of accounting rules and
standard deviation of the real GDP growth firms’ ownership structures on earnings
rate as additional controls. The results (not management.
reported) are consistent with our original First, accounting rules can both limit a
findings in Table 24.4. In particular, outside manager’s ability to distort reported
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 639
earnings, and affect the properties of Germany or Italy. Similarly, d’ Arcy (2000)
reported earnings. But the extent to which shows that Anglo-American countries have
accounting rules influence reported earn- stricter accounting rules with respect to
ings and curb earnings management explicit accounting choices than Continental
depends on how well these rules are European countries with less effective
enforced. Moreover, accounting rules likely investor protection.
reflect the influence of a country’s legal Ultimately, however, it is an empirical
and institutional framework and are there- matter whether our results are robust to the
fore endogenous in our analysis. Countries inclusion of controls for countries’ stated
with strong outsider protection are expected accounting rules. To address this issue, we
to enact and enforce accounting and secu- re-estimate the main regression and include
rities laws that limit the manipulation an accrual rules index constructed by Hung
of accounting information reported to out- (2001) as a control variable. This index
siders. Consistent with this view, Enriques captures the use of accrual rules to acceler-
(2000) argues that U.K. and the U.S. laws ate the recognition of economic transactions
on director self-dealing are stricter and are (e.g., R&D activities or pension plans) in
more reliant on disclosure than those in accounting, and it proxies for the extent
Table 24.5 Earnings management and outside investor protection: Controlling for differences in
the accounting rules and ownership concentration
The table presents coefficients and two-sided p-values (in parentheses) from rank regressions of the
Aggregate Earnings Management Measure on Outside Investor Rights and Legal Enforcement controlling
for other institutional factors. Outside Investor Rights is the anti-director rights index from La Porta
et al., (1998), which ranges from zero to five. Legal Enforcement is measured as the average score across
three legal variables used in La Porta et al. (1998): (1) the efficiency of the judicial system, (2) an assess-
ment of rule of law, and (3) the corruption index. All three variables range from zero to ten. The Accrual
Rules variable captures the extent to which accrual rules accelerate the recognition of economic trans-
actions (e.g., R&D activities or pension obligations) in accounting. It is constructed by Hung (2001).
Ownership concentration is measured as the median percentage of common shares owned by the largest
three shareholders in the ten largest privately owned non-financial firms (La Porta et al., 1998). The
regressions in columns 2 and 4 are estimated using two-stage least squares. Instrumental variables are
the rank of the country’s real per capita GDP averaged from 1980 to 1989, and three binary variables
indicating an English, German, French, or Scandinavian legal origin based on the classification in
La Porta et al. (1998).
Aggregate Aggregate
Aggregate Earnings Aggregate Earnings
Earnings Management Earnings Management
Management Controlling for Management Controlling for
Controlling for Accounting Rules Controlling for Ownership
Accounting Rules -2SLS- Ownership -2SLS-
Consistent with this hypothesis, the 1 While the investor protection literature
regression results show that earnings acknowledges the importance of accounting
management is negatively associated with information, it typically treats the quality of
the quality of minority shareholder rights this information as exogenous and does not
distinguish between stated accounting rules and
and legal enforcement. The findings high-
firms’ actual reporting practices (e.g., La Porta
light an important link between investor et al., 1998).
protection and the quality of accounting 2 Outsiders are also expected to price protect
earnings reported to market participants, themselves, leading to more internal financing,
and complement both finance research that smaller arm’s length financial markets and
treats the quality of corporate reporting higher cost of outside capital (see, for example,
as exogenous and accounting research that La Porta et al., 1997). Bhattacharya et al.
documents systematic patterns in the (2002) replicate our earnings management
relation between stock returns and measures and provide evidence that firms’
accounting numbers. earnings management activities appear to be
priced in capital markets.
Our findings are robust to the inclusion
3 See also Basu et al. (1998) and Hope
of controls for country wealth, economic (2003) relating the properties of analyst
heterogeneity across countries, and inter- forecasts to institutional factors.
national differences in accounting rules and 4 As accounting systems likely underreact to
ownership concentration. They should economic shocks, insiders using accruals to
nevertheless be interpreted cautiously as signal firm performance induce on average a
earnings management is difficult to measure less negative (and in specific cases even
and the theoretical relations among institu- positive) correlation with cash flows.
tional factors are not yet well understood 5 Our loss avoidance results may appear to
and hence difficult to disentangle. contradict the finding of Brown and Higgins
(2001) that earnings surprise management is
more pronounced in the US than in other coun-
tries. However, the two findings are compatible.
NOTES Brown and Higgins (2002) show that US firms
engage in more expectations management,
* This paper was previously circulated and i.e., downward guidance of analysts, to meet or
presented under the title “Earnings beat analysts’ earnings forecasts, rather than
Management Around the World’’ at the 2000 earnings management.
European Accounting Association meetings in
Munich. It has also benefited from presentations
at the Duke/UNC Fall Camp, the EAA meetings
in Athens, the European Finance Association REFERENCES
meeting in Berlin, the Financial Economics and
Accounting conference at the University of
Alford, A., Jones, J., Leftwich, R., Zmijewski, M.,
Michigan, Goethe University Frankfurt, MIT,
London Business School, New York University, 1993. The relative informativesness of
University of Vasa, University of Virginia, the accounting disclosures in different countries.
Wharton School, and the College of William Journal of Accounting Research 31 (Suppl.),
and Mary. We are grateful to Stan Baiman, 183–221.
Sudipta Basu, Phil Berger, Larry Brown, Willem Ali, A., Hwang, L., 2000. Country-specific
Buijink, Elroy Dimson, Simeon Djankov, factors related to financial reporting and
Richard Frankel, Wayne Guay, Juha Kinnunen, the value relevance of accounting data.
David Larcker, Christian Laux, Martien Journal of Accounting Research 38, 1–23.
Lubberink, Peter Pope, Bill Schwert, Weiling Ball, R., Kothari, S., Robin, A., 2000. The effect
Song, Ross Watts, and especially Bob
of international institutional factors on
Holthausen and an anonymous referee for
helpful comments. properties of accounting earnings. Journal of
† Corresponding author. Tel.: 1-215-898- Accounting and Economics 29, 1–52.
2610; fax: 1-215-573-2054. E-mail address: Ball, R., Robin, A., Wu, J., 2003. Incentives
leuz@wharton.upenn.edu (C. Leuz). versus standards: properties of accounting
642 GOVERNANCE, PERFORMANCE AND FINANCIAL STRATEGY
income in four East Asian countries. Journal Enriques, L., 2000. The law on company
of Accounting and Economics, forthcoming. directors’ self-dealing: a comparative analysis.
Basu, S., Hwang, L., Jan, C., 1998. International International and Comparative Corporate
variation in accounting measurement rules Law Journal 2, 297–314.
and analysts’ earnings forecast errors. Journal Fan, J., Wong, T., 2001. Corporate ownership
of Business, Finance and Accounting 25, structure and the informativeness of account-
1207–1247. ing earnings in east asia. Journal of
Beck, T., Demirgüe-Kunt, A., Levine, R., 2003. Accounting and Economics 33, 401–426.
Law, politics, and finance. Journal of Healy, P., Wahlen, J., 1999. A review of the
Financial Economics, forthcoming. earnings management literature and its
Bhattacharya, U., Daouk, H., Welker, M., 2002. implications for standard setting. Accounting
The world price of earnings opacity. Unpub- Horizons 13, 365–383.
lished working paper, Indiana University. Hope, O., 2003. Disclosure practices, enforcement
Brown, L., Higgins, H., 2001. Managing of accounting standards, and analysts’
earnings surprises in the U.S. versus 12 other forecasts: an international study. Journal of
countries. Journal of Accounting and Public Accounting Research 41, 235–272.
Policy 20, 373–398. Hung, M., 2001. Accounting standards and
Brown, L., Higgins, H., 2002. Managers’ value relevance of financial statements: an
guidance of analysts: international evidence. international analysis. Journal of Accounting
Unpublished working paper, Georgia State and Economics 30, 401–420.
University and Worcester Polytechnic Institute. Joos, P., Lang, M., 1994. The effects of
Burgstahler, D., Dichev, I., 1997. Earnings accounting diversity: evidence from the
management to avoid earnings decreases and European Union. Journal of Accounting
losses. Journal of Accounting and Economics Research 34(Suppl.), 141–168.
24, 99–129. La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Claessens, S., Djankov, S., Fan, J., Lang, L., Vishny, R., 1997. Legal determinants of external
2002. Disentangling the incentive and finance. Journal of Finance 52, 1131–1150.
entrenchment effects of large shareholdings. La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
Journal of Finance 57, 2741–2772. Vishny, R., 1998. Law and finance. Journal of
d’Arcy, A., 2000.The degree of determination of Political Economy 106, 1113–1155.
national accounting systems — an empirical La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
investigation. Schmalenbach Business Review Vishny, R., 1999. Corporate ownership
52, 45–67. around the world. Journal of Finance 54,
Dechow, P., 1994. Accounting earnings and 471–517.
cashflows as measures of firm performance: La Porta, R., Lopez-de-Silanes, F., Shleifer, A.,
the role of accounting accruals. Journal of Vishny, R., 2000. Investor protection and
Accounting and Economics 18, 3–42. corporate governance. Journal of Financial
Dechow, P., Skinner, D., 2000. Earnings man- Economics 58, 3–27.
agement: reconciling the views of accounting Land, J., Lang, M., 2002. Empirical evidence on
academics, practitioners, and regulators. the evolution of global accounting. Accounting
Accounting Horizons 14, 235–250. Review 77(Suppl.), 115–133.
Dechow, P., Sloan, R., Sweeney, A., 1995. Lang, M., Raedy, J., Yetman, M., 2003. How
Detecting earnings management. The representative are cross-listed firms? An
Accounting Review 70, 193–225. analysis of firm performance and accounting
Degeorge, F., Patel, J., Zeckhauser, R., 1999. quality. Journal of Accounting Research 41,
Earnings manipulation to exceed thresholds. 363–386.
Journal of Business 72, 1–33. Levine, R., 1999. Law, finance and economic
Doidge, C., Karolyi, A., Stulz, R., 2003. Why are growth. Journal of Financial Intermediation
firms listed in the U.S. worth more? Journal 8, 8–35.
of Financial Economics, forthcoming. Nenova, T., 2000. The value of corporate votes
Dyck, A., Zingales, L., 2002. Private benefits and control benefits: a cross-country analysis.
of control: an international comparison. Unpublished working paper, Harvard
Unpublished NBER working paper (8711). University.
EARNINGS MANAGEMENT AND INVESTOR PROTECTION 643
Shleifer, A., Vishny, R., 1997. A survey of Watts, R., Zimmerman, J., 1986. Positive
corporate governance. Journal of Finance 52, Accounting Theory. Prentice-Hall, Englewood
737–783. Cliffs, NJ.
Skinner, D., Myers, L., 1999. Earnings momentum Zingales, L., 1994.The value of the voting right:
and earnings management. Unpublished a study of the Milan stock exchange experi-
working paper, University of Michigan and ence. Review of Financial Studies 7,
University of Illinois. 125–148.
Part 6
On takeover as disciplinary
mechanism
INTRODUCTION
REFERENCE
Manne, Henry G. (1965), ‘Mergers and the market for corporate control’ Journal of Political
Economy, 73: 110–120.
Chapter 25
ABSTRACT
This study empirically examines the relationships between a firm’s takeover defenses and its
ownership structure and stock returns. Analyzing data of Dutch listed companies, we find that
multiple antitakeover defenses are increasingly adopted when firms are characterized by
relatively lower ownership concentration. The evidence supports the hypothesis that more
concentrated ownership of shares provides more effective monitoring of managers. As defense
by issuing preferred share has recently been the most widely adopted mechanism in the
Netherlands, its impact on shareholders’ wealth is also analyzed. We observe the presence of
two opposing effects of this antitakeover measure.
be one group of stakeholders in a firm next also suggest that differences in firms’
to employees, suppliers and customers. The ownership structure (internal control aspect)
equity ownership is also concentrated in can explain observed variations in anti-
the hands of a few investors. Although the takeover defenses (external control aspect).
pattern of cross-shareholdings in German The notion can be illustrated in the
and Japanese companies may look similar, following way.
the governance structures are quite dissim- Shareholders with large stakes are
ilar. German firms have close relationships expected to participate actively in manage-
with banks which supply both equity capital rial decision making (Demsetz, 1983;
and debt. In contrast, Japanese firms are Shleifer and Vishny, 1986).They will not in
characterized by large industrial groups their own interest allow managers to adopt
with interlocking directorships. Hostile defensive measures, as disciplining will be
takeovers are virtually nonexistent in more difficult. The same is true for large
Germany and the Netherlands, but due to but passive shareholders who will also
two different reasons. Extensive cross- try to resist any attempt by managers to
shareholdings provide German companies adopt defenses. This is because any future
with a strong defense, while Dutch compa- possibility of gain through facilitating a
nies are protected by multiple antitakeover third-party takeover will then be reduced.
devices. These and other differences imply Shareholders with small holdings, on the
that the influence of various disciplinary other hand, may not take an active interest
mechanisms will vary from country to in monitoring management, perhaps
country. The takeover market is a relatively because of the ‘free-rider’ problem. Defense
more important disciplinary mechanism in measures are then relatively easily adopted
the U.S.A. and the U.K. But, for Germany by managers because there are no large
and the Netherlands, concentrated ownership shareholders to counteract management’s
and supervisory boards exert a relatively attempt. The purpose of the study is, there-
more important role. Various antitakeover fore, to test empirically this theoretically
measures are adopted in the U.S.A. to predicted relationship between firms’ own-
protect the interests of shareholders during ership concentration and takeover defense
takeover bids. But, in the Netherlands these measures.1
measures are primarily directed to limit the Incentives as well as the degree of
power of common shareholders. monitoring can vary depending on the stakes
A vast literature addresses the inter- and the types of shareholders. One may
relationship between ownership structure be interested to know how institutional
and different corporate governance devices. shareholders, as a separate group, affect
Walsh and Seward (1990) examine different corporate decision making.These investors—
internal and external mechanisms of cor- usually banks, insurance companies, pension
porate control used in aligning the diverse funds and mutual funds—are expected to
interests of managers and shareholders. play a more active role in the affairs of a
Important internal control mechanisms company. They are in a better position to
include the control function of the board invest resources for increased monitoring
of directors, competition within the mana- so that management’s inclination to adopt
gerial team, and the monitoring role of large defense mechanisms decreases. On the
shareholders. The external control mecha- other hand, some institutional investors
nisms, on the other hand, are the market may align with management because of
for corporate control and the competition commercial ties and profitable business
in the product market. Walsh and Seward opportunities. The role actually played
(1990) argue that the failure of one by institutional shareholders, therefore,
control mechanism triggers the presence of becomes an empirical issue.2
another mechanism. Studies by Jarrell and Although it has been argued that large
Poulsen (1987), Ambrose and Megginson shareholders who are effective monitors
(1992), and Gordon and Pound (1993) will prevent managers from adopting
TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 649
defensive measures, one can not be sure if in target shareholder returns. Given these
shareholders in general are harmed by such findings, we examine if the shareholders
adoptions. In fact, adoption of takeover wealth effect of takeover defenses is
defenses is usually explained under two related to ownership structure.
competing hypotheses (DeAngelo and Rice, The remainder of the paper is organized
1983; Mahoney and Mahoney, 1993). as follows. Important takeover defense
According to the managerial entrenchment measures are first discussed with particular
hypothesis, defense measures primarily emphasis on those prevailing in the
protect poorly functioning management Netherlands. The following two sections
by reducing the probability of potential describe the sample and the methodology.
takeover. These measures help incumbent The results are presented in the next
management to abuse their power by acting section. A brief summary of the study and
in their own interest at the expense of the research implications are presented in
shareholders. On the other hand, the share- the final section.
holder interest hypothesis postulates that
adoption of defense measures allows
current management to focus on long-term TAKEOVER DEFENSE MEASURES
strategies of the firm while remaining
protected from the worry of hostile Takeover defense measures help to make
takeovers. Through a strong negotiating acquisition of a company more difficult,
position, managers can also help share- if not impossible, and thereby serve to
holders to obtain a fairer/higher premium if insulate managers from the free market
a takeover does take place. for corporate control. These measures vary
Empirical studies from the U.S.A. from country to country depending on
document that while some defense mecha- institutional features and corporate gover-
nisms are harmful for shareholders, others nance systems (Franks and Mayer, 1990).
are not.3 This study, therefore, reexamines Moerland (1995) distinguishes two basic
the valuation impact of defense measures, types of corporate systems: the market-
using the Dutch data. If shareholders of oriented system (prevailing in the U.S.A.
Dutch companies interpret the adoption and the U.K.) and the network-oriented
of defense measures as managerial system (prevailing in, for example, the
entrenchment, stock prices should decline. Netherlands, Germany, France and Japan).
Alternatively, if these measures allow The former is characterized by relatively
management to bargain for a higher developed financial markets, large-scale
takeover premium, share prices should presence of corporations with widely
increase. dispersed ownership, and active markets for
The wealth effect of defense measures corporate control.The latter system, on the
needs to be examined in conjunction with other hand, features closely held corpora-
the ownership structure of firms. Jarrell tions, group membership of corporations,
and Poulsen (1987) document that value- and substantial involvement of banks in
reducing takeover defenses are adopted by corporate financing and corporate control.
firms with larger insider holdings and These differences in governance systems
smaller institutional holdings. McWilliams are also reflected in differences in adoption
(1990) finds that defense measures induce of specific defense devices.
positive effects on shareholder wealth for There exist a variety of ways to classify
firms with low insider share ownership. takeover defenses. These can be either
Agrawal and Mandelker (1990) report structural or technical.The first type arises
that the effect is more favorable the larger from prevailing structures of stock market
the level of institutional ownership. Song and equity ownership (e.g., relative impor-
and Walkling (1993) find that managerial tance of debt financing, crossholdings). The
ownership is related both to the probability second type of defenses are specifically
of being a takeover target and to increments directed to impede hostile takeover attempts
650 ON TAKEOVER AS DISCIPLINARY MECHANISM
unfriendly takeover attempt. These three Table 25.2 Distribution of different takeover
steps follow one after another, but do not defenses
necessarily take place simultaneously. A
Official Parallel
company may create the possibility to issue
marketa marketa Total
preferred defense shares at a certain point
of time, while the shares are actually issued Total number of firms 141 36 177
several years later (depending on any
Priority shares 63 16 79
threat of hostile takeover). Preferred shares 89 16 105
Binding appointment 49 15 64
Limited voting power 7 0 7
DATA Depository receipts 41 19 70
Inspired by the European Community a The official market is the first-tier market for larger
companies, while the parallel market is for smaller
initiative, shareholders with holdings of companies.
5 percent or more in Dutch listed companies
have been required to disclose their holdings
publicly since February 1992. Before that, firms are protected by three or more
there was no mandatory disclosure of share defense devices. Table 25.2 presents a list
ownership, and no way existed even to of widely used antitakeover measures in the
identify shareholders.5 The data on block- Netherlands. We find that 105 (32%)
holdings are collected from the Dutch firms have adopted defense mechanism
financial daily Het Financieele Dagblad. with preferred shares, 79 (24%) firms
In total, we obtained a sample of 177 have issued priority shares, 70 (22%)
companies listed on the Amsterdam Stock companies have issued depository receipts,
Exchange.These companies represent more 64 (20%) firms have made binding
than 90 percent of the Dutch stock market appointments of directors, and seven (2%)
capitalization. Data on takeover defense companies have imposed restrictions on
measures associated with these companies voting rights.
are collected from Voogd (1989) and other After searching sources like the stock
publications. Our findings are presented in exchange publication Beursplein 5 and the
Tables 25.1 and 25.2. financial daily Het Financieele Dagblad,
Table 25.1 shows that more than we find that 79 new defense mechanisms
90 percent of Dutch companies are were announced by Dutch companies dur-
protected by at least one defense measure. ing 1984–90. Defense with preferred share
We find that while only 16 (9%) companies was the most frequently announced mecha-
are without any of these defenses, 52 (29%) nism—on 52 occasions, which represents
companies have one defense mechanism, 66% of the total. The next most important
62 (35%) firms have two defense antitakeover devices were the issues of
mechanisms, and as many as 47 (27%) priority shares and depository receipts. Both
were adopted on seven occasions each. No
new defense measure was announced during
Table 25.1 Number of takeover defenses 1991–92 because of restrictions imposed
adopted by Dutch companies by the Amsterdam Stock Exchange. Many
companies accelerated the adoption of new
No. of measures No. of firms % antitakover devices before the restriction
took effect.
0 16 9.1
1 52 29.4
Out of the 52 announcements of defense
2 62 35.0 with preferred shares during 1984–90,
3 39 22.0 we select a sample of 47 to analyze share-
4 8 4.5 holder wealth effect.6 The sample is further
Total 177 100 divided into three groups based on the
announcement of three steps followed
652 ON TAKEOVER AS DISCIPLINARY MECHANISM
during the issuing process. The statutory used to calculate ownership concentration.
possibility to defend with preferred shares In addition to the four concentration meas-
was created for the first time during ures defined earlier, we use the logarithmic
1984–90 by 17 companies. During the transformation of these variables as well
same period, the announcement of granting as the Herfindahl measure of concentration
an option allowing friendly parties to own in the regression analysis.9
preferred shares was made by 12 companies To examine whether shareholders
(these 12 companies have taken the first experience any change in their wealth
step either during 1984–90 or earlier). when new takeover defense measures are
Finally, during the period of our investiga- announced, we follow the conventional
tion, there were 18 companies which actually event study methodology.This methodology
announced the issue of preferred shares. has been widely used in the financial
We collect daily share price data from economics literature (e.g., DeAngelo and
Datastream. These are adjusted for stock Rice, 1983; Linn and McConnell, 1983).
splits and other capital changes. We also Recently, it has also become popular in
adjust for cash dividends and then compute the strategic management literature
continuously compounded stock returns for (e.g., Mahoney and Mahoney, 1993). The
the analysis. purpose of this method is to estimate
the deviation of actual stock returns
(consequent upon the announcement of a
METHODOLOGY specified event) from expected stock
returns. We employ the Market Model and
We first divide the aggregate sample into the Market Adjusted Returns Model to
groups with cumulative takeover defense estimate these deviations for each stock.
measures, and then determine the average The Market Model supposes that the
ownership concentration for each group. return on an individual stock is linearly
We calculate ownership concentration of related to the market return.The relationship
a firm in several ways: the percentage is written as follows:
of shares held by the largest block- Rjt j jRmt ejt
holder7 (C1), the share of the three
largest blockholders (C3), and the share of where
all blockholders (Cblock). We also separately Rjt the continuously compounded
calculate a concentration measure (Cinst.)
return of stock j in period t;
to represent institutional ownership (esti-
mated by blockholdings held by major Rmt the continuously compounded
Dutch banks and insurance companies). On market return in period t;
the basis of a t-test we then find out j j security specific and time
whether average ownership concentration independent parameters;
significantly varies among groups of com- ejt the error term of stock j in
panies with different takeover defenses. period t.
The above analysis is performed by
comparing two sample averages at a time. The period to estimate the Market Model
In order to examine the effect of firms’ parameters is selected as the period of
ownership structure on the likelihood of 100 days before the start of the event (or
adopting individual takeover defense, we announcement) period. We also estimate
estimate the following logistic regression:8 the parameters using 100 days of data
from the postevent period. A period
p (defense measure) f (ownership of 20 days before the announcement until
concentration) 20 days after the announcement is selected
as the event period. The impact of takeover
Here the dependent variable is equal to 1 defense announcements on stock returns is
if a firm has a particular defense measure measured over this period. The parameters
and 0 otherwise. Several new proxies are are estimated by using the ordinary least
TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 653
squares method. We use the ‘CBS-Total The average abnormal returns and
Return Index’ to calculate the market 0cumulative average abnormal returns
returns used in the model.10 The abnormal are then computed as described previously.
return (also called excess return or predic-
tion error) is the difference between the
actual return during the event period (20, EMPIRICAL RESULTS
20) and the return predicted from the
estimation period:
Ownership structure
ARjt Rjt
ˆj ˆ jRmt A descriptive analysis on Dutch ownership
The abnormal returns for individual stocks structure is presented in Table 25.3. We
are then averaged across all stocks to find that blockholders hold more than half
obtain average abnormal returns for each of all shares in Dutch companies. The
day. The excess returns for each stock are average share of the largest blockholder is
also compounded over different time inter- 31 percent, that of the three largest block-
vals around announcement date to calculate holders is 45 percent, and the average share
cumulative abnormal returns. A t-test is of all blockholders together is 51 percent.
performed to test whether the average It appears that the group with the three
abnormal returns are significantly different largest shareholders dominates ownership
from zero. The t-value is obtained by divid- concentration of Dutch firms. The correla-
ing average daily abnormal returns by its tions between these variables are, as
standard deviation calculated from the expected, very high. Our results show that
estimation period. ownership concentration in the Netherlands
In order to check the robustness of our is higher than in the U.S.A., the U.K.
results, we also perform the stock return and Japan, but lower than in Sweden.11
analysis using the Market Adjusted Returns The variation within each measure of owner-
Model. The model predicts individual stock ship concentration is also higher in the
return to be equal to the corresponding Netherlands. The standard deviation of
market return, or in other words, percentage of shares held by the top five
blockholders in our sample is 26 percent
Rjt Rmt compared with Prowse’s (1995) findings
This model is distinct from the Market of 16 percent in the U.S.A. and the U.K.
Model in the sense that here all stocks are and 14 percent in Japan.
assumed to be of average risk. The abnor- Analyzing the distribution of sharehold-
mal returns are calculated as the difference ings, we find that the largest shareholder
between the actual stock return and the has more than 25 percent of shares in
corresponding market return: 52 percent of the firms in the sample, and
more than 50 percent of shares in 22 percent
ARjt Rjt Rmt of the firms. A majority of the companies
The table reports the results of different ownership concentration variables: C1, C3, C5, Cblock and Cinst. represent
the percentage of shares held by the largest blockholder, the three largest blockholders, the five largest
blockholders, all blockholders, and institutional blockholders, respectively.The sample consists of 177 industrial
companies listed on the Amsterdam Stock Exchange in 1992.
654 ON TAKEOVER AS DISCIPLINARY MECHANISM
has a blockholding in excess of 50 percent. 11 percent age points higher than that for
After searching the identity of these firms with two devices. Both differences in
blockholders, we find that the average concentration are statistically significant.
shares of management and family members, In general, we find that the lower the
companies, and individual blockholders ownership concentrations are, the more
are 8 percent, 20 percent and 5 percent, takeover defenses companies adopt. This
respectively. The average share of financial phenomenon is valid for all three measures
institutions (banks and insurance compa- of ownership concentration. Our evidence is
nies) in our sample is almost 10 percent. consistent with Bergström and Rydqvist
The combined share of these investors is (1990), who observe that Swedish firms with
less than 25 percent for 90 percent of high concentration of equity ownership rarely
the companies. The sample contains adopt antitake over devices. These findings
18 companies in which banks and insurance suggest that firms adopt multiple takeover
companies are the only blockholders. defenses when shareholdings are diffuse.
The average share of other institutional Table 25.4 also reports the results
blockholders is 6 percent. for institutional blockholders (banks and
In Table 25.4 we present the average insurance companies).The concentration of
ownership concentrations of companies these institutional shareholders does not
with cumulative defense mechanisms. show any particular relationship with
We also report in the lower panel corre- multiple takeover defenses. The share of
sponding t-values testing the difference in these investors in firms with one defense
average ownership concentrations. Our mechanism is five percentage points
results show that the concentration of the higher than in firms without any defense.
largest shareholder for firms without any Afterwards, as institutional ownership
defense measure is almost 13 percentage concentration declines, firms adopt a higher
points higher than that for firms with only number of defenses. These differences are
one measure. Similarly, for companies not statistically significant.
with one takeover defense device, the Next, we examine if the general finding
concentration of the largest shareholder is on the negative relationship between
Table 25.4 The difference in ownership concentration of firms with cumulative takeover defenses
Measures of concentration
Table 25.5 Estimates of logistic regressions relating the likelihood of adopting a specific takeover
defense mechanism to ownership concentration
negative abnormal returns dominates the The almost negligible stock price impact
sample.This is also found to be statistically with respect to the second step announce-
significant at the 5 percent level after ment is not surprising, since granting a
conducting a sign test (Z-statistic purchase option to a friendly party is an
2.34). Over the 6-day postannouncement obvious outcome of the charter amendment.
period, the cumulative abnormal return is Another interesting finding is obtained
2.27 percent (with a t-value of 2.97).14 when we look at the third step of the defense
Although the above result tends to process. The negative announcement effect
support the managerial entrenchment of the aggregate sample is in fact deter-
hypothesis, further analysis of the sample mined by the issue of the preferred share
reveals some interesting findings. We split itself. We find that the announcement of
the aggregate sample into three subsamples a preferred share issue is associated with a
based on the three steps followed in the strong excess decline in stock returns
issuing process. With the announcement of (4.09%), which is statistically significant
the first step towards defense (creating the (with a t-value of 4.94). For the 6-day
possibility of preferred share issue), a period [0, 5], the excess decline in share-
positive and statistically significant stock holders’ wealth amounts to 6.40 percent.16
price effect is observed. This evidence does However, we find a significant price increase
not support the managerial entrenchment before the announcement of the preferred
hypothesis. Shareholders do not experience share issue.This increase in share price could
any wealth decline from the charter be an indication of a takeover attempt
amendment leading to takeover defense. On that eventually led managers to issue
the contrary, they appear to benefit as there the preferred shares.17 Interestingly, the
is a signficant increase in stock returns post-announcement periods indicate a
(1.23% in 2 days).15 All other postan- significant decline in shareholders’ wealth.
nouncement intervals also reveal positive This decline might provide an estimate of
(but not significant) price increases. This the lost premium incurred by common
result indicates that defense measures are shareholders—since the chance of eventual
indeed adopted allowing shareholders to takeover was eliminated by actually issuing
benefit from increased takeover premiums. preferred shares.18 Our finding is consistent
Table 25.6 Cumulative abnormal returns around the announcement of defense with preferred
share issue
Return intervals
(20, 1) (0, 1) (1, 1) (0, 5) (20, 5) (0, 20)
with prior studies showing that stock prices when a firm is characterized by diffuse
increase with takeover bids but then decline shareholdings.We do not find any significant
if they do not materialize. In sum, the relationship associated with institutional
evidence provided here suggests that, stock ownership. The evidence provided
although defense measures are beneficial here, therefore, does not strongly support
to a certain extent, the benefits do not the hypothesis that institutional share-
remain when they are used to fend off holders provide better monitoring than
takeover attempts. other blockholders.
We also examine whether there is a We also conduct a stock return analysis in
difference in the results between the first the case of defense with preferred share—
preferred share issue and a subsequent the most widely used takeover defense
issue. The issue sample is further divided device in recent years in the Netherlands.
into a subsample of 13 companies that Our results indicate two opposing effects
issued preferred shares for the first time of defense on shareholders’ wealth: in one
and a subsample of four companies with a situation, the stock market reacts
subsequent issue. We find that the first positively, seemingly to allow managers to
issue is more damaging for shareholders. bargain for a higher premium in takeover
The abnormal return in the 5-day post- bids. In another situation, the stock market
announcement period is 7.03 percent reacts negatively as potential takeover
(t-value –3.88) in case of the first-time attempt appears to be eliminated.
issue, compared to 4.37 percent (t-value Alternative disciplinary mechanisms
2.71) in case of a subsequent issue. This have been an area of extensive scrutiny. In
difference is statistically very significant this paper, we document that low (high)
(with a t-value of 6.24).19 Other post- ownership concentration is associated with
announcement return intervals show greater (smaller) use of antitakeover devices
similar results. which affect the functioning of the market
for corporate control. We also provide
evidence on the existence of positive and
CONCLUSIONS AND negative share price effects of takeover
IMPLICATIONS defense measures. Some implications of
our findings are discussed below.
This paper empirically analyzes the The empirical results obtained in this
relationship of takeover defenses with study reconfirm the need of analyzing
firms’ ownership structure and shareholders’ firms’ ownership structure as a mechanism
wealth. A sample of Dutch industrial to control the agency conflict between
companies is selected for the study. The shareholders and managers. Other such
Dutch scenario is particularly interesting mechanisms include the capital market, the
because almost all listed companies have market for corporate control, the manage-
adopted multiple takeover defenses. rial labor market and the product market.
We find that firms with a relatively lower Similar to Walsh and Seward (1990), we
ownership concentration are the ones with believe that much can be learned about one
a larger number of defense measures. Our control mechanism when it is analyzed in
analysis suggests that firms with disperse and around another mechanism. Managers
ownership adopt more defense tactics. The in the Netherlands seem to be immune
analysis also shows that the likelihood for from the disciplinary threat of the market
a firm to adopt takeover defenses is inversely for corporate control. Since the takeover
and significantly related to ownership con- market is just one disciplinary mechanism,
centration. The result is robust to different we would expect other control mechanisms
ways of measuring ownership concentration. to be at work too. Our results in this
Overall, our evidence is consistent with the paper demonstrate this—for example,
hypothesis that company management is monitoring by concentrated ownership. As
more likely to adopt defensive measures Prowse (1995) points out, concentrated
658 ON TAKEOVER AS DISCIPLINARY MECHANISM
shareholdings are important because they substituted by another, these do not imply
provide investors with both the incentive one mechanism being better than another.
and the ability to monitor and influence the Analyzing the cost—benefit aspects of
management. Without such concentration, each monitoring mechanism is an issue
again other mechanisms of corporate which was beyond the scope of the paper.
control must be relied upon. According to Moerland (1995), there exists
It is usually believed that institutional a trade-off in the sense that the advantages
investors find it in their best interest to of one mechanism will generally be the
more effectively monitor company managers. disadvantages of the other. Questions also
In the U.S.A., institutional shareholdings remain relating the effectiveness of moni-
have increased over the last years, and a toring with corporate performance. Are
few institutional shareholders have emerged firms with concentrated ownership more
as very active monitors. The findings of profitable than firms in which ownership is
Duggal and Millar (1994) suggest that not concentrated? If incumbent managers
researchers should better split aggregate use defense tactics to deter takeovers,
institutional ownership into different cate- are they completely insulated from the con-
gories to obtain correct results. The results sequences of poor firm performance? The
of this study show that in the Netherlands usual finding has been that takeover threat
institutional shareholders like banks and improves performance of a firm. But, as has
insurance companies do not have large been mentioned by Walsh and Seward
holdings, and these have no relationship (1990), if managers in a poorly performing
with the adoption of antitakeover devices. firm anticipate a takeover contest, they
An implication of this finding is that active may adopt antitakeover devices to protect
monitoring by institutional shareholders them. On the other hand, managers could
may not take place in many countries. It is be more frequently dismissed when they
highly unlikely that Dutch institutional perform poorly because another disciplinary
shareholders lack the expertise and the mechanism—the managerial labor market—
ability to serve as effective monitors. then starts functioning. A further analysis of
Rather, the presence of small stakes may the role of different corporate governance
explain why passivity remains the norm. It mechanisms represents an interesting area
is also possible that active institutional of future research.
monitoring may not be a representation of
the general pattern in the U.S.A.
There are many types of defense
measures, and their effects also depend ACKNOWLEDGEMENTS
on situation like the manner in which a par-
ticular device is introduced. Our analysis We would like to thank Tom Berglund,
shows that it is difficult to say a priori Associate Editor Richard Bettis, Marc
whether defense measures are good or bad Zenner and two anonymous referees for
for shareholders. Even takeover defenses useful comments and suggestions. We have
that are approved by shareholders have the also benefited from the comments of semi-
potential to reduce shareholder wealth. nar participants at Tilburg University, the
This lack of conclusive evidence regarding Workshop on Corporate Finance at the
the stock price reaction could also be European Institute for Advanced Studies in
attributed to the limitations of event- Management (Brussels), the International
study methodology and the additional Meeting of the French Finance Association
need of analysis using other characteris- (La Baule), the Symposium on Money,
tics of firms that propose takeover defense Finance, Banking and Insurance (Karlsruhe)
measures. and the Eastern Finance Association
Although the results presented here show Meeting (Boston) where previous versions
that one monitoring mechanism has been of this paper were presented.
TAKEOVER DEFENSES, OWNERSHIP STRUCTURE AND STOCK RETURNS 659
activity, and firm value: Evidence from the 1989 Honey well proxy solicitation’, Journal
death of inside blockholders’, Journal of of Financial Economics, 34, pp. 101–132.
Finance, 48, pp. 1293–1321. Voogd, R. P. (1989). ‘Statutaire beschern-
Song, M. H. and R. A. Walkling (1993). ‘The ingsmiddelen bij beursvennootschappen’,
impact of managerial ownership on acquisition doctoral dissertation, University of Nijmegen,
attempts and target shareholder wealth’, Kluwer.
Journal of Financial and Quantitative Walsh, J. P. and J. K. Seward (1990).
Analysis, 28, pp. 439–457. ‘On the efficiency of internal and external
Van Nuys, K. (1993). ‘Corporate governance corporate control mechanisms’, Academy
through the proxy process: Evidence from the of Management Review, 15, pp. 421–458.
Chapter 26
ABSTRACT
This article uses clinical evidence to show how the German system of corporate control and
governance is both more active and more hostile than has previously been suggested. It
provides a complete breakdown of ownership and takeover defence patterns in German listed
companies and finds highly fragmented (but not dispersed) ownership in non-majority
controlled firms. We document how the accumulation of hostile stakes can be used to gain control
of target companies given these ownership patterns.The article also suggests an important role
for banks in helping predators accumulate, and avoid the disclosure of, large stakes.
benefits: minorities are typically bought out opportunity for hostile control changes to
at a premium to the block price, despite the be effected via stakebuilding and for such
absence of any legal compulsion (Barclay behaviour to be profitable. Such loose reg-
and Holderness, 1991, 1992). In contrast, ulation exists in a number of continental
in Germany, minorities are vulnerable to European countries including Germany.
expropriation by majority owners: there is Hence, the behaviour that we focus on in
no legal requirement to buy-out minority this article is, in a sense, quite predictable,
stakes, and when offers to minority share- although no previous paper has, to our
holders are made, they are typically at a knowledge, provided systematic evidence
large discount to the price paid by the on the importance of hostile acquisitions in
controlling shareholder. Germany via stakebuilding. Our article com-
There are some important differences plements that of Franks and Mayer (2001),
between the block trades discussed in the who consider the importance of share
US literature and our cases. First, US Secu- stakes in Germany. They document a large
rities and Exchange Commission (SEC) incidence of stake changes and suggest that
regulations have tended to result in block such changes are the most significant influ-
trades being negotiated and publicised ence on the turnover of supervisory board
rather than covert. Second, partial acquisi- members. However, Franks and Mayer con-
tions or trades of existing blocks take place sider the incidence and observed response
against the background of an ownership to sales of share stakes in general, without
structure that is significantly more dispersed distinguishing between the nature of these
in the US than in Germany, particularly for sales. In contrast, we focus exclusively on
large firms.6 Therefore, the economically hostile stakebuilding, where the objective is
significant block size is likely to be differ- to influence control.
ent in the US and Germany, and should The remainder of the article is organised
depend on the concentration and composi- as follows. In Section 2, we start by dis-
tion of the remaining ownership structure. cussing the rules governing corporate con-
Third, three-quarters of the US cases trol in Germany, in particular, the control
involve a simple sale of one existing block rights associated with share blocks of vari-
(Barclay and Holderness, 1991), while our ous sizes. We also provide a snapshot of the
article documents a much richer dynamic actual ownership patterns in Germany, which
of blocks being built from open-market demonstrates the importance of large
purchases acquired from existing block- share blocks. In Section 3, we outline our
holders and combined with other blocks. In research methodology and data sources.
that sense, we look at block-building, rather Section 4 presents evidence on hostile stake
than block transfers. There has, to date, accumulation for a number of recent cases,
been little analysis of block accumulation detailed chronologies of which are provided
tactics, in the US or elsewhere.7 in Appendix A (in alphabetical order by
Finally, our article is also related to the target company, for ease of reference).
literature on optimal takeover regulation, Section 5 concludes.
since we focus particularly on the ability to
build hostile stakes without the knowledge
of other market participants—something 2. OWNERSHIP AND CONTROL OF
that the existing block literature assumes GERMAN COMPANIES
away. For example, as Burkart (1996)
notes, it is critical whether stakebuilding, Ownership of German companies is highly
and hostile intentions, can be obscured concentrated, certainly in comparison with
because as soon as such a strategy becomes US or UK companies. We document later
transparent, free-riding (along the lines on in this section just how concentrated
suggested by Grossman and Hart, 1980) is ownership structures are, but we begin by
likely to result. Loose disclosure and regu- explaining the control rights associated
lation of acting in concert will provide the with different-sized share blocks in Germany.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 665
In many cases, the hostile stakebuilding contracts have no special legal standing,
that we document later in the article does and are typically renewable every 4–5 years.
not involve building a majority stake, but We start by considering the rights asso-
rather takes the form of building a stake with ciated with dispersed shareholdings. In
sufficient control rights to exert a major Germany, no special rights are attached to
influence on the incumbent managers and share blocks of below 25%. Indeed, in
other shareholders. The rules regarding common with many other EU countries,
these control rights therefore play a critical small minority shareholders are afforded
role in hostile stakebuilding strategies. little in the way of protection, either from
expropriation by controlling shareholders,
or equal treatment. For example, there is
2.1. Control rights no legal requirement to buy-out minority
There are five economically significant stakes when a party acquires a stake in
block sizes as defined by the rights of excess of 50%. While the recently intro-
minorities and the discretionary powers of duced, and voluntary, Takeover Code does
the dominant blockholders. These are sum- introduce such a requirement, it is notice-
marised in Table 26.1. One important fea- able that the offer to minority shareholders
ture of corporate control in Germany is the can be at a discount (of up to 25%) to the
existence of pooling contracts. Pooling price paid in acquiring the controlling stake
contracts are agreements between share- (see footnote 27 for more details).
holders that oblige them to pool their votes. The first important threshold in terms of
Hence, share blocks can be built directly control rights applies to blocks of 25% or
via purchase and/or indirectly via a pooling more. Such a shareholding is referred to as
contract. Either way, the control rights a blocking minority and gives veto powers
summarised in Table 26.1 would apply to on corporate charter amendments, supervi-
the single or pooled block. The pooling sory board changes, and profit transfer and
25% No major formal rights other than voting pro rata. Shareholders who control
at least 5% of the equity capital can demand that an annual general meeting
be called.
25% Blocking minority control. Power of veto over corporate charter amendments,
supervisory board changes, and profit transfer and control agreements. Stakes
above 24.9% must be reported to the Cartel Office and to the target company,
which in turn has a duty to publish a notification in certain newspapers.
50% Management control, although the existence of another block of 25% would
limit discretion.
75% Super-majority control. Enables the blockholder to amend the corporate
charter, change the composition of the supervisory board, and enter into
profit transfer and control agreements. The lack of any blocking minority
(25%) stake thus significantly increases effective control.
95% The blockholder can force a merger and minority shareholdings can be
compulsorily purchased. The law requires that minority shareholders be paid
‘adequate’ compensation in the form of cash or shares in the parent company.
The level of compensation is largely left unregulated.
666 ON TAKEOVER AS DISCIPLINARY MECHANISM
control agreements. The latter agreements prices have nothing to do with fundamental
are an important way by which the claims value—which can ‘only’ be found by
of minority shareholders can be diluted. accountants acting as expert witnesses.
Subject to ensuring the economic survival Even when the courts do find in the plain-
of the dominated company, a controlling tiffs’ favour, the average imposed increase
shareholder can dilute minorities in a vari- in bid prices or guaranteed dividends of
ety of ways. First, group losses can be 25% (Wenger and Hecker, 1995) is not
foisted disproportionately onto minorities, sufficient to close the discount to the mar-
while profits can be transferred out of the ket price before the buy-out announcement.
dominated company. Second, hidden reserves Finally, the law explicitly gives the majority
(typically land and share stakes whose owner the right to cancel the control and/or
book value is below market value) can profit transfer agreement if the court’s
be sold and the proceeds transferred to the decision is unfavourable.11
majority owner. Third, group assets can be Hence, the ability to veto a profit trans-
bought and sold at prices that are advanta- fer and control agreement represents an
geous to the dominant shareholder.8 Fourth, important control threshold. It is also
while minority shareholders must be offered worth noting that, in the absence of other
the alternative of a guaranteed fixed divi- large shareholders, a 25% stake can in
dend should they wish not to sell out to the practice provide more than mere veto influ-
controlling shareholder, neither the bid ence, given an average presence of no more
price nor the guaranteed dividend need in than 57% of votes at AGMs (Baums and
any way be related to the share price, or the Fraune, 1994).
price the bidder paid to acquire control.9 The next threshold in terms of control
Both options are, in most cases, relatively rights occurs with blocks of 50% or more.
unattractive: Wenger and Hecker (1995) Such a block gives management control of
show that for 45 buy-out bids made to the company, but is subject to limits on the
minority shareholders between 1983 and controlling party’s discretion due to the
mid-1992, the price offered was 27.1% existence of a blocking 25% minority. In
below the market price 2 days before the practice, 50% may not be enough to dis-
announcement. miss incumbent management quickly if the
Minority shareholders have no right of supervisory board is not on the controlling
legal appeal against any such transfer of party’s side: it is the supervisory board which
assets or profits per se, but can litigate appoints and dismisses the management,
against the level of compensation (dividend and the supervisory board in turn can only
or buy-out bid) offered within 2 months of be dismissed with a 75% super-majority.
the date of the offer. Many such offers Given the various control rights that
(such as the Boge case we examine below) accrue to blocking minority stakes of 25%
do indeed result in lengthy litigation during or more, the next important control thresh-
which minority shareholders attempt to old occurs with blocks of more than 75%.
increase the attractiveness of the terms Such super-majority blocks give the
they are offered. The average court case controlling party complete discretion in
takes 5.3 years to conclude (Wenger and matters of supervisory board elections and
Hecker, 1995).There is, in general, no right profit transfer and/or control agreements.
of final appeal to the Supreme Court, However, the incremental control rights
which has led to different second-level associated with blocks of 75% accrue
courts of appeal passing contradictory ver- entirely from the impossibility of a rival
dicts, all of which have binding character. blocking minority stake, rather than
Neither the law nor the courts have estab- additional legal rights. The final control
lished any consistent framework for valuing threshold occurs with 95% ownership,
a company or assessing its risk.10 As to when the remaining minority share-
offers being made below trading prices, the holders can be compulsorily bought out on
courts have adopted the line that market disadvantageous terms.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 667
Source: Own calculations based on “Saling, 1992,” Hoppenstedt’s stock market yearbook.
Notes: The law requires disclosure of blocks of more than 25% and more than 50%. Frequently, smaller blocks
are also disclosed (as in Panel C). Non-disclosure need not imply non-existence.
Table 26.3 Defensive share structures of German stock exchange listed companies (1991)
Grand total 37 21 25 32
(% of all German listed companies) 6.6% 3.8% 4.5% 5.7%
Source: Own calculations based on “Saling, 1992,” Hoppenstedt’s stock market yearbook.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 669
Table 26.4 Potential targets for hostile stakebuilders
Notes: This table is based on an analysis of the companies which appear in Panels B and C in Table 26.1. In the final column, the number of potential takeover targets
was reduced if and only if the existing limits to share transferability or multiple voting rights applied to a sufficiently large number of votes to effectively rule out hostile
stakebuilding.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 671
First, a hostile bidder could attempt to build stock, one had sufficiently privileged stock
stakes in firms that have a free float of to block hostile approaches. Depending on
50% or more, without buying any existing the effectiveness of each of these takeover
blocks.The companies that potentially fall into barriers, the number of firms at risk
this category are widely held firms; compa- from hostile stakebuilding could fall to 110
nies with one blocking-minority stake; and (if all takeover strategies are considered)
firms with one or more non-blocking- or as low as 64—just 11% of all German
minority stakes. Both Mannesmann and listed firms—if only the first two strategies
Thyssen belong in this category. Checking are considered realistic.
for the presence of additional disclosed
stakes and computing free float, we find 79
companies with a free float of 50% or
3. DATA AND METHODOLOGY
more.13 Second, a bidder could take over a
company by buying one or more or all
existing non-blocking stakes; there were The cases for our clinical analysis were
seven firms potentially at risk from this identified using the Financial Times
strategy. A third takeover strategy is to Mergers and Acquisitions (FTMA) data-
buy-out one existing blocking-minority stake base. FTMA contains structured templated
and to make up the difference to 50% via information on changes in the ownership
open-market purchases. Thirty firms had structure of European firms, covering
one or two blocking minority stakes and (i) listed and unlisted firms; (ii) takeover
enough free float for this strategy poten- bids and stake purchases; and (iii) transac-
tially to allow a majority takeover.The final tions which are completed as well as those
strategy involves a bidder buying out two merely rumoured or still under negotiation.
existing blocking minority stakes, for All ‘forms’ compiled by FTMA between
instance, in companies which have two or January 1988 and December 1996 were
three such stakes and too little free float to obtained for transactions involving German
make any of the other strategies viable. companies as targets. This yielded 2511
Therefore, of the 141 non-majority con- forms, some of which pertain to the same
trolled, some are much more vulnerable to firm at different points in time, or to differ-
hostile stakebuilding than others. In prac- ent bidders for the same firm at the same
tice, the first two strategies look much more point in time.Two types of filters were used
feasible than the last two, suggesting the to manage this very large amount of raw
potential takeover targets might number data. ‘Negative’ filters were used to
around 86, which represents just 15.4% of eliminate, without further investigation,
the total number of listed German firms. FTMA-reported transactions with a low
If the defensive features summarised in likelihood of being motivated by ‘hostility’.
Table 26.3 are effective, some of these 86 These negative filters were:
firms are unlikely to be taken over.Table 26.4
takes this into account by looking in detail 1. cases of apparent initiation of cross-
at the nature of any takeover defences.Two shareholdings (which require both
companies, for instance, listed only non- parties’ agreement)
voting preference shares, making them vir- 2. “participation in capital increases
tually immune to hostile stakebuilding.14 A via subscription” (which presumably
further 20 had voting right restrictions, are agreed between buyer and tar-
though we will argue later that the empiri- get, and thus not hostile)
cal effectiveness crucially depends on the 3. unlisted companies other than those
level of the cap. Of the 11 companies with captured using a positive filter (see
limited share transferability, eight below)
restricted transferability sufficiently to 4. complete or partial disposals of
make a bidder’s life difficult. Finally, of the divisions or other operating units
21 companies that had multiple classes of 5. privatisations by the Treuhandanstalt
672 ON TAKEOVER AS DISCIPLINARY MECHANISM
requirement is to the target company in the industry cases. In only two instances are
first instance. Second, the way blocks are the control contests of a conglomerate
acquired is often, and deliberately, covert, nature.This suggests that the bidders’ moti-
even where formal notification would in vation may often be industrial or strategic
principle be required. Finally, block trades rather than disciplinary: stakebuilders may
often receive press coverage only once seek to reduce overcapacity in their indus-
further buying activity or target responses try or increase market power rather than
add up to a pattern, resulting in substantial ‘punish’ wayward managers. Since target
reporting delays. managers would expect to lose their jobs
whatever the stakebuilders’ motivation,
their resistance to the control changes is
4. THE CASE STUDIES not surprising. The strong horizontal (and
vertical) bias results in the Cartel Office
The 17 cases of hostile stakebuilding pro- being involved in a relatively high propor-
vide an insight into the way corporate gov- tion of the cases. Indeed, the Cartel Office’s
ernance is exercised in Germany. Brief frequent involvement supports the view
summaries of each individual case are that market concentration or consolidation
provided in Appendix A. In this section, we motives play an important role in our cases.
first describe the sample companies in The final column of Table 26.5 attempts
terms of firm characteristics, shareholder to categorise the stakebuilders’ apparent
structure, and defences in place. We then motivations. In their study of the US block
turn to the various strategies employed by market, Bethel et al. (1998) distinguish
hostile stakebuilders, including the forma- between ‘activist’ block-buyers (usually
tion of coalitions, the role of banks and individuals whose announced intention is to
ways to avoid disclosure. We next consider influence firm policy) and ‘strategic’ block-
the defensive actions taken by targets, buyers (other companies whose purchases
including voting restrictions, denial of are unopposed by target management). Our
board representation and the use of defen- stakebuilders do not easily fit this distinction:
sive coalitions and white knights. Finally, most of the stakebuilders are companies—
we discuss the outcomes of the different not individuals as documented in the US by
types of control battle that are observed. Holderness and Sheehan (1985) and
Bethel et al. (1998)—but we cannot call
them ‘strategic’ since they were opposed
4.1. Company characteristics by target management and announced
‘activist’ style intentions. The classifica-
Table 26.5 lists the 17 targets and their tions we offer are instead based on the
respective suitors.The control contests were extent to which the stakebuilders sought to
initiated between 1987 (Axel Springer become involved—ranging from gaining
Verlag) and 1994 (Kolbenschmidt) though, influence over corporate decision-making
due to the often covert nature of stake- via blocking minority stakes (as in the AMB,
building, precise dating is difficult. The tar- Buderus, Th. Goldschmidt and Holzmann
gets are invariably from ‘traditional’ cases) to gaining majority control (as in
industries, ranging from construction the Axel Springer Verlag case) or effecting
(Philipp Holzmann and Dywidag) to insur- a takeover (as in the Asko, BIFAB, Boge,
ance (AMB and DBV). One of the most Bopp and Reuther, Conti, DBV, Dywidag,
noticeable characteristics of the case stud- Hoesch, and Kolbenschmidt cases). We
ies is that the stakebuilders mostly operate label six of the nine intended takeovers as
in the same industry as their targets. ‘consolidating takeovers’ because the
Indeed, many cases are reminiscent of stakebuilders—all of whom operated in the
Jensen’s (1993) over-capacity argument, for same industry—made public statements
instance, the two construction cases, the concerning the need to remove overcapacity
Krupp–Hoesch steel case, or any of the car (steel being an obvious example). The large
Table 26.5 Stakebuilders and targets
Industrial Stakebuilder’s
Base Country relationship apparent
year a Target company Industry Stakebuilder of origin Industry with target motive
1991 AMB Aachener und insurance AGF France insurance horizontal gaining influence over
Münchener Beteiligungen corporate decision-making
1992 Asko retailing Metro Switzerland retailing horizontal consolidating takeover
1987 Axel Springer Verlag media/publishing Kirch group vs. Burda all from Germany media/publishing horizontal gaining control over
brothers vs. Springer heirs corporate decision making
1988 Bibliographisches Institut publishing Maxwell UK media/publishing horizontal takeover
and F.A. Brockhaus (BIFAB) Communication
1988 Boge car parts Sogefi Italy car parts horizontal consolidating takeover
1989 Bopp and Reuther process engineering IWKA Germany process engineering horizontal consolidating takeover
1994 Buderus heating systems/ Bilfinger and Berger Germany construction vertical “long-term strategic
metal products investment” via 25%
block
1989 Continental tires Pirelli Italy tires horizontal consolidating takeover
1993 Deutsche Beamten– insurance Commerzbank/Winterthur Germany/ bank/insurance horizontal takeover
674 ON TAKEOVER AS DISCIPLINARY MECHANISM
a Base year is defined as the year in which the control contest began.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 675
number of such ‘consolidating takeover’ bids Feldmühle Nobel and Hoesch). This is not
lends further support to the importance of surprising as such caps clearly make it
‘industrial’ rather than disciplinary motiva- more difficult (although as we shall see
tions as the primary driver of hostile stake- later, not impossible) for a stakebuilder to
building. Only the two cases involving establish a controlling interest.
targets that were conglomerates appear to The final block of Table 26.6 provides
have been motivated by a desire to remove some performance indicators prior to the
‘underperforming’ management. Finally, control contests. The targets are mostly
APV’s stakebuilding in the SEN case is large companies, with nominal annual sales
perhaps best understood as an instance of averaging DM 3.6 billion in the 3 years
intended greenmail, as we will show later on. prior to the control contests. The smallest
A final striking feature of the stake- target (BIFAB) had annual sales of DM 44
builders is their nationality: nine cases million, and the largest target (Asko) had
involved non-German stakebuilders, includ- DM 13.2 billion. All but four of the targets
ing most of the earliest cases. This is con- had experienced positive sales growth in
sistent with ‘outsiders’ undermining the the 3 years leading up to the control
prevailing ‘governance consensus’. contests, with double-digit growth rates in
In line with our discussion in Section 2.1, five cases. The earnings dynamic is more
in Table 26.6, we classify the target mixed, with three firms moving into losses
companies according to their shareholding over the 3 years pre-contest, three firms
structure in the run-up to their control con- experiencing negative earnings growth, and
tests.Three of the companies were majority the remaining 11 seeing sometimes quite
controlled prior to the control contest. substantial increases in earnings. In the
SEN had a 50.01% owner but still became year their control contests began, four
the target of a control contest. Wünsche firms were loss-making and six firms
was majority controlled by two brothers earned less than an 8% return on equity.
who later fell out. Bopp and Reuther, the The remaining seven, however, would appear
only non-listed target in our sample, was to be performing healthily, with return on
controlled by a large number of family equity ranging from 10% to 22%.
shareholders and one outside 25%
blockholder, who originally pooled their 4.2. Stake accumulation tactics
votes. Of the remaining 14 companies, there
was a blocking minority shareholder in six Hostile stakes can be formed either via
cases, and in the remaining eight companies, open-market purchases or by purchasing
ownership was initially dispersed. blocks from existing shareholders. As
The middle block of Table 26.6 docu- Table 26.7 shows, stakes are often built
ments the takeover defences in place in the partially through open-market purchases,
run-up to the control contests. Eight of the but limited secondary market liquidity typ-
17 targets had some form of takeover ically precludes a controlling stake being
defence. Not surprisingly, the three majority accumulated without also buying (or form-
controlled firms in Panel A had no formal ing coalitions with) existing blocks. In con-
takeover defences. Amongst the six trast to the predictions of the theoretical
companies in Panel B, which initially had literature (such as Zwiebel, 1995) there
blocking minority shareholders, two are usually multiple large blockholders in
(Axel Springer Verlag and DBV) limited our cases. This greatly increases the com-
the transferability of shares, and one plexity of the game for each individual
(Th. Goldschmidt) had differential voting blockholder. In particular, each blockholder
rights. The most popular defence for will fear not being part of a controlling
companies with fragmented ownership coalition when a hostile stakebuilder
structures appears to be caps on voting emerges, given the high risk of being diluted
rights, which were in place in four of the eight in Germany. It is not altogether surprising,
companies in Panel C (Asko, Continental, therefore, that in a large number of cases
Table 26.6 Pre-contest target company ownership structure, takeover defenses, and performance
Source: accounting data comes from Hoppenstedt Bilanzdatenbank and various issues of the Saling stock market yearbook.
a None of the target firms we analyse have only non-voting shares listed.
b Measured as the average in the base year and the previous 3 years.
c Measured as the average growth rate in the base year and the previous 2 years.
d 1.3% of shares have 11.6% of votes.
e Adopted in response to stakebuilding rumours.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 677
Table 26.7 Stakebuilding strategy
Bank
assists
Acting in Bank stake- Disclosure
Target Source of stake concert stake builder problems
existing pooling agreements break down in given the horizontal nature of many of
the face of a hostile stakebuilder. For exam- these takeovers, the threshold beyond
ple, in the Th. Goldschmidt case, the family which Cartel Office approval is required.
shareholders had a pooling agreement with Disclosure problems are of particular
Allianz that gave the coalition majority concern since most German companies—
control. When VIAG emerged as a stake- unlike British or American ones—issue
builder, Allianz let its pooling contract expire bearer shares, making it hard to identify
and sold out to VIAG, who thereby gained one’s shareholders. During the period of our
control. In the Bopp and Reuther case, the investigation, company law required only
pooling agreement that established major- holdings in excess of 25% and 50% to be
ity control was actually violated by certain revealed. This contrasts with the 5%
shareholders, which precipitated the control threshold in the US (Regulation 13-D) and
battle in the first place. the 3% threshold in the UK. The ability of
It is also interesting to observe the role a predator to build a stake without other
of banks during the control contests. As market participants realising is important
Table 26.7 shows, banks held significant as it reduces the likelihood of other share-
stakes in 12 of the 17 cases, and often holders eliminating potential takeover
sided with the predator (either through sale gains by free-riding,17 and it prevents the
or pooling). In some cases, the bank stakes potential target from initiating some form
appear to have been held for a long time, of anti-takeover protection such as limiting
although there are also examples where the votes that can be cast by any individual
investment banks essentially built stakes in shareholder. Germany’s disclosure rules
partnership with a predator, and appear to have been progressively tightened in recent
play a pivotal role in brokering a control- years,18 but it is unclear whether this in
ling coalition. Perhaps the most interesting itself limits the ability to build a secret
example is the Hoesch case, where three stake: as our cases demonstrate, banks or
banks held significant stakes: Deutsche Bank other friendly parties can act in concert to
(Hoesch’s house bank);WestLB (Krupp’s— help companies build large stakes by com-
the stakebuilder’s—house bank) and Credit bining a number of smaller stakes that indi-
Suisse (who was acting for Krupp). Credit vidually do not have to be disclosed. Note
Suisse, on behalf of Krupp, secretly accu- also that the law requires no notification
mulated an initial 24.9% stake. Later in for block increases between 25% and
the battle, Credit Suisse accumulated a fur- 49.9%, so a stakebuilder could secretly
ther 20% stake, pledging it in support of build a near-controlling block.
Krupp. While WestLB did not declare its Our case studies show how weak, in prac-
12% stake in support of Krupp during the tice, disclosure rules are in Germany. For
battle (to avoid voting restrictions), it later example, by splitting up his 38% stake
worked with Krupp to have voting restric- between himself and some of his companies
tions removed and seal the takeover. and two children, Walter avoided revealing
Interestingly, Krupp also enjoyed the support his ownership interest in Dywidag. In the
of Deutsche Bank—Hoesch’s house bank— Feldmühle Nobel case, the Flick brothers
which was estimated to control around 12% were able to assemble a secret 40% stake
of the votes (partly via proxies).16 via open-market purchases and undisclosed
The secretive nature of stakebuilding in agreements with institutional blockholders.
the Krupp case is a common feature of our Indeed, having sold this stake to Veba, the
case studies. There is clearly much to be Flick brothers secretively assembled a second
gained from not alerting other sharehold- stake, which they used to extract further
ers that a significant stake, or coalition of surplus from Veba. Effective stakes can also
stakes, is being assembled.Two particularly be hidden by the use of option contracts, such
relevant considerations in this respect are, as in the Axel Springer Verlag case, where
first, the rules governing disclosure of the predator had a disclosed stake of 10%
stakes and acting in concert, and second, but an undisclosed option on a further 16%.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 681
DBV
Th. Goldschmidt ✓ ✓
Kolbenschmidt ✓ ✓
on the supervisory board. Initially Veba and significantly, at no stage declared its support
VIAG were demanding a “change of direc- for Krupp—thus avoiding accusations of
tion” but their frustration ultimately acting in concert. Surprisingly, Deutsche
resulted in VIAG assembling a controlling Bank, who controlled about 12%, also
stake, without offering to buy-out the fam- sided with Krupp despite being Hoesch’s
ily blockholders who had denied it a seat on house bank. Hence, this case illustrates the
the supervisory board. significant difference between voting
A more potent form of defence, fre- restrictions of 5%, which can be a major
quently employed by companies with frag- impediment to hostile stakebuilders, and
mented ownership, can be the use of voting restrictions of 15%, which are typically
restrictions. While it is not unheard of that much less effective in defending a target
a company institutes voting restrictions company.22 It also illustrates that a target
after a hostile stakebuilder has emerged, as cannot, as a matter of course, rely on its
in the Feldmühle Nobel case, more usually house bank for its defence. In general, proxy
restrictions are already in place.20 Such votes give banks both substantial influence
restrictions typically limit the proportion of (because voting restrictions do not apply to
votes that can be cast by any individual proxy votes which are ultimately not owned
shareholder or in some cases, groups of by the banks themselves) and valuable
shareholders operating under a pooling information about changes in ownership,
agreement. However, voting restrictions can which could be used to obstruct, or accel-
also take the form of limited transferability erate, a bid’s progress. However, proxies are
of voting rights. For example, Continental not always effective. While Deutsche Bank
had a restriction that no shareholder could successfully used its proxy votes to defend
exercise more than 5% of the votes. Not Continental against Pirelli, its proxies
surprisingly, this resulted in numerous proved ineffective in defending Feldmühle
blockholders holding precisely 5% blocks. Nobel once shareholders themselves had
When the Pirelli bid emerged, a group of decided to accept Stora’s bid.
minority shareholders demanded an extra-
ordinary general meeting to remove the 4.4. Outcomes
voting right restriction.This required a sim-
ple majority and was duly carried. However, Classifying the outcomes of the control
this decision was later overturned in the contests requires some care, as the ‘success’
courts when Continental alleged that Pirelli of a stakebuilder’s control challenge must
had acted in concert with its allies.21 The be measured against his initial intentions.
predator group should, therefore, have been For instance, where stakebuilders intended
subject collectively to a 5% voting limit. to gain ‘influence’ (or, perhaps, co-operation)
Continental continued to deny Pirelli and rather than majority control, a measure of
its allies votes in excess of 5%, and this their success is whether they gain seats on
restriction was clearly critical in thwarting the supervisory board, or whether they can
the attempted takeover. effect changes in management or corporate
Voting restrictions appear, however, to be strategy. In this spirit, Table 26.9 sum-
much less potent as a defence when they marises the outcomes of the 17 control
are set at slightly higher levels. For exam- contests. Judged relative to the initial
ple, the 15% voting right restriction in intentions, the original hostile stakebuilder
Hoesch’s corporate charter was not an was successful in 10 of the 17 cases. It is
effective defence against Krupp. The 15% noticeable that there is no apparent
restriction meant that it was only neces- relationship between initial ownership
sary for Krupp (with its 24.9% stake) to structure and the success of the stake-
convince two other significant blockholders building: the two main reasons for failure in
to vote in favour of removal of the voting our cases are either the intervention of the
restriction. One of these was WestLB, its Cartel Office or the appearance of a second
house bank, who had a 12% stake but, ‘white knight’ stakebuilder.
Table 26.9 Outcomes
remaining stakes
SEN rival stakebuilder possibly stake- control management rival bidder yes, n.a. 1.6 0.3 0.0
opposed greenmail builder change board vetoed at 18%
controlling succeeded opposed seats discount
shareholder forced for
Klöckner Werke, merger controlling
but ultimately shareholder
sold its stake to
the latter
Wünsche a family feud disciplining stake- control CEO 21.2 24.0 1.5 2.1
resulted in one target builders change resigned
brother being management succeeded
removed from
the management
board. He then
sold his 26%
stake to two
outsiders, who
pressured the
CEO to resign.
Rumours about
the CEO change
led to 28%
share price rise.
The stakebuilders
sold out once the
CEO was removed
Source: for accounting data: Hoppenstedt Bilanzdatenbank and various issues of the Saling stock market yearbook.
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 691
An alternative way to assess the outcome changes resulting directly from the control
of the control contests is to look at actual contest.We find senior management changes
changes in control. Consider first the three in seven23 of the 17 cases.This rate of post-
companies that were initially majority con- contest management turnover is roughly
trolled (see Panel A). It is noticeable that comparable to the finding of Bethel et al.
control changed hands in each case, which (1998) that about 22% of US CEOs are
demonstrates that pooling agreements can replaced following ‘activist’ or ‘strategic’
break down when rival stakebuilders appear, block purchases. German CEOs appear to
and that blocking minority stakes can have be at risk even when the stakebuilders
considerable influence even when there is a ultimately do not gain control, as in the
controlling stake or coalition. The stake- Continental case.
builders’ apparent motivations in this group How were minority shareholders treated
were very heterogeneous, ranging from in those cases where majority control was
IWKA’s consolidating takeover bid to APV’s gained? As noted in Section 2, German
greenmail and the disciplinary actions corporate law neither requires that minor-
taken in the Wünsche case. Four of the six ity shareholders be made a buy-out offer
targets in Panel B, which initially had one nor that they receive ‘equal treatment’ in
or more blocking minority stakes (but no the terms offered. As can be seen from
controlling stake), saw control change, Table 26.9, in the few cases where tender
though in two cases it was parties other offers were made to minority shareholders,
than the original stakebuilders who gained such offers have typically been at a signifi-
control. In one case (Axel Springer Verlag) cant discount to the price offered to block
the target finally agreed to ‘co-operate’ sellers in the course of gaining control.
with the stakebuilder. In the final case Specifically, in three of the five cases where
(Buderus) the Cartel Office effectively offers were made to minorities, the discounts
blocked a change in control (the Cartel ranged from 15% to 64%. Only in one
Office also blocked T & N’s stakebuilding in case (BIFAB) were minorities offered a
Kolbenschmidt, although this simply led to premium to the market price, of 10%. The
control passing to a third party). implications of the rules regarding minori-
Of the eight cases in Panel C, where the ties can be seen in the takeover of SEN by
initial ownership structure was fragmented, KW. Rival stakeholder APV had no realistic
the initial stakebuilder took control in three chance of taking majority control of SEN
cases, a rival stakebuilder took control in since KW owned 50.01%. However, its 40%
two cases, and in one case (AMB) the tar- stake had considerable value as it could be
get company agreed to cooperate with the used to block KW’s merger with SEN.
stakebuilder. In the remaining two cases Blocking minority stakes can, therefore,
(Holzmann and Continental), control did potentially be used for greenmail: in the
not change hands. In the Holzmann case, event, KW bought APV’s stake at a
the attempt by Hochtief to gain control was premium to its purchase price and then
effectively blocked by the Cartel Office. In proceeded to offer a coercive dilution deal
the remaining case, involving Continental, to the remaining (non-blocking) minorities.
court decisions on voting restrictions The Feldmühle Nobel case similarly involved
were the decisive factor in thwarting the a substantially lower bid to minorities once
predator—though Continental’s supervi- Stora had gained control.
sory board nevertheless dismissed the CEO This evidence contrasts with the US,
for his alleged intransigence in dealing with where most of the blockholders investigated
Pirelli’s bid. by Barclay and Holderness (1992) volun-
Overall, therefore, we observe control tarily buy out the minorities at a premium
changes in 12 cases, co-operation with the to the block price. The ability to dilute the
stakebuilder in two cases, and no change value of minority shareholdings in Germany
in control in the remaining three cases. does, of course, provide a powerful incentive
Table 26.9 also documents management for takeover and is likely to reduce Grossman
692 ON TAKEOVER AS DISCIPLINARY MECHANISM
derive, in the main, from the ability to block corporate control we observe? A full answer
transactions such as profit transfer and is beyond the scope of the present article,
control agreements that can be used to but there are certainly a number of areas
reduce the value of minority shareholdings. of concern. First, the ability of a controlling
Establishing a blocking minority stake is, blockholder to expropriate minorities could
therefore, frequently the initial stage in a significantly raise the cost of capital. It is
control contest, and is often followed by an interesting to observe that the recently
attempt to build a coalition with other introduced Takeover Code in Germany—the
large stakeholders with a view to gaining adoption of which is voluntary26 for firms—
control. As we see from the cases, the makes some attempt to protect minorities.
emergence of a rival coalition can even Companies acquiring stakes in excess of
cause existing pooling agreements between 50% are now required (provided they have
stakeholders to collapse. adopted the Code) to make an offer for the
Third, we have illustrated the complex outstanding shares, but the regulations
role that banks can play in corporate gov- regarding the terms of such offers provide
ernance. The dominant role accorded to very little, if any, additional protection for
German banks in much of the academic lit- minorities and are far weaker than those
erature is as major providers of finance and operating in other countries.27
also—via their representation on supervi- Second, corporate governance in Germany
sory boards, their direct equity ownership is both unpredictable and lacking in trans-
and their control of proxy votes—as impor- parency. Battles often involve a protracted,
tant monitors of corporate performance. and clandestine, shuffling of stakes between
We believe that this stereotype is, at best, rival coalitions and the revising of pooling
only partially accurate. There is growing agreements. Even large blockholders can
evidence that banks do not provide a higher find themselves, apparently without warning,
proportion of finance for investment in as members of the suppressed minorities.
Germany than elsewhere, and that the Furthermore, once a hostile stakebuilder
effectiveness of the monitoring role has appears there is frequent recourse to the
often been overstated.25 In this paper, we courts, whose decisions are, on occasion,
identify another important role for banks, unpredictable and lack consistency. As a
namely their role in assisting companies result, some of the bids considered in this
pursuing a strategy of hostile stakebuild- paper took over 5 years to reach their con-
ing. We document many cases where banks clusion. If a guiding principle for the design
play a pivotal role in building, brokering of corporate governance systems is
and concealing stakes. In contrast, it is reasonable speed and certainty, the German
striking how few examples we find of banks system frequently fails to achieve either.
actively defending target companies from a Germany’s market for stakes is becom-
hostile stakebuilder. Such behaviour may, of ing more liquid. Banks are, in general,
course, be compatible with the view that reducing their stakes in companies, in
banks actively monitor German companies response to tax changes and a general shift
and help to effect changes in corporate in opinion regarding their role in corporate
governance in the case of failing firms. governance. The latter has many roots: the
However, it is important to recognise that recent string of embarrassing failures of
this role is performed not by the compa- control and monitoring at Schneider,
nies’ house banks (who are often believed Balsam, Metallgesellschaft and Klöckner–
to be acting as monitoring intermediaries Humboldt–Deutz; the increasingly global
drawing on their privileged information), financial outlook of many large German
but by the banks assisting the predator.This companies; and the new focus on share-
role has not previously been recognised. holder value permeating many German
An important question remains: given companies and banks. As banks reduce their
the ownership structure that exists in stakes, and their influence over corporate
Germany, how efficient is the system of governance, cases of hostile stakebuilding
694 ON TAKEOVER AS DISCIPLINARY MECHANISM
foundation close to members of its supervi- and the family came to an arrangement,
sory board, implying that a new owner electing first one of his associates and later
could not control Asko’s operations (this Kirch himself onto the supervisory board.
was part of a defence structure erected in Throughout the 10 years of control battles,
response to two hostile takeover attempts there was extremely high turnover amongst
by rival retailer co-op AG in 1978 and top executives, including four fired CEOs in
1981). Asko itself had a 5% voting restric- one 12-month period.
tion. The battle ended once the supervisory
board removed its chairman in a vote of
confidence, unravelled the special rights of A.4. Bibliographisches Institut and
the foundation, and removed the voting F.A. Brockhaus (BIFAB)
right restriction. Metro then increased its
stake to 55%, in the open-market and from The control contest began when Maxwell
other blockholders (believed to be its asso- Communications privately approached
ciates, though the sellers’ identity was BIFAB’s board with a takeover offer at a
never confirmed). 53% premium to the share price. At the
time, there were three blockholder groups:
the Brockhaus family with around 15%, its
A.3. Axel Springer Verlag pooling partner the Meyer family with
around 38%, and Rheinpfalz Verlag with
The control contest began when, following 27.34%. While it is not known what
the flotation of the company and the death prompted Maxwell’s bid, the fact that the
of its founder, three rival blockholders Meyer family did eventually sell out hints at
emerged at this publishing house: the family either of: (i) its desire to divest its stake for
(which, though in a minority, controlled the some unknown reason, which put BIFAB in
boards), the Burda brothers Franz, Frieder play and prompted Maxwell’s bid; (ii) the
and Hubert (to whom the late founder had imminent breakdown of the controlling
tried to give majority control, which was coalition of the two families; or (iii) the
blocked by the Cartel Office), and an out- possibility that the large premium Maxwell
sider, Leo Kirch. Initially, Franz and Frieder was willing to offer tempted the Meyer
Burda co-operated with the family to con- family into considering a sell out. BIFAB’s
tain Kirch’s influence, denying him a seat board responded to Maxwell’s approach by
on the supervisory board and not removing saying it “resisted a takeover by a foreign
the CEO, as he demanded. Then they sur- firm” and was reported to be looking for a
prisingly agreed to pool their 25.9% stake (German) white knight to stave off
with Kirch’s declared 10% and his unde- Maxwell’s “hostile takeover bid.” The
clared option on a further 16%, giving the Brockhaus family added to this opposition
new coalition majority control. Only a claiming it would not sell under any cir-
month later, however, Franz and Frieder cumstances, nor would it tolerate a sale to
Burda sold their block to the family. The Maxwell by the Meyer family. A white
family was not yet safe, though: claiming he knight was quickly found in rival publishing
had first refusal on his brothers’ stake, house Langenscheidt, which bought the
Hubert Burda challenged the sale in court. Meyer family’s 38%, giving it majority
As defensive measures, the Springer family control in coalition with the Brockhaus
refused to register Kirch’s additional 16% family (whose patriarch joined Langen-
of votes, signalled they would not register scheidt’s board). Langenscheidt shortly
Hubert’s block should he win in court afterwards increased its stake to 65.34%
(which he finally did not), and entered into by buying out Rheinpfalz. One year later,
a cross-shareholding agreement with Monti Langenscheidt and Brockhaus jointly took
of Italy (in a further twist, the family and BIFAB private; the buy-out offered minor-
Monti later fell out, with Monti threatening ity shareholders a 10% premium to the
to sell its 10% to Kirch). Eventually, Kirch trading price.
696 ON TAKEOVER AS DISCIPLINARY MECHANISM
family shareholders and HF (though that this option had been rejected to ensure
contract was due to expire within 11 months Buderus’ independence. However, shortly
and even the board had to admit that the afterwards, Commerzbank and Dresdner
pool did not agree on the desirability of revealed they had each taken a 10% stake
IWKA’s new stake; the 54% figure also “as a long-term financial investment and
suggests that 3.1% of the family holdings not for resale to potential takeover bid-
did not rally around Reuther’s defence). ders.” A month later, Bilfinger and Berger
Reuther himself was engaged in negotiating (B & B) revealed at Buderus’ AGM that it
the sale of the combined 54% block held had assembled a 15% block and planned
by the remaining family shareholders and to discuss its long-term strategic vertical
HF to Britain’s Siebe (it later emerged involvement with Buderus’ management.
that the family had offered to buy-out HF While Buderus did not openly condemn the
at 170% of book value, and that Siebe’s emergence of this block, its management
195% bid for the 54% majority stake val- did declare their intention to keep the com-
ued the company more highly than IWKA’s pany independent of outside influence
180%). However, within 2 weeks, IWKA (curiously motivated as being in the interest
had secured HF’s support (in spite of the of Buderus’ core customers!). Significantly,
latter’s pooling commitment) and intended Dresdner Bank acted as B & B’s house
to dismiss B & R’s board at the forthcom- bank, was its only declared blockholder
ing extraordinary annual meeting. The deal (25.1%), and chaired its supervisory
with Siebe fell through once HF switched board. This constellation led to speculation
its support to IWKA, and HF was later that Dresdner Bank had facilitated the
sued by family members for violation of the assembly of B & B’s 15% stake (possibly
pooling contract (the outcome of this suit in connection with the IPO) and might have
is unknown). Following the EGM, the taken its own 10% stake for the benefit—
remaining family shareholders gave up and if not on behalf of—B & B. Not surpris-
sold their 32% stake to IWKA at the lower ingly, therefore, B & B applied to the
price of 180% of book value. IWKA Federal Cartel Office to increase its
exchanged the management, began to Buderus stake to 25%, prompting Buderus
restructure the company, and a few years to declare that it could not see any synergy
later also bought out HF. gains and did not need B & B as a strategic
partner. Only once the Cartel Office ruled
against the stake increase did Buderus
A.7. Buderus soften its tone and entered into a dialogue
with its 15% shareholder and supervisory
The control contest began shortly after board member B & B. To date, B & B has
Metallgesellschaft floated its 79.9% stake not divested its Buderus stake.
in Buderus in a public offering lead-managed
by Deutsche Bank and co-managed by
Dresdner Bank and Commerzbank with a A.8. Continental (Conti)
mandate to spread the shares widely. At
the time of the book-building, Buderus The control contest followed a year of
announced that a number of institutional heavy trading (7.6 times total Conti share
investors had taken stakes of between 1% capital) and persistent stock market
and 3% and jointly (though not in coali- rumours—repeatedly denied by Conti and
tion) controlled a majority of votes. the subsequent bidder, Pirelli—that a
Buderus’ management welcomed the fact takeover bid was imminent. In the run-up to
that the company no longer had a majority Pirelli’s intentions being confirmed,
owner; it is known that Buderus’ manage- the Italian tyre manufacturer—faced
ment and Metallgesellschaft had consid- with worldwide over-capacity in the tyre
ered selling to a single investor instead of industry—privately approached Conti with
placing the shares in the market, but that a merger proposal, and reportedly received
698 ON TAKEOVER AS DISCIPLINARY MECHANISM
encouragement from two members of the launched a public tender offer. Events took
supervisory board: Ulrich Weiss, of Deutsche an unexpected turn when a group of minor-
Bank (Deutsche Bank was Conti’s house ity shareholders demanded an extraordi-
bank, chaired the supervisory board, and nary general meeting to repeal the voting
held 5% of its shares) and Friedrich right restriction (which required a simple
Schiefer, a management board member of majority) and force a decision on the merger
5% shareholder Allianz, the insurance proposal (which required the approval of
company. Having assembled a 5% stake in 75% of votes at the EGM). Deutsche Bank
open-market purchases, Pirelli eventually and Morgan Grenfell put together a defen-
announced its bid for control in the form of sive coalition of banks, proxy votes and car
a reverse takeover by Conti of Pirelli’s tyre manufacturers large enough to block the
division, adding that it had already secured merger proposal: Daimler Benz (in which
the support of an unnamed majority of Deutsche Bank held a 28.3% stake and
Conti’s shareholders; Conti’s share price whose supervisory board it chaired),
fell by 7%. Pirelli was at pains to stress Volkswagen (whose CEO was Urban’s
that its proposal was friendly. Behind the predecessor at Conti), and BMW; Conti’s
scenes, Conti CEO Horst Urban enlisted share price fell by 5.5%. Nevertheless, at
Morgan Grenfell (Deutsche’s investment the EGM, the voting restriction was over-
banking division) to advise on the takeover turned with a 65.97% majority (though
defence and lobbied his board to reject the this was later to be opposed in the courts
proposed merger. Urban also promised his by a minority shareholder as well as Conti
unions that there would be no redundancies itself); Conti’s share price rose 5.2%.
if Conti stayed independent—which is a Conti still vowed to defend its independ-
significant form of defence as employee ence. Interestingly, the arithmetic of the
representatives controlled half the supervi- EGM indicates that Pirelli did not in fact
sory board. Though allegedly originally in control a majority of votes. Shortly after-
favour of Pirelli’s approach, supervisory wards, Conti’s supervisory board relieved
board chairman Weiss helped defeat (with CEO Urban of his duties, reportedly for his
the votes of the employee representatives) continued opposition to talks. Uncondi-
the other board members’ suggestion that tional talks were resumed and continued
management be instructed to negotiate over the next 8 months, though the man-
with Pirelli. Urban then publicly declared agement board was still publicly divided on
the proposal “a hostile takeover attempt, the merits of a merger. Conti began to
despite all assurances to the contrary.” The restructure by closing overseas factories
share price continued to fall. Over the fol- and selling off non-core divisions. Just
lowing few months, stakes of up to 5% before a cross-shareholding deal between
each were disclosed by some of Pirelli’s Conti and Pirelli was to be announced, it
backers, including two of Pirelli’s own emerged that Pirelli had given its backers
shareholders (Italian merchant bank indemnity guarantees to reimburse any
Mediobanca and Sopaf), Fiat, Allianz’ losses on Conti shares. As Conti shares had
Italian subsidiary and Merrill Lynch, lost roughly 45% of their value since
Pirelli’s advisors. After an unusually public Pirelli and its partners bought their stakes,
war of words, Conti demanded a standstill Pirelli came under pressure from its banks
agreement as a pre-condition for talks, to find funds to cover its position and was
including a commitment that Pirelli abstain eventually forced to call off the deal and
from attempting to remove Conti’s 5% vot- restructure and refinance its balance sheet.
ing right restriction. Pirelli rejected this However, Pirelli also bought options on its
demand. Morgan Grenfell was actively allies’ combined 32.4% in Conti. At that
engaged in finding a white knight or at least point, Conti appealed to the courts alleging
a 25% blocking coalition, a solution publicly Pirelli had broken securities laws by acting
favoured by Urban. Pirelli had still neither in concert and not disclosing the contracts
named its alleged majority supporters nor with its allies.This challenge, later accepted
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 699
by the court, invalidated the EGM’s decision the overall stake would be reduced back to
to remove the voting right restriction. Over 25.1% in due course, and, it appeared
the next 4 months, new Pirelli allies bought later, its agreement that DBV could veto
stakes in Conti and Pirelli tried again: Commerzbank’s choice of buyer. When
at the next AGM, it moved to have the Zurich Insurance emerged as a possible
restriction lifted, a proposal deemed buyer, DBV’s management expressly
“hostile” by Conti’s new CEO who called declared they would block any attempt at a
Pirelli’s intentions “sinister.” Conti refused hostile takeover and that they would not
to let Pirelli and its allies vote all their allow any shareholder to build up a major-
shares at the AGM, on the grounds that ity or dominant stake —which would seem
they constituted a concert party and as a credible threat given (i) the restrictions
such were limited to 5%; Pirelli’s motion on votes, (ii) the free float of below 50%,
was defeated, though Pirelli managed to and (iii) Commerzbank’s earlier commit-
obstruct Conti’s proposed capital increase. ment to ensure DBV’s independence. The
When finally a superior court resurrected a talks with Zurich Insurance collapsed,
decision of a pre-bid AGM to raise the partly because DBV opposed Zurich
majority required to remove the voting Insurance’s insistence on majority control.
right restriction from 50% to 75%, Pirelli Four months later, in spite of its long-
gave up and let its 5% stake and the standing commitment not to hold more
options it held be placed by Deutsche Bank than a blocking minority stake, Com-
with German companies friendly to Conti. merzbank increased its stake to 50% 1
These placements were supported by a share via open-market purchases. Once it
financial guarantee from the state of Lower controlled DBV, the bank swiftly proceeded
Saxony where Conti is headquartered. The to sell majority control to Winterthur, a
control contest lasted 2.5 years in total. Swiss insurer. As DBV’s CEO was chosen to
head all of Winterthur’s German sub-
A.9. Deutsche Beamten–Versicherung sidiaries, it seems unlikely that DBV
objected to the sale; however, since
(DBV) Commerzbank had acquired majority con-
The control contest began soon after public trol opposition on DBV’s part was, one pre-
sector insurer DBV was privatised via a sumes, no longer an attractive proposition.
public offering of 50% minus 2 shares.The
IPO was lead managed by Commerzbank, A.10. Dyckerhoff und Widmann
which not only took a 25% 1 share (Dywidag)
stake (another 25% 1 share block still
being in the public sector) but also bound The control contest began at a time when
itself not to increase the size of its block Dywidag’s ownership structure was highly
and agreed that DBV was to remain inde- fragmented, with a maximum free float of
pendent. One way to ensure independence only 21% and the two largest blocks
was the choice of restricted transferability owned by Holzmann, a rival construction
of shares, which requires the board’s regis- company whose shareholding was viewed
tration before votes can be exercised. as friendly, and industrialist Max Aicher.
However, when DBV’s share price soon began During the late 1980s, Ignaz Walter,
to fall below the offer price, Commerzbank, acquisitive owner of several regional con-
perhaps in an effort to support the share struction companies, embarked on what he
price, acquired a further 23.3% over the later called a ‘strategy of slow takeover’ of
following few months from an unnamed Dywidag by clandestinely buying up minor-
investor. This stake increase ostensibly had ity blocks from various sources, including a
the approval of DBV’s board, and anyway stake held by his house bank, Bayern LB,
was accompanied by Commerzbank’s and 2% bought on the open-market. While
pledge not to have the votes of the addi- there was speculation that Walter was
tional shares registered, its assurance that behind the stake purchases, he repeatedly
700 ON TAKEOVER AS DISCIPLINARY MECHANISM
denied being a blockholder in Dywidag. By the sale price of various assets. When a
the end of 1991, Walter controlled 40% of hostile tender offer was rumoured, FeNo’s
Dywidag’s votes, which he had previously management restricted voting rights to a
failed to declare, and explicitly denied hav- maximum of 5% per shareholder, aided by
ing; it appears that technically, he never Deutsche Bank which controlled about
crossed the 24.9% disclosure rule simply 55%, mostly via proxies but also via its
by spreading the 40% stake over various 8% stake in FeNo. A year later, the Flicks
associated parties, including his children. and five associated parties sold a previ-
Once Walter declared his stake as well as ously undisclosed 40% block (assembled
his intention to take over the company, with the help of Merrill Lynch via open-
Dywidag’s management strongly and pub- market purchases and secret direct agree-
licly resisted his hostile endeavours. ments with institutional blockholders) to
Despite being the largest shareholder, Veba, which Veba then boosted to 51%.
Walter was denied a seat on the supervi- However, when Veba failed to remove the
sory board. In December 1991, a financier 5% restriction and take control of FeNo,
called Dieter Bock proposed a “friendly” the Flicks assembled another stake of
takeover of Dywidag, a move that was wel- between 10% and 20% and began to
comed by the board. Rather than inviting oppose Veba for not launching a full bid.
all Dywidag’s shareholders to tender their The Flicks’ actions are consistent with a
shares, Bock proposed to consummate the strategy of trying to maximise the bid value
takeover via the negotiated friendly acqui- in the ensuing auction, which they were
sition of two key shareholders’ large stakes: well-placed to do given that their 10–20%
the 24.9% stake held by Holzmann, and stake, spread over several parties, gave
the Aicher block, which by now had been them more clout than Veba derived from its
increased to 24.7%, along with the small 51% stake. Two further suitors emerged in
stake owned by Dumez of France via a the form of Sweden’s Stora Kopparberg
cross-shareholding arrangement. This 51% and SCA, both of which bid for the company.
stake would block the 40% stake owned by Veba eventually sold its block to Stora, who
Walter. However, events did not turn out as also bought out the block jointly held by the
expected. Bock duly acquired the Flicks and Merrill Lynch, as well as SCA’s
Holzmann stake, but then announced, in 5% toehold stake. Once Stora owned 85%,
May 1992, that Advanta had failed to com- minority shareholders were offered a buy-
plete the assembly of a controlling stake in out price 15% below the bid price paid to
Dywidag, blaming Max Aicher for reneging Veba. The firm was split up and restruc-
on the sale contract (a view disputed by tured, in spite of the management’s opposi-
Aicher). Bock also disclosed that Advanta tion. The control contest lasted 15 months.
had sold the 24.9% stake in Dywidag it
had acquired from Holzmann to Walter, A.12. Th. Goldschmidt
finally giving Dywidag’s hostile suitor
majority control! Dywidag’s CEO resigned. The control contest began when two con-
No buy-out offer was made to minority glomerates, VIAG and Veba, independently
shareholders.The Cartel Office subsequently and potentially in rivalry, bought stakes
fined Walter DM 500,000 for failure to from an existing corporate shareholder and
register changes in its ownership interest in part of the family. When management and
Dywidag. the remaining family shareholders affirmed
their desire to remain independent, the new
A.11. Feldmühle Nobel (FeNo) blockholders pooled their 45.8% stakes
and demanded “a change of direction,” but
The control contest began when the Flick were kept at bay by the insiders who
brothers, Friedrich Christian and Gert- refused to grant representation on the
Rudolf (the former owner’s grand-nephews) supervisory board.The contest was resumed
accused the management of not maximising when (i) Veba came under pressure from
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 701
its shareholders to sell its 27.95% stake Krupp made its public announcement.
(25.02% of votes) and (ii) the family’s Initially, it was only Hoesch’s unions which
coalition with insurer Allianz broke down, called Krupp’s bid a hostile takeover.
when Allianz chose to let its pooling con- Hoesch’s CEO, Kajo Neukirchen, subse-
tract (via which the family controlled the quently developed a more confrontational
firm) expire in 1996 and put its 10.38% tone in public, and eventually Hoesch’s
stake (9.29% of votes) on the market. supervisory board declared its opposition
VIAG beat the family to both stakes, and in to the clandestine nature of Krupp’s stake-
April 1997, controlled 50.34% of the building, though it was in principle willing
votes to the family’s 39.94%. No offer to to consider merger plans on the basis of a
the minority shareholders was made. The voluntary discussion amongst equals. At
control battle lasted 5 years in total. the same time, it took the unusual step of
suspending two members of the manage-
A.13. Hoesch ment board who were believed to favour
Krupp’s bid, widely seen as a hardening of
The control contest began when high trad- positions. For a brief moment, it looked as
ing volumes in Hoesch shares throughout if British Steel might step in as a white
1991 prompted speculation of a possible knight to rescue Hoesch. Krupp’s Cromme
(foreign) takeover bid. In October 1991, reacted by noting that he would take over
Krupp, a rival steel maker, revealed it had Hoesch whether or not Hoesch co-operated.
bought a 24.9% stake secretly accumu- Significantly, he also claimed Deutsche
lated on its behalf by Credit Suisse. Hoesch’s Bank had been notified of the impending
share price fell by 9.7% in response, while bid some 2 weeks in advance, and had
Krupp’s rose by 8.6%. Krupp made clear welcomed it, which contradicted Deutsche
its intention to acquire a majority block, a Bank’s public insistence on its uninformed
plan for which it claimed to have received and neutral role. When Hoesch demanded
the prior support of various banks and evidence of Krupp’s alleged majority coali-
financial institutions with holdings in Hoesch. tion, Cromme provided notary evidence of
One of these was WestLB, Krupp’s house the support of a further 30.4% of the
bank and chair of its supervisory board, votes, including a 20% block held at Credit
which declared it had a 12% Hoesch stake Suisse, but excluding WestLB’s 12%.
on its trading books, though it denied to Shortly afterwards, Krupp announced it
have pledged the shares or the votes to had bought a further 26% (likely to have
Krupp. This is not altogether surprising, included at least part of Credit Suisse’s
since a formal agreement would have 20% stake) at an undisclosed price,
reduced Krupp’s influence given Hoesch’s thus increasing its 24.9% stake to a
corporate charter, which capped the votes majority block; Hoesch’s share price fell by
of any stake or formal pooling of votes at 4%. Krupp had still not made a formal
15%. Deutsche Bank was Hoesch’s house tender offer. Over the next few months,
bank and chaired its supervisory board. Krup and its allies removed the voting
Nevertheless, there were persistent right restriction, sealed a merger agree-
rumours that Deutsche Bank controlled a ment and saw off legal challenges from
block of perhaps 10% which was friendly three minority shareholders. Amongst its
to Krupp! Krupp’s CEO, Gerhard Cromme, allies were not only WestLB, its own
was at pains to stress this was no hostile house bank, but also Deutsche Bank,
bid, but a defensive move as Krupp would Hoesch’s house bank, which controlled an
have suffered had Hoesch been taken over estimated 12% of Hoesch via proxies.
by a (foreign) rival. The initial reaction Hoesch’s CEO Neukirchen resigned.
from Hoesch’s management was muted Krupp’s CEO Cromme was elected
with no particular indication of opposition, “Manager of the Year” by TopBusiness and
except perhaps grumblings about not hav- Manager Magazin. The contest only lasted
ing been informed until a few days before a few months.
702 ON TAKEOVER AS DISCIPLINARY MECHANISM
defend its client from takeover is not generally 21 This information came to light when it
accurate. In an interview with the Frankfurter emerged that Pirelli had entered into contracts
Allgemeine Zeitung, Hilmar Kopper, then chief with some of its allies guaranteeing to compen-
executive of Deutsche Bank, said he had known sate them for any fall in Continental’s share
of Krupp’s planned takeover of rival steel group price. Fall it duly did (by around 45%) and
Hoesch beforehand. “It has my full support Pirelli’s exposure became known when its banks
because it makes good industrial sense,” Kopper exerted pressure to cover its position.
said. In his view, Deutsche Bank (whose man- 22 Limiting the transferability of shares can
agement board member, Herbert Zapp, headed also be a major impediment to a predator, as in
Hoesch’s supervisory board) had no obligation the case of AGF’s stake in AMB. In this case,
to defend Hoesch against the bid: “How [should the court upheld the refusal of AMB’s board to
Deutsche Bank have defended Hoesch]? register the shares on the grounds that a board
Everyone knew someone was buying shares, but has a right to defend itself against a hostile
no one knew who. Secondly, why should stakebuilder.
Deutsche Bank defend Hoesch? Does Hoesch 23 Counting Asko’s supervisory board chair-
have a right to a defence? Or is Deutsche Bank man as ‘management’, which he de facto
obliged to maintain [industrial] structures?” appeared to be.
17 As suggested, for example, by Shleifer and 24 The Takeover Code was introduced in
Vishny (1986). However, incentives for minori- October 1995, and includes a requirement that
ties to free-ride in Germany are likely to be minorities should receive an offer when major-
small relative to the fears of expropriation. ity control changes hands. However, its adoption
18 The new securities trading law has intro- is voluntary, and the rules provide little protec-
duced two lower reporting thresholds of 5% tion for minorities. See footnote 27 for more
and 10%. In addition, the Berlin Supreme Court details.
ruled in 1991 that if asked at their AGMs, com- 25 See, for example, Rajan and Zingales
panies have a duty to disclose shareholdings of (1995), Corbett and Jenkinson (1996) and
(i) 10% or more or (ii) a minimum market Edwards and Fischer (1994).
value of DM 100 million. The market value 26 By September 1996, 1 year after it came
threshold means that share stakes of less than into force, only around one-third of listed com-
1% in DAX-30 companies are declarable. panies in Germany had actually adopted the Code.
19 Hochtief later went a step further and The Takeover Commission reviewed 12 cases in
signed a purchase agreement for the stake with its first year, only one of which led to a public
Commerzbank, effectively raising its holding in censoring.
Holzmann to 35.15%. The agreement and 27 A party that acquires a stake in excess of
exchange of title and consideration were contin- 50% (via open-market purchases or a private
gent on Hochtief winning its appeal against the deal) must offer to buy out minority sharehold-
Cartel Office’s ruling that it should not be ers unless the acquiring company has merged
allowed to increase its stake above 20%. This with the target company or entered into a profit
ruling prevented Hochtief voting the additional transfer and/or management control agreement
stake pending regulatory approval, but the within the first 18 months after gaining control.
purchase agreement bound Commerzbank ‘not In the absence of a merger or control agree-
to act against Hochtief’s interests’, a clause ment, the controlling company is required to
which would almost certainly tie Com- make a public offer for the remaining shares
merzbank’s hands at Holzmann’s AGM. When within the next 3 months. If, during the initial 18
these transactions came to light, the Cartel month period after having gained control, the
Office started an investigation into the legality acquiring company has not bought additional
of Hochtief’s dealings with Commerzbank and shares, the price offered to minorities must not
another bank (BfG), threatening the three be less than the price paid on purchases during
companies’ managers with fines of up to DM 1 the 6 months before gaining control minus 25%
million each. (denote this price P1). If the acquiring company
20 A Frankfurt court recently ruled a charter has made additional purchases since gaining
amendment inadmissible, that would have control, the price offered to minorities must be
allowed the Dresdner Bank management board the maximum of P1 and the weighted average
to institute a contingent 10% voting right of the prices paid on such additional purchases.
restriction in the event of a hostile bidder To those who have grown accustomed to observing
emerging, subject only to the supervisory bid premia being offered to shareholders in the
board’s approval. event of a takeover, such arrangements will
706 ON TAKEOVER AS DISCIPLINARY MECHANISM
hardly appear too onerous for the bidder! This control. Journal of Financial and Quantitative
contrasts, for example, with the U.K. City Code Analysis 26, 391–407.
on Takeovers, which demands a full bid once a Comment, R., Schwert, G.W., 1997. Hostility in
shareholder obtains a stake in excess of 30%, at takeovers: In the eyes of the beholder?
a price no worse than the highest price that the mimeo, University of Rochester.
bidder paid during the previous 12 months.
Corbett, J., Jenkinson, T.J., 1996. The financing
of industry, 1970–1989: an international
REFERENCES comparison. Journal of the Japanese and
International Economies 10 (1), 71–96.
Allen, F., Gale, D., 1994. A welfare comparison Denis, D.J., Serrano, J.M., 1996. Active
of the German and US financial systems, LSE investors and management turnover following
Financial Markets Group Discussion Paper unsuccessful control contests. Journal of
no. 191. Financial Economics 40 (2), 239–266.
Barclay, M.J., Holderness, C.G., 1989. Private Edwards, J.S.S., Fischer, K., 1994. Banks,
benefits from control of public corporations. Finance and Investment in West Germany
Journal of Financial Economics 25 (2), Since 1970. Cambridge Univ. Press,
371–395. Cambridge.
Barclay, M., Holderness, C.G., 1991. Negotiated Franks, J., Mayer, C.P., 1990. Capital markets
block trades and corporate control. Journal and corporate control: a study of France,
of Finance 46 (3), 861–878. Germany and the U.K. Economic Policy 10,
Barclay, M., Holderness, C.G., 1992. The law 191–231.
and large-block trades. Journal of Law and Franks, J., Mayer, C.P., 2001. Ownership and
Economics 35, 265–294. control of German corporations. Review of
Baums, T., Fraune, C., 1994. Institutionelle Financial Studies, forthcoming.
Anleger and Publikumsgesellschaft. Die Grossman, S., Hart, O., 1980. Take-over bids,
Aktiengesellschaft 3, 106–124. the free-rider problem and the theory of the
Becht, M., Roell, A., 1999. Blockholdings in corporation. Bell Journal of Economics 11,
Europe: an international comparison. European 42–64.
Economic Review 43, 1049–1056. Grundfest, J.A., 1990. Subordination of
Berle, A., Means, G., 1932. The Modern American capital. Journal of Financial
Corporation and Private Property. Macmillan, Economics 27 (1), 89–114.
New York. Holderness, C.G., Sheehan, D.P., 1985. Raiders
Bethel, J.E., Liebeskind, J.P., Opler, T., 1998. or saviors? The evidence on six controversial
Block share purchases and corporate perform- investors. Journal of Financial Economics 14
ance. Journal of Finance 53 (2), 605–634. (4), 555–579.
Bolton, P., von Thadden, E.-L., 1998. Blocks, Jenkinson, T.J., Mayer, C.P., 1994. Hostile
liquidity and corporate control. Journal of Takeovers—Defence, Attack and Corporate
Finance 53 (1), 1–26. Governance. McGraw-Hill, London.
Bulow, J., Huang, M., Klemperer, P., 1999. Jensen, M., 1993.The modern industrial revolu-
Toeholds and takeovers. Journal of Political tion, exit, and the failure of internal control
Economy 107 (3), 427–454. systems. Journal of Finance 48 (3),
Burkart, M., 1996. Economics of takeover reg- 831–880.
ulation, mimeo. Stockholm School of Mikkelson, W., Ruback, R., 1985. An empirical
Economics. analysis of the interfirm equity investment
Butz, D.A., 1994. How do large minority share- process. Journal of Financial Economics 14,
holders wield control? Managerial and 523–553.
Decision Economics 15 (4), 291–298. Mørck, R., Shleifer, A., Vishny, R.W., 1989.
Carney, W.J., 1997. Large bank stockholders in Alternative mechanisms for corporate con-
Germany: saviours or substitutes? Journal of trol. American Economic Review 79 (4),
Applied Corporate Finance 9 (4), 74–81. 842–852.
Choi, D., 1991. Toehold acquisitions, share- Rajan, R.G., Zingales, L., 1995. What do we
holder wealth and the market for corporate know about capital structure? Some evidence
ROLE OF HOSTILE STAKES IN GERMAN CORPORATE GOVERNANCE 707
from international data. Journal of Finance Wenger, E., Hecker, R., 1995. Übernahme-und
50 (5), 1421–1460. Abfindungsregeln am deutschen Aktienmarkt—
Roe, M.J., 1994. Strong Managers, Weak Eine kritische Bestandsaufnahme im
Owners: The Political Roots of American internationalen Vergleich. Ifo Studien
Corporate Finance. Princeton Univ. Press, 41, 51–87.
Princeton. Wruck, K.H., 1989. Equity ownership concen-
Shleifer, A., Vishny, R., 1986. Large sharehold- tration and firm value: evidence from private
ers and corporate control. Journal of Political equity financings. Journal of Financial
Economy 94 (3), 461–488. Economics 23, 3–28.
Spencer, C., Akhigbe, A., Madura, J., 1998. Zwiebel, J., 1995. Block investment and
Impact of partial control on policies enacted partial benefits of corporate control.
by partial targets. Journal of Banking and Review of Economic Studies 62 (2),
Finance 22, 425–445. 161–185.
Chapter 27
Jens Köke
THE MARKET FOR CORPORATE
CONTROL IN A BANK-BASED
ECONOMY: A GOVERNANCE
DEVICE?
Source: Journal of Corporate Finance, 10(1) (2004): 53–80.
ABSTRACT
This study examines whether changes in ultimate firm ownership (control) play a disciplinary
role in a bank-based economy. We focus on Germany as the prototype of a bank-based system.
We find that poor performance makes a change in control more likely; this suggests a
disciplinary role. Tight shareholder control acts as a substitute for control changes, strong
creditor control as a complement. Following a change in control, management turnover
increases, but not as a consequence of poor performance, and performance does not improve
significantly.These findings are inconsistent with a disciplinary role of the market for corporate
control in bank-based Germany.
relevant are tight shareholder control, changes in block ownership and their
continuous creditor control, and control by relation to management turnover for a
the supervisory board. These nonmarket small sample of listed firms. This study is
mechanisms might act as a substitute for the first to provide systematic evidence on
disciplinary control changes. Therefore, we the causes and consequences of control
examine whether their presence makes changes, based on a large sample of listed
control changes less likely to occur. and nonlisted firms.Third, this study identi-
Another characteristic of bank-based fies changes in ultimate firm ownership
countries is the large number of nonlisted based on individual shareholder data.
corporations. We take advantage of this Therefore, it allows for changes within
characteristic. If a stock market listing shareholder categories and in this sense
provides for an extra monitoring tool (e.g. extends the work of Denis and Sarin
via regular analysts’ reports), we should (1999) who examine changes between
expect nonmarket governance devices to be shareholder categories. It improves on
stronger for nonlisted firms. Therefore, we Franks and Mayer (2001) whose work is
investigate whether the causes and conse- based on changes in direct ownership.
quences of control changes differ between The study proceeds as follows: Section 2
listed and nonlisted firms. reviews the literature on control changes
The empirical analysis is based on a and other governance devices to formulate
sample of almost 1000 listed and non- testable hypotheses. Section 3 describes
listed German corporations for the years the data used for this study and defines the
1987–1994. Our main findings are: concept of control that is applied to iden-
(1) poor performance makes a change in tify the ultimate owner. Section 4 describes
control more likely, being a necessary the frequency of control changes in
condition for control changes to be discipli- Germany as well as the type of buyers and
nary; (2) tight shareholder control (creditor sellers of control blocks. Section 5 analyzes
control) has a negative (positive) impact the causes and consequences of control
on the probability of control change, indi- changes and compares the results with
cating that shareholder control (creditor evidence from market-based countries.
control) acts as a substitute (complement) Section 6 concludes.
for control changes; (3) following a change
in control management turnover increases,
but irrespective whether performance was 2. HYPOTHESES
good or poor before the transaction; and
(4) performance does not improve signifi- As a starting point of our analysis, we
cantly after a change in control. We con- hypothesize that changes in control have a
clude from these results that the market for disciplining function in bank-based as well
corporate control in Germany does not as market-based economies. To make this
work as a governance device. working hypothesis testable, we now
This study contributes to the literature in develop more specific hypotheses. Basically
several ways. First, previous evidence on they say when firms are likely to experience
ownership changes almost exclusively a change in control and what the conse-
refers to market-based economies; closely quences of a control change are likely to
related studies are Bethel et al. (1998) be. We pay particular attention to key
and Denis and Sarin (1999). This study characteristics of the German system of
examines a typical bank-based economy. corporate governance.
Second, previous studies on Germany
provide valuable insights into the nature of Hypothesis 1. Poorly performing firms are
ownership changes; Jenkinson and more likely to experience a change in control.
Ljungqvist (2001) assemble case study It is widely recognized that firm perform-
evidence on hostile stake-building, and ance may suffer if management does not
Franks and Mayer (2001) examine act in the interest of shareholders. The
710 ON TAKEOVER AS DISCIPLINARY MECHANISM
market for corporate control moderates concentrated ownership also has costs:
this divergence of interest because First, shareholders owning large blocks are
managers who fail to create shareholder poorly diversified. This may render their
value can be disciplined by shareholders monitoring too risk-averse (Fama and
acquiring control. This discipline can take Jensen, 1983). Second, the interests of the
the form of a takeover, closer shareholder concentrated shareholder do not need to
monitoring, or dismissing management coincide with those of other investors. This
(Shleifer and Vishny, 1986; Jensen, 1988; may allow the concentrated owner to
Scharfstein, 1988). Hence, we expect redistribute wealth from others (Shleifer
control changes to affect particularly and Vishny, 1997). Third, disciplinary
poorly performing firms. action via takeovers becomes more difficult
because incumbent large shareholders
Hypothesis 2. The availability of nonmarket must be willing to sell their blocks.
governance devices makes firms less likely Regardless of whether concentrated
to experience a change in control. ownership is good or bad, we expect
changes in control to be less likely if
For market-based economies, the market ownership is concentrated. For Germany,
for corporate control is often described as this negative relation between shareholder
a critical governance mechanism (Jensen, concentration and the probability of a
1988; Shleifer and Vishny, 1997). For a change in control should be particularly
bank-based economy such as Germany, non- pronounced because share concentration is
market monitoring devices can be expected comparatively high (see, for example, Becht
to play a larger role because hostile control and Böhmer, in press).
transactions are rare (Franks and Mayer, The second characteristic concerns
1998) and because other constituencies lending relationships. Shleifer and Vishny
such as large shareholders or large credi- (1997) argue that large creditors are
tors typically have considerable power similar to large shareholders because they
(Mayer, 1988; OECD, 1995). Key charac- have large investments in the firm and
teristics of the German system of corporate therefore a strong incentive to monitor.
governance are: (a) highly concentrated Additionally, large creditors typically have
share ownership, allowing for tight share- a variety of control rights and therefore
holder control, (b) close relationships to sufficient power to monitor. Hence, agency
banks, allowing for tight creditor control problems should be smaller for firms under
and long-term lending, and (c) a two-tier strong bank influence, making a discipli-
board structure with representatives of nary change in control less profitable. For
banks and shareholders on the supervisory Germany, we might expect that bank
board, supporting tight control by creditors influence is particularly strong. One reason
and shareholders. We expect that a firm is that historically German banks have
will be less likely to experience a change in acted as so-called house banks, providing
control if these alternative governance long-term loans to long-term clients
mechanisms are present. (Baums, 1994). Another reason is that
Concentrated share ownership is an German banks traditionally have owned
important vehicle to overcome the tradi- large equity positions in other German
tional free-rider problem (Shleifer and corporations and therefore have been
Vishny, 1986). A large shareholder has the represented on many supervisory boards.
incentive to collect information and to Mayer (1988) characterizes these lending
monitor management because the expected relationships as an integral part of the
private benefits from monitoring exceed the German system of corporate finance.
expected costs. Hence, agency problems The third characteristic concerns the
should be smaller for firms with concen- two-tier board structure of German
trated ownership, making a disciplinary corporations (see Hopt, 1997 for a detailed
change in control less profitable. But description). The management board
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 711
of complexity are well documented for 2 years after change in ownership. In Section
Germany (La Porta et al., 1999). Finally, 3.2, we investigate whether dropping firms
large firms may be less likely to experience with fewer continuous years of data intro-
a change in control due to wealth duces a selection bias, particularly which
constraints or diversification strategies of variables are most likely to be affected.
investors. Shleifer and Vishny (1992) show The sample is fairly representative for
that the market for corporate control is the universe of large German corporations.
less liquid as firm size increases. Taking the number of all incorporated
German firms in the year 1992 as a bench-
mark, coverage is high for listed firms
3. DATA AND MEASUREMENT (66.6%), all of which are public corpora-
ISSUES tions.1 For nonlisted firms, coverage is
small for public corporations (13.9%) and
weak for private corporations (0.03%).
We examine the frequency, causes, and
Controlling for firm size, the sample
consequences of control changes using a
includes 48% of all public corporations
sample of listed and nonlisted German
with total sales exceeding 100 million DM,
firms for the years 1987–1994. Firms from
and more than 3% of large private
former Eastern Germany are included after
corporations. Industry coverage is suffi-
1990. We use only incorporated firms
ciently representative for manufacturing
because disclosure requirements are the
industries. There is also a significant
strictest for them.The inclusion of nonlisted
number of firms from other industries such
firms is essential because the determinants
as wholesale trade and construction.
of control changes might differ between
Details on sample selection and data
listed and nonlisted firms, particularly for a
sources are provided in Appendix A.
bank-based economy such as Germany. To
This study does not distinguish between
identify changes in control, we apply a
private and public corporations because
concept of control that is based on ultimate
they are similar in many regulatory charac-
share ownership (see Section 3.3). This is
teristics (e.g., liability status, disclosure
significant because German corporations
requirements, and taxation).2 But it does
are often owned through pyramids or cross-
distinguish between listed and nonlisted
ownership structures (La Porta et al.,
corporations because their governance
1999), making changes in direct share
structures are likely to differ.3 As argued
ownership less meaningful.
above, we expect tighter shareholder
control and creditor control for nonlisted
firms, reflected in differences of ownership
3.1. Sample
and capital structure. For the present data
In total, the analysis of the frequency of set, we find that shareholder concentration
control changes (Section 4) is based on is significantly higher for nonlisted firms
data from 946 firms (4882 firm years). (Table 27.1). But we do not find that
This sample (in the following, Sample I) creditor control is tighter, at least when
includes firms with at least two continuous measured by leverage or the ratio of bank
years of data because changes in ownership debt to total debt. However, we find major
must be calculated. The analysis of causes differences between listed and nonlisted
and consequences of control changes firms regarding ownership complexity and
(Section 5) is based on a subsample and types of owners. At this point, we take these
encompasses data from 664 firms (4433 observed differences as indicative for
firm years). This sample (in the following, differences in governance structures. The
Sample II) includes firms with at least four observed differences are generally robust
continuous years of data because analyzing to excluding nonlisted subsidiaries, which
the consequences of control changes were increasingly sampled (by the data
requires data on the year prior and the provider) during the early 1990s. Whether
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 713
Ownership structure
Largest block 60.6% 85.9%*** 70.0%***
Ownership concentration 47.3% 80.3%*** 58.1%***
Cross ownership 6.6% 9.5%*** 6.8%
Pyramid 35.6% 57.0%*** 50.9%***
Type of owner
Individual 33.5% 28.1%*** 31.5%
thereof: insider 22.5% 12.0%*** 12.9%***
Financial firm 9.2% 9.2% 7.1%***
Nonfinancial firm 24.0% 48.6%*** 38.2%***
Government 1.9% 4.7%*** 3.4%***
Dispersed shares 31.5% 9.3%*** 19.8%***
Capital structure
Leverage 41.3% 39.0%*** 39.0%***
Bank debt 35.8% 20.7%*** 27.4%***
Number of observations 2460 2422 1141
Comparison of listed and nonlisted firms, based on the mean of various firm-specific characteristics. All
characteristics are defined in Appendix A.The types of owners are: individual (family, partnership, or foundation),
financial firm (bank, insurance company, or investment fund), nonfinancial firm, government authorities, and
dispersed shares (shares not held in a block). Insiders are executive and nonexecutive directors (or members of
their families). The test statistics are heteroscedastic t-tests of equal means; the tests compare listed firms
(column 2) with nonlisted firms (columns 3 and 4, respectively).
The sample comprises 946 firms (Sample I).
***
Significant at the 0.01 level.
this sampling procedure induces some type we find some differences that are
of sampling bias is examined next. statistically as well as economically
significant: For nonlisted firms, firms in
Sample II have lower leverage (measured by
3.2. CHARACTERISTICS OF ENTERING debt to assets), better performance (measured
AND EXITING FIRMS by industry-adjusted return on assets, return
on equity, or earnings loss), and larger firm
Next we check for a sampling bias. For size (measured by total assets). For listed
example, if poor performance increased the firms, we find no systematic difference
likelihood of firm failure or takeover, the between the two samples except for firm
sample could contain systematically fewer size, which is larger in Sample II.4
poorly performing firms. This would bias Second, we examine what type of firms
any observed correlation between perform- enter and exit the full sample (Sample I)
ance and the likelihood of a control change during the years 1987–1994 (Table 27.2).
towards zero. For firms entering in year t, the comparison
First, we examine whether Sample II, group is firms in year t that are in the
which is used for the analysis of causes and sample at least since year t 1. For firms
consequences of control changes (Section exiting at the end of year t, the comparison
5), systematically excludes some type of group is firms in year t that are still in the
firms. Comparing firms with at least four sample in year t 1. We find that entering
continuous years of data (Sample II) with firms tend to be smaller and have higher
firms that have fewer than four observations ownership concentration. Listed firms
714 ON TAKEOVER AS DISCIPLINARY MECHANISM
Firms in sample
Firms entering at least since Firms exiting Firms still in
in year t year t 1 at end of year t sample in year t 1
Listed firms
Largest block 67.3% 60.8%*** 84.0% 60.3%*
Ownership concentration 54.5% 47.5%*** 73.8% 47.2%
Debt-to-assets ratio 41.0% 41.6% 41.1% 41.2%
Return on assets (ROA) 4.6% 1.2%*** 0.1% 1.5%
Return on equity (ROE) 6.9% 1.3%*** 0.2% 1.8%
Change in ROA 0.4% 0.3% 4.6% 0.6%
Change in ROE 0.4% 0.9% 3.3% 1.6%
Earnings loss 3.5% 10.1%*** 20.0% 8.9%
Total assets (mn DM) 527.5 2714.2*** 27.0 2518.4***
Number of observations 171 2094 5 2094
Nonlisted firms
Largest block 89.1% 85.7%*** 82.1% 84.8%
Ownership concentration 85.1% 79.8%*** 75.5% 78.6%
Debt-to-assets ratio 43.8% 37.9%*** 47.2% 37.8%**
Return on assets (ROA) 0.1% 0.8% 0.4% 1.0%
Return on equity (ROE) 1.7% 1.3% 3.9% 1.6%
Change in ROA 0.9% 0.2%* 2.3% 0.4%
Change in ROE 1.7% 0.2% 1.3% 0.6%
Earnings loss 20.0% 13.0%*** 19.3% 12.3%
Total assets (mn DM) 692.4 1453.6*** 461.8 1379.3***
Number of observations 448 1842 60 1842
Characteristics of firms entering the sample after 1987, the first year of the sample, or exiting the sample before
1994, the last year of the sample, compared with characteristics of firms that do not enter or exit the sample,
separately for listed and nonlisted firms. For firms entering in year t, the comparison group is firms in year t
that are in the sample at least since year t 1. For firms exiting the sample at the end of year t, the comparison
group is firms in year t that are still in the sample in year t 1. Firms are compared at the mean of each firm
characteristic. All characteristics are defined in Appendix A. All performance measures except the loss indicator
are calculated as the difference to median industry performance (two-digit industry level).The test statistics are
heteroscedastic t-tests of equal means.
The sample comprises 946 firms (Sample I).
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
typically show better performance when reduce the explanatory power of poor
entering, whereas nonlisted firms show performance in a regression explaining the
poorer performance. Regarding exiting probability of control change. Second, a
firms, the main difference to firms remain- systematic bias due to sample attrition is
ing in the sample is firm size, which tends unlikely because, with the exception of
to be smaller for listed and nonlisted firms firm size, exiting firms are not signifi-
leaving the sample. cantly different from other sample firms.
There are two lessons from this sample This result could be expected because
composition analysis. First, it is possible 60% of sample exits are cases of nonre-
that Sample II is somewhat biased towards porting due to name changes, and not due
well-performing firms, but only for non- to bankruptcy or ownership change (see
listed firms. As discussed above, this can Appendix A).
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 715
100.0%
Dornier GmbH
57.6% 42.4%
Deutsche Aerospace AG Dornier family
17.0% 83.0%
Dispersed shares Daimler-Benz AG
distribution of share blocks for German the precondition for bank voting power
corporations shows large peaks at exactly based on proxy rights. Hence, if there is a
these thresholds, indicating that sharehold- discrepancy between German banks’
ers hold these blocks in awareness of the ownership rights and voting rights, it is
associated control rights (Köke, 2001; Becht probably small. In summary, the concept of
and Böhmer, in press). Therefore, varying control has potential weaknesses. But these
these thresholds should not significantly weaknesses do not appear to induce a
change the main findings of this study. systematic bias in the identification of the
Another weakness could be that the ultimate owner for each sample firm.
concept of control is based on officially
reported ownership rights instead of voting
rights.7 Ownership rights do not need to
coincide perfectly with voting rights for two 4. FREQUENCY OF CONTROL
reasons: First, shares might have voting CHANGES
caps or carry multiple voting rights.
Second, shares that are officially reported In this section, we determine the frequency
as dispersed might be under de facto of changes in ownership, the size of traded
control of banks because the German blocks, and the types of buyers and sellers
system of proxy voting gives banks the right of control blocks. We distinguish between
to execute voting rights of deposited shares. changes in the ultimate owner of each
Unless shareholders have given instruc- direct share block (block trades) and
tions, banks can multiply their influence changes in the ultimate owner of each sam-
this way. We believe that these weaknesses ple firm (control changes). For illustration,
do not induce a systematic bias. Although consider the following example (Figure 27.2).
legal during 1987–1994, limitations on In the year 1990, Boge, a large nonlisted
voting rights do not seem to be widespread. firm in the rubber industry, is owned by four
Gorton and Schmid (2000) report that large shareholders. Applying the concept of
only 14 of the largest German firms had control, we identify Carlo de Benedetti as
voting rights restrictions. Hence, for the the ultimate owner of Boge. In the year
vast majority of sample firms, voting rights 1991, Mannesmann becomes the new
do coincide with ownership rights. As for ultimate owner of VDO Adolf Schindling, a
proxy voting, recent evidence suggests that direct shareholder of Boge. This is a block
proxy voting is extremely unlikely to trade from the Schindling-Rheinberger
enhance German banks’ voting power family to Mannesmann. Note that de
considerably (Edwards and Nibler, 2000). Benedetti remains the ultimate owner of
In addition, only few of our sample firms Boge. In the year 1992, Mannesmann
have a dispersed shareholder base, which is acquires almost all shares of Boge, becoming
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 717
1990: Boge AG
100.0% 100.0%
Carlo de Benedetti Schindling-Rheinberger family
1991: Boge AG
1992: Boge AG
98% 2%
Mannesmann AG Dispersed Shares
its new ultimate owner. These block trades (Renneboog, 2000) and 9.0% for the UK
lead to a change in control. (Franks et al., 2001). For the US, Bethel et
In total, the sample encompasses 2460 al. (1998) report a rate of 6.6% for blocks
firm years for listed firms and 2422 firm larger than 5%. Hence, the frequency of
years for nonlisted firms. This is the total block trades in Germany is similar to other
number of cases in which block trading countries. For nonlisted firms, the respec-
could occur. For listed firms, in 258 cases tive annual rate is 7.0%; no comparable
(10.5% on average in any given year), we figures from other countries are available.
observe that a block is purchased by a new As exemplified by the case of Boge (Figure
shareholder, who did not own a block before 27.2), the trade of a large block does not
the purchase. For nonlisted firms, the need to be associated with a change in
respective rate of block purchases is 7.3%. control over the entire firm. An example of
Table 27.3 shows the size distribution of this is a new shareholder purchasing a large
blocks purchased by new shareholders. For block, while the majority of shares remains
listed firms, the purchased block is a in the hands of the ultimate owner. To iden-
majority block in 107 cases (41.5%), and tify changes in control, we check for name
a minority block in 62 cases (24.0%). This changes of the ultimate owner. We find that
strongly suggests that new shareholders a change in control takes place in 205 cases
come in because they want to take over in listed firms (8.3% on average in any
control of the respective firm. given year), and in 146 cases in nonlisted
To put these results into perspective, we firms (6.0%).8 Hence, the frequency of con-
compare them with evidence from other trol changes is somewhat lower than sug-
countries. For Germany, we find that a new gested by the number of blocks traded. We
shareholder purchases a block larger than conclude that not all block trades lead to
10% in 9.9% of listed firms, on average in change in control but that a large part does.
any given year. This rate of block pur- To see whether a particular shareholder
chases compares to 12.4% for Belgium type dominates control changes, we explore
718 ON TAKEOVER AS DISCIPLINARY MECHANISM
Table 27.3 Size distribution of blocks purchased by new shareholders
Size distribution of blocks that are purchased by a new shareholder (i.e. a shareholder that did not own a block
before the transaction) for the period 1987–1994, separately for listed and nonlisted firms. Block size is
observed at the direct level of ownership, block ownership is determined at the ultimate level, applying the
concept of control. Therefore, purchases by new shareholders refer to changes in ultimate ownership of direct
share blocks. Column 2 shows the total number of block purchases by new shareholders. Columns 3, 5, 7, and 9
disaggregate these purchases into four size classes. Columns 4, 6, 8, and 10 relate the disaggregate number of
blocks to the number of all block purchases by new shareholders (column 2).
The sample comprises 946 firms (Sample 1).
Characteristics of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6),
compared with characteristics of firms that do not experience such a change in any year (columns 3 and 5),
separately for listed and nonlisted firms. Firms are compared at the mean of each firm characteristic. For firms
with a change in control, statistics are calculated for the 2 years preceding the change (year – 1 and year – 2)
and the year of change (year 0). For other firms, statistics refer to the whole period of observation. All charac-
teristics are defined in Appendix A. The test statistics are heteroscedastic t-tests of equal means.
The sample comprises 664 firms (Sample II).
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
(change in control) with firms that do not change (8.0%). These purely descriptive
experience such a change in any of the results suggest that poor profitability gen-
sample years (no change). We find that erally does not induce a change in ultimate
profitability (measured by industry- ownership. Only in case of an earnings loss
adjusted return on assets) is weaker during a change in control becomes more likely.
the 2 years preceding a change in control We take these results as indicative that the
(year –1, year –2). But the difference to German market for corporate control
firms not experiencing a change in control (measured by changes in ultimate owner-
is statistically insignificant. The proportion ship) addresses governance problems
of loss-making listed firms is significantly (reflected in poor performance) and thus
higher among firms in which a control potentially works as a disciplining device.
change is going to occur within 1 year Next, we consider three alternative
(13.7%) than in firms without a control governance devices: (1) shareholder control,
720 ON TAKEOVER AS DISCIPLINARY MECHANISM
(2) creditor control, and (3) supervisory board (or their families), we find that firms
board control. Concerning shareholder con- not experiencing a change in control tend
trol, we find that ownership is significantly to show higher insider ownership; but this
less concentrated in firms experiencing a holds only for listed firms. This result sug-
change in control. This finding is consistent gests that insiders try to hinder changes in
with the view that concentrated ownership control. Finally, measuring ownership
ensures good governance, making a discipli- complexity by a dummy that indicates
nary change in control less profitable and whether the ultimate owner belongs to the
therefore less likely. It is also consistent well-known web of German financial and
with the view that concentrated ownership nonfinancial firms (Wenger and Kaserer,
makes it more difficult for outside 1998)9, we find no evidence that complex
investors to assemble a controlling block. ownership structures deter control purchases.
Concerning creditor control, we find that
the ratio of bank debt to total debt is sig- 5.1.2. Multivariate analysis
nificantly higher among firms experiencing
a change in control. Taking bank debt as a The above univariate analysis provides a
proxy for bank influence, this implies that first indication of the determinants of
control changes are more likely for firms control changes. To take into account that
under bank influence. This result is incon- different variables can simultaneously
sistent with the view that strong creditors affect the probability of a control change,
continuously monitor borrowers and we examine the determinants of control
thereby reduce the need for disciplinary changes using a logistic regression model
control changes. Rather, banks appear to (Table 27.5). This model predicts
assist control changes. control changes using 1-year-lagged meas-
The supervisory board is mandatory for ures of performance, alternative governance
listed firms. Thus, we cannot determine devices, and other deterrents of control
whether its presence is related to the changes. To investigate the difference
probability of a control change for listed between listed and nonlisted firms, we run
firms. For nonlisted firms, there is weak the regression on the pooled sample and
evidence that a larger proportion of firms include interaction terms for nonlisted
experiencing a change in control has a firms, i.e. a dummy that indicates a
supervisory board than firms not experi- nonlisted firm multiplied by the respective
encing a change in control. Hence, the firm characteristic. Hence, the columns for
presence of a supervisory board does not nonlisted firms in Table 27.5 show to which
appear to make control changes less likely. extent their slope parameters differ from
This finding is inconsistent with the view those for listed firms.
that monitoring by a board modifies the Model (1) confirms that firms making
need for disciplinary action. As argued in earnings losses are more likely to experi-
Section 2, some firm characteristics can be ence a change in control. Tight shareholder
expected to reduce the probability of a control, measured by ownership concentration,
change in control, irrespective whether these serves as an alternative governance device,
changes are disciplinary or not: insider reducing the probability of a control
ownership, ownership complexity, and firm change. Tight creditor control does not
size. Measuring firm size by total assets, we reduce the probability of a control change.
find that firm size tends to be larger for To the contrary, firms with strong creditors
firms that do not experience a change in are more likely to experience a change in
control. This supports the view that firm control. We find no evidence that the
size deters control purchases, for example supervisory board acts as a substitutive
due to wealth constraints of investors. governance device to changes in control. As
Furthermore, measuring insider ownership argued in Section 2, this lack of monitoring
by the fraction of shares owned by mem- power may be due to co-determination of
bers of the management or supervisory supervisory boards. Among the other
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 721
deterrents of control changes, only insider creditor control for addressing governance
ownership is significantly negatively related problems in large German firms. Our finding
to the probability of a control change. The that strong creditor control enhances the
determinants of control changes do not probability of a control change is consis-
differ between listed and nonlisted firms as tent with recent case-study evidence:
the column for nonlisted firms shows. The Jenkinson and Ljungqvist (2001) docu-
only exception is the dummy indicating ment many cases where German banks play
earnings losses, which is significantly smaller an assisting role in building hostile stakes.
for nonlisted firms. In fact, taking the sum Finally, we do not find evidence that the
of both earnings loss coefficients, we obtain determinants of control changes differ
the respective slope parameter for non- between listed and nonlisted firms.Thus, we
listed firms; this parameter is not statisti- do not find support for H3.
cally different from zero at the 0.10 level.
Hence, for nonlisted firms, poor performance 5.1.3. Sensitivity of results
does not affect the probability of a change
in control. As argued in Section 3.2, this To test the sensitivity of our results, we
result might be due to a selection problem conduct a number of robustness checks
because our sample contains systematically (Table 27.6). Our benchmark is Model (2)
fewer poorly performing nonlisted firms. from Table 27.5. First, we examine whether
Model (2) investigates the positive our results depend on the choice of the
impact of bank debt on the probability of a performance measure. Choosing a dummy
control change more closely. The aim is to that indicates low interest coverage instead
see whether this effect is due to the credi- of an earnings loss leaves all results quali-
tor position of German banks or to their tatively unaffected (Model 2a). When using
shareholder position. As argued in Section industry-adjusted return on assets, we find
2, German banks are often large creditors that the performance variable looses its
as well as large shareholders. To examine significance, but the other results remain
this issue, we interact bank debt with the qualitatively unaffected (Model 2b).10
total burden of debt (leverage) and with These findings confirm that poor perform-
the fraction of shares ultimately owned by ance makes a control change more likely,
banks (bank ownership). Table 27.5 shows but only if performance is very poor. Model
that bank debt looses its explanatory (2c) examines whether our results depend
power. Instead, the interaction of leverage on the choice of the measure for share-
and bank debt is significantly positive. This holder control. We find that measuring
indicates that the banks’ role in promoting shareholder concentration by the size of the
control changes comes from their strong largest share block instead of the
creditor position. In comparison, banks do Herfindahl concentration index does not
not appear to enhance their monitoring affect our findings qualitatively. This result
role based on their shareholder position. In could be expected because shareholder
summary, we find support for H1 that poor concentration in Germany is typically very
performance makes firms more likely to high (see also Table 27.1).
experience a change in control, but only if Next, we examine the impact of owner-
the firms’ financial difficulties are severe. ship complexity more closely. Replacing
We find partial support for H2 because the dummy that indicates cross ownership
tight shareholder control functions as a with a dummy that indicates pyramid
substitutive governance device to control structures, we find essentially the same
changes. But we find no support for the results (not tabled). In particular, the
view that creditors or the supervisory impact of ownership complexity remains
board act as alternative governance insignificant. These findings stand in con-
devices. These findings are consistent with trast to the popular view that complex
Edwards and Nibler (2000) who show that German ownership structures hinder
shareholder control is more important than control changes.
722 ON TAKEOVER AS DISCIPLINARY MECHANISM
Table 27.5 Causes of control changes: multivariate results
Earnings loss 0.863** (0.034) 1.300* (0.052) 0.791** (0.050) 1.261* (0.062)
Ownership 0.992*** (0.000) 0.353 (0.384) 0.981*** (0.000) 0.352 (0.388)
concentration
Bank debt 1.072*** (0.001) 0.191 (0.701) 0.160 (0.815) 1.040 (0.346)
Bank debt 2.114** (0.031) 1.310 (0.460)
leverage
Bank debt 0.378 (0.695) 0.176 (0.932)
bank
ownership
Supervisory 0.013 (0.952) 0.018 (0.934)
board
Insider 0.797*** (0.006) 0.845 (0.165) 0.821*** (0.005) 0.836 (0.172)
ownership
Cross ownership 0.053 (0.874) 0.342 (0.459) 0.014 (0.968) 0.327 (0.486)
Log (total 0.050 (0.130) 0.017 (0.293) 0.040 (0.234) 0.013 (0.435)
assets)
Number of 4181 4181
observations
Log 1030.9 1028.4
likelihood
Results of logistic regression models predicting changes in control by various firm characteristics. The models
estimate the slope parameters for listed firms (column listed firms) and the difference in slope parameters com-
pared to nonlisted firms (column nonlisted firms). All characteristics are defined in Appendix A. Both models
include an intercept and time dummies (not reported). The p-values (reported in parentheses) are based on
robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance–
covariance matrix.
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
Earnings loss 0.791** (0.050) 1.261* (0.062) 0.717* (0.079) 1.199* (0.076)
Low interest 0.952*** (0.006) 1.383** (0.021)
coverage
Industry 0.405 (0.443) 0.873 (0.212)
adjusted ROA
Ownership 0.981*** (0.000) 0.352 (0.388) 0.979*** (0.000) 0.408 (0.321) 0.927*** (0.001) 0.443 (0.276)
concentration
Largest block 0.923*** (0.001) 0.760* (0.091)
Bank debt 0.160 (0.815) 1.040 (0.346) 0.190 (0.779) 1.084 (0.326) 0.246 (0.717) 1.198 (0.273) 0.125 (0.854) 1.019 (0.356)
Bank debt 2.114** (0.031) 1.310 (0.460) 2.139** (0.029) 1.318 (0.458) 2.113** (0.032) 1.404 (0.425) 2.117** (0.030) 1.268 (0.475)
leverage
Bank debt 0.378 (0.695) 0.176 (0.932) 0.367 (0.703) 0.033 (0.987) 0.279 (0.771) 0.283 (0.891) 0.393 (0.683) 0.267 (0.897)
bank ownership
Supervisory 0.018 (0.934) 0.012 (0.957) 0.005 (0.980) 0.006 (0.977)
board
Insider 0.821*** (0.005)0.836 (0.172) 0.829*** (0.005) 0.843 (0.171) 0.826*** (0.005) 0.823 (0.179) 0.838*** (0.004) 0.770 (0.205)
ownership
Cross ownership 0.014 (0.968) 0.327 (0.486) 0.031 (0.927) 0.370 (0.432) 0.013 (0.969) 0.296 (0.527) 0.031 (0.928) 0.280 (0.551)
Log (total 0.040 (0.234) 0.013 (0.435) 0.033 (0.329) 0.015 (0.364) 0.044 (0.185) 0.011 (0.523) 0.050 (0.135) 0.027 (0.158)
assets)
Number of 4181 4145 4181 4181
observations
Log likelihood 1028.4 1015.3 1029.6 1029.1
Results of logistic regression models predicting changes in control by various firm characteristics.The models estimate the slope parameters for listed firms (column listed firms)
and the difference in slope parameters compared to nonlisted firms (column nonlisted firms). All characteristics are defined in Appendix A.The benchmark model for the robust-
ness checks is Model (2), which is taken from Table 27.5. All models include an intercept and time dummies (not reported). The p-values (reported in parentheses) are based
on robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance–covariance matrix.
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 723
724 ON TAKEOVER AS DISCIPLINARY MECHANISM
Table 27.7 Corporate restructuring following control changes
Characteristics of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6),
compared with characteristics of firms that do not experience such a change in any year (columns 3 and 5),
separately for listed and nonlisted firms. Firms are compared at the mean of each firm characteristic. For firms
with a change in control, statistics are calculated for the year preceding the change (year –1), the year of
change (year 0), and the 2 years following the change (year 1 and year 2). For other firms, statistics refer to
the whole period of observation. All characteristics are defined in Appendix A. The test statistics are
heteroscedastic t-tests of equal means.
The sample comprises 664 firms (Sample II).
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
observe that the replacement rates of the For the supervisory board, we also find
CEO and other executive directors strongly strongly increasing turnover rates, both for
increase following control changes. For the chairman and the other board members
example, for listed firms that experience a (not tabled).
change in control CEO turnover rates more Concerning asset divestitures, we find
than double from 10.2% in the year pre- that control changes are accompanied by
ceding the change to 21.1% in the year of increasing sales of fixed assets, but only
change. This turnover rate is significantly for listed firms. Specifically, sales rates of
above average CEO turnover in listed firms fixed assets increase from 6.9% in year – 1
that do not experience a change in control. to 8.8% in year 0. The difference in the
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 725
rate of asset sales to firms without a contrary, labor costs per employee
change in control is significant at the 0.05 significantly increase after a change in
level. Regarding employment, we find that ultimate ownership. This might reflect that
in the second year after the control the new owner is working to improve
change, the average growth rate of employ- incentives for employees.
ment is negative for listed and nonlisted Finally, the evidence does not support
firms, but the difference in growth rates the view that control changes improve
between firms with and without a change performance (Table 27.8). Neither for
in control is significant only for listed listed firms nor for nonlisted firms we
firms. Hence, employee layoffs increase, observe significant improvements in
but not before 2 years after the change in operating performance following control
control. This time lag suggests that adjust- changes. Only the fraction of loss-making
ment costs play a role in reorganizing a listed firms declines to a level that is
firm. Concerning cost cutting, we find no comparable to the industry benchmark. In
evidence of decreasing labor costs. On the summary, we find no support for H4.
Performance of firms that experience a change in ultimate ownership (control) in year 0 (columns 4 and 6),
compared with performance of firms that do not experience such a change in any year (columns 3 and 5),
separately for listed and nonlisted firms. Firms are compared at the mean of each performance measure. For
firms with a change in control, statistics are calculated for the year preceding the change (year 1), the year
of change (year 0), and the 2 years following the change (year 1 and year 2). For other firms, statistics refer to
the whole period of observation. All performance measures are defined in Appendix A; all except for the loss
indicator are calculated as the difference to median industry performance (two-digit industry level). The test
statistics are heteroscedastic t-tests of equal means.
The sample comprises 664 firms (Sample II).
*
Significant at the 0.10 level.
**
Significant at the 0.05 level.
726 ON TAKEOVER AS DISCIPLINARY MECHANISM
Results of logistic regression model predicting CEO turnover by two governance devices (change in control, tight
shareholder control) and their interaction with poor performance (earnings loss in previous period). The model
estimates the slope parameters for listed firms (column listed firms) and the difference in slope parameters
compared to nonlisted firms (column nonlisted firms). All characteristics are defined in Appendix A. The model
includes an intercept and time dummies (not reported). The p-values (reported in parentheses) are based on
robust standard errors, which are calculated using the White/Huber sandwich estimator for the variance–
covariance matrix.
**
Significant at the 0.05 level.
***
Significant at the 0.01 level.
6. SUMMARY AND CONCLUSIONS not make control changes less likely. This
suggests that the board cannot substitute
This paper sheds light on the market for for control changes; this stands in contrast
corporate control in a bank-based country. to evidence from the US (Kini et al.,
We focus on Germany because high owner- 1995). Codetermination may explain the
ship concentration and tight bank relation- weak monitoring power of German supervi-
ships make it the prototype of a bank-based sory boards. Contrary to our expectation,
system. We examine the frequency, causes, we find that control changes are more
and consequences of changes in corporate likely for firms under strong creditor influ-
control and compare our findings to those ence. This suggests that creditor control
from market-based countries. and control changes are complementary;
In contrast to the common perception, this assertion is supported by recent
Germany has an active market for large case-study evidence on hostile stake-building
share blocks. We find that the frequency of in Germany (Jenkinson and Ljungqvist,
block trades is similar to the US and the 2001).
UK. Since the traded blocks are typically Whereas the observed causes of control
very large, block trades usually lead to a changes are consistent with a disciplinary
change in control. The most active traders role of control changes, the observed
of control blocks are nonfinancial firms. consequences are not. We find that follow-
This stands in contrast to market-based ing a change in control management,
economies where nonfinancial firms play turnover increases, but irrespective
only a minor role in control transactions. whether performance was good or poor
Regarding the causes of control changes, before the control transaction. For listed
we find that poor performance makes a firms, also sales of fixed assets and layoffs
change in control more likely.This indicates increase. In contrast to evidence from the
that control changes in Germany may play US (Bethel et al., 1998; Denis and Sarin,
a disciplinary role as in market-based 1999), we cannot confirm that sales of
economies. We also find that high owner- financial assets increase or labor costs
ship concentration makes control changes decrease. Similarly, we cannot confirm that
less likely. This is consistent with the view performance significantly improves after a
that tight shareholder control acts as a change in control. In summary, although
substitute for disciplinary control changes. there is an active market for corporate
The presence of a supervisory board does control in Germany and although there are
728 ON TAKEOVER AS DISCIPLINARY MECHANISM
Consolidated balance sheet data for the years 1986–1994 5222 22,732
Mining, manufacturing, construction, and trade 3470 15,148
Ownership data 1485 6367
No missing values 1455 6205
Sample I (at least two continuous years of data) 946 4882
Sample II (at least four continuous years of data) 664 4433
Hypobank). These reports contain informa- nonlisted firms, mainly for private
tion on direct ownership of common stock corporations (GmbH). Third, some firms
(Stammaktien) for all listed German changed their names during the period of
corporations. In addition, Hypobank observation, for example following takeovers
provides information on direct ownership of or restructuring of conglomerates. Since
common stock for large nonlisted corpora- changes in ownership are crucial to the data
tions (Stammaktien for the nonlisted AG collection procedure, we adjust for name
and Gesellschafteranteile for the GmbH changes and obtain a panel on ownership
which cannot be listed). Hypobank reports structures through the years 1986–1994.
the size and the name of a direct owner In consequence, we are left with owner-
when the size of the ownership block ship data for 1485 firms (6367 firm
exceeds 5%. In general, ownership rights years). Because of missing values for
as reported by Hypobank correspond to important balance sheet items, another 30
voting rights (see Section 3.3). firms must be eliminated. This selection
Ownership information from Hypobank procedure generates a sample of 1455
cannot readily be used in our analysis for firms (6205 firm years) with at least
three reasons. First, ownership information 1 year of balance sheet and ownership
from Hypobank only refers to the direct level data during the years 1986–1994. As this
of owners. But this analysis requires to study examines changes in ownership, we
identify the ultimate owner of each sample further eliminate 509 firms for which we
firm. In contrast to Becht and Böhmer (in have only 1 year of data. For almost all of
press), who rely on voting rights information these 509 firms, the year 1994 is the
provided by the German Securities Exchange single year. This is due to the fact that
Commission (Bundesaufsichtsamt für den Hoppenstedt substantially increased firm
Wertpapierhandel), this study reconstructs coverage in 1994. The resulting sample
voting rights information in a bottom-up that contains 946 firms (4882 firm years)
approach from information on direct is labeled ‘Sample I’ in this study. For the
ownership rights. Our concept of control, as analysis of causes and consequences of
outlined in Section 3.3, represents the control changes (Section 5), we require
methodological tool to achieve this aim. A firms to display at least four continuous
second drawback is that Hypobank does not years of data. This generates ‘Sample II’
directly reveal ownership information on which contains 664 firms (4443 firm
medium-sized nonlisted firms. Therefore, we years). Table 27.10 summarizes the
construct the relevant ownership structures sample selection procedure.
by searching the information on investments To test for a potential sample selection
in subsidiaries and affiliated companies, bias, we collect data on the firms’ survival
which is given in the appendix to each status. For firms leaving the sample before
company in Hypobank. To further enlarge 1994, information is obtained from the
our sample, we search the Mannheim BSD and MUP databases and from tele-
Company Database (MUP) located at the phone interviews. We find that 39 out of 65
ZEW in Mannheim. We thereby obtain firms that exit the sample before 1994 still
ownership structures on many medium-sized existed in 1994 without a change in ultimate
730 ON TAKEOVER AS DISCIPLINARY MECHANISM
Table 27.11 Definition of variables
Performance
Return on assets (ROA) Earnings before interest, taxes, depreciation and amortization
(EBITDA) divided by total assets
Return on equity (ROE) EBITDA/total equity
Change in ROA ROAt ROAt – ROAt-1
Change in ROE ROEt ROEt – ROEt-1
Earnings loss EBITDA 0
Ownership structure
Largest block Size of largest share block
n
Ownership concentration Herfindahl index H j1P2j ,
with Pj proportion of a firms’ shares owned by shareholder j
Insider ownership Fraction of shares owned by executive and nonexecutive
directors or their families
Cross ownership Cross 1 if ultimate owner is part of the web of firms
identified by Wenger and Kaserer (1998) and if ultimate owner
indirectly owns a share block in itself, 0 otherwise
Pyramid Pyramid 1 if ultimate owner is located on the second or
higher level in the ownership structure
Capital structure
Bank debt Bank debt to total debt
Leverage Total debt to total assets
Interest coverage EBITDA/interest payments; interest coverage is low if
EBITDA/interest payments 1
Board structure
Turnover of CEO CEOt 1 if chairman of management board (Vorstand)
changes from year t – 1 to t, 0 otherwise.
Turnover of executive directors (Jt Lt)/(Et Et–1) with Jt (Lt) number of joining (leaving)
directors executive directors and Et number of executive
directors
Supervisory board Board 1 if firm has a supervisory board, 0 otherwise
Restructuring
Sales of assets sjt Sjt/Tjt with Sjt sales of asset class j in year t and
Tjt stock of asset class j at the beginning of year t
Growth rate of employment nt (Nt – Nt–1)/(Nt–1) with Nt total number of employees in
year t
Growth rate of labor costs ct (Ct – Ct–1)/(Ct–1) with Ct total labor expenditures in year t
ownership, but simply changed their name not generate a random sample of nonlisted
or quit reporting due to reasons determined medium-sized firms. Given that we include
within the firm. In 11 cases, operation was only medium-sized firms that are sub-
shut down due to liquidation or bankruptcy. sidiaries of conglomerates, our sample
In 14 cases, the respective firm had been might be biased. For example, we lack
taken over by another entity. And in one medium-sized firms that are directly owned
case, operation was shut down voluntarily. by families. But given the advantage of a
Hence, 60% of firm exits during the sample smaller average firm size in the total sam-
period are not related to bankruptcy or ple, we regard this as a necessary cost to be
ownership change. incurred.
We recognize that the selection proce- All variables used in this study are
dure of nonlisted medium-sized firms does defined in Table 27.11.
MARKET FOR CORPORATE CONTROL IN A BANK-BASED ECONOMY 731
Short, H., 1994. Ownership, control, financial model which should not be imitated.
structure, and the performance of firms. In: Black, S.W., Moersch, M. (Eds.),
Journal of Economic Surveys 8 (3), 203–247. Competition and Convergence in
Wenger, E., Kaserer, C., 1998. The German Financial Markets. Elsevier, Amsterdam,
system of corporate governance: a pp. 41–78.
Chapter 28
ABSTRACT
This paper examines the disciplining function of hostile takeovers in the U.K. in 1985 and
1986. We report evidence of high board turnover and significant levels of post-takeover
restructuring. Large gains are anticipated in hostile bids as reflected in high bid premiums.
However, there is little evidence of poor performance prior to bids, suggesting that the high
board turnover does not derive from past managerial failure. Hostile takeovers do not there-
fore perform a disciplining function. Instead, rejection of bids appears to derive from
opposition to post-takeover redeployment of assets and renegotiation over the terms of bids.
successful hostile takeovers and failed basis. In contrast, the post-bid asset sales
hostile takeovers. Failed takeovers provide of targets of unsuccessful bids are readily
evidence of target management’s response observable. Consequently, we also analyze
to the threat of a hostile bid and can be the asset sales of targets of unsuccessful bids
compared with the bidder’s response in a over a period of five years around the bid.
successful takeover.
2.2. Anticipated bid benefits and
2.1. Board turnover and prebid performance measures
post-takeover restructuring
If hostile bids serve a disciplinary function
If hostile takeovers perform a disciplinary there must also be evidence of anticipated
role, we should see a high level of manage- gains associated with the restructurings
rial turnover followed by large-scale and poor performance prior to the takeover
restructuring. These actions are necessary Anticipated merger benefits are measured
but not sufficient conditions for takeovers using bid premiums based on abnormal
to be disciplinary because they could reflect share price returns in the month of the first
disagreement over a strategic redeploy- bid. If bids are revised. we examine the
ment of assets. To evaluate management month of the first bid to the month of the
turnover, we collect data on all members of last bid.
the board of directors of the target firm, We use four measures of pre-bid finan-
including the name of the director, the date cial performance. The first is share price
of resignation or retirement, and the even- returns over five years prior to the month of
tual position of the director in the merged the announcement of the initial bid.The bid
company. announcement date is from Financial Times
Our information comes from company Mergers & Acquisitions (FTM&A). If bids
accounts of the target and bidder firms and are revised, the date of the first bid is
from company announcements issued by obtained from FTM&A, London Stock
the Stock Exchange. The number of target Exchange Microfiches, and Textline.Textline
company board members promoted to provides extracts from newspapers such as
the acquiring company’s board comes from the Financial Times. Returns are based
accounts of the target company and the upon share price and dividend data collected
acquiring firm. Telephone interviews with from LSPD and Datastream. Abnormal
company representatives corroborated and returns are calculated using a market
supplemented the data. The directors’ his- model ( 0, 1).The second measure
tories were followed for two years after the of financial performance is based on
takeover. changes in dividends per share and dividend
We use two measures of post-takeover omissions in each of the two years prior to
restructuring.The first measure is significant the financial year in which the takeover
sales of assets during the two years after occurred. The third measure of pre-bid
takeover. Asset sales for successful hostile financial performance is the cash flow rate
and friendly bids are classified as signifi- of return on assets employed in 1980 and
cant if more than 10% of the total fixed in the last accounting year before the bid,
assets of the merged firm are sold in the calculated from the cash flow gross of
corresponding year. Data on asset sales are interest and depreciation divided by the
based on annual accounts and telephone sum of the market value of equity and the
interviews with company secretaries to the book value of debt.The last measure is based
board of directors or equivalent persons. on Tobin’s Q ratio, the ratio of the market
Post-bid studies of successful takeovers value to replacement cost of a firm’s
suffer from the difficulty of separately assets. The ratio is calculated for 1980 (or
identifying asset sales in the target firms the first year of the listing) and the last
from those of the bidder, since the financial accounting year before the bid. The market
accounts are reported on a consolidated value of equity and book value of debt proxy
HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 737
for the numerator. The replacement cost of significant at the 1% level for both total
each company’s assets was taken from its and inside board resignations.
annual accounts. Until 1985, companies in The levels of inside director resignations
the U.K. were required to report the cur- in both successful and unsuccessful hostile
rent cost of their assets. Current cost is the bids are higher than in nonmerging firms.
replacement cost based on an up-to-date For example, in a sample of ten nonmerg-
cost of the assets, after providing for depre- ing firms drawn from the lowest decile of
ciation to reflect the age and usage of the share price performance in 1985, only
assets. Replacement costs are based on an 19% of the board resigned during the year
index of prices of new assets. of bad performance and the subsequent
year, compared with 90% in successful
hostile bids and 39% in unsuccessful bids.
3. TEST RESULTS Clearly, hostile takeovers are associated
with a high level of board turnover. These
Sections 3.1 and 3.2 examine the extent to results are consistent with Klein and
which hostile takeovers give rise to man- Rosenfeld (1988) who find that firms suc-
agement board turnover and post-takeover cessfully defending themselves through
restructuring. Sections 3.3 and 3.4 evalu- greenmail experience above-average man-
ate the extent to which hostility and mana- agement turnover. The high turnover of the
gerial control changes are associated executive members of the boards in unsuc-
with higher bid premiums and poor pre-bid cessful hostile bids is also consistent with
performance. Hirshleifer and Thakor’s (1994) prediction
that hostile bids reveal information about
the quality of management to the monitors
of the firm.
3.1. Board turnover
Table 28.1 reports that 50% of the direc- 3.2. Post-takeover asset sales
tors resign after accepted bids, compared
with 90% after successful hostile bids. If Hostile takeovers are also associated with
outside directors are excluded, the differ- restructurings. Based on accounts of merged
ence barely changes. According to panel 2, companies and telephone interviews, asset
the differences in proportions between sales exceed 10% of the value of total
successful hostile and accepted bids are fixed assets of the merged firms in 53% of
Table 28.1 Proportion of directors who resigned in the targets of accepted and successful and
unsuccessful hostile bids (size of sample, n. in parentheses)
Panel 1. Resignations
Accepted Successful hostile Unsuccessful hostile
Source: Annual reports. Stock Exchange microfiches. Part of the source data for 1986 is contained in appen-
dices A and B of Franks and Mayer (1990). It includes resignations in the two years subsequent to takeover.
a Significant at the 1% level.
738 ON TAKEOVER AS DISCIPLINARY MECHANISM
successful hostile bids. In contrast, only create greater shareholder, value than in
26% of accepted bids had significant asset other takeovers. Furthermore, if takeovers
sales, a difference that is statistically sig- are disciplinary, then takeovers which give
nificant at greater than the 5% level. The rise to managerial control changes should
high level of asset disposals is consistent have higher bid premiums than those which
with the results of Dann and DeAngelo do not. Table 28.3 finds that anticipated
(1988, p. 88), who find that ‘the median benefits are higher in successful hostile bids
restructuring involves assets constituting than in other bids: bid premiums average
23% of the target’s pre-offer equity value’. 30% in the month of a bid compared with
It is also consistent with Bhide (1989) who only 18% for accepted bids. [By way of
finds more evidence of sales of assets comparison, Franks and Harris (1989)
arising from past acquisitions of targets in report premiums of 42% for hostile bids
hostile than in friendly bids. and 28% for uncontested and unrevised
Table 28.2 provides a detailed examina- bids in the U.S.] The difference of 12
tion of the proportion of asset disposals for percentage points is statistically significant
unsuccessful hostile bids. The table sug- at the 5% level. When bid premiums are
gests that prior to the bid and in the year of computed for a period which includes any
the bid, annual sales of assets total 8–9% bid revisions, the difference is 15 percentage
of total fixed assets. Post-bid asset sales points and is significant at the 1% level.
are larger, with the proportion increasing There is therefore some support for the
from 13% in the year after the acquisition proposition that bid premiums are higher in
to 19% three years after takeover. In the hostile bids. Bradley, Desai, and Kim
third year after the acquisition, asset sales (1988) suggest that competition in the
are significantly greater than 10% (at the bidding market may produce benefits to the
5% significance level), with 60% of com- target shareholders at the expense of the
panies having sold more than 10% of their shareholders of the bidder.
assets. This finding is consistent with the Unsuccessful hostile bids have bid pre-
observation of Dann and DeAngelo (1988, miums that are considerably lower than
p. 88) that ‘hostile bids rarely prevail in those in successful hostile bids. In fact, they
cases in which the planned restructurings are much closer to the bid premiums in
were implemented by target managers’. accepted bids, suggesting that the market is
anticipating some restructuring after
3.3. Anticipated benefits and bid unsuccessful bids but not at as high a level
premiums as in successful hostile bids.
We also examine whether the high level
It might be expected that the higher levels of board turnover in hostile bids (reported
of restructurings in hostile takeovers would earlier) is associated with high bid premiums.
Table 28.2 Asset disposals for a five-year period around the bid for unsuccessful bids occurring in
1985 and 1986 (size of sample. t-statistic, and proportion 10% in parentheses)
B 1a B B1 B2 B3 Averageb
Asset 8.5% 9.2% 13.3% 17.2% 19.4% 11.0%
disposals (32, 1.0,0) (24, 0.4,25) (19,1.0,42) (17,1.6,53) (15,2.2.60) (33,0.7.42)
as a % of
tangible
fixed
assets
Source: Annual reports. t-statistic indicates whether the figure is different from 10%.
a B is the year of the bid.
b Firms for which data do not exist for some of the five years have been included for the calculation of the
average.
HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 739
Table 28.3 Bid premiums for three samples: targets of accepted bids, successful hostile bids, and
unsuccessful hostile bids (sample size, t-statistic, and proportion positive in parentheses)
Source: LSPD. Bid premiums in panel 1 are calculated using a market model with parameter values of 1
and 0.
a Significant at the 5% level.
b Significant at the 1% level.
c Significant at the 10% level.
Source: LSPD and own calculations. Abnormal returns in panel 1 are calculated using a market model with
parameter values of 0 and 1.
a Based upon the same industry using the Stock Exchange Industrial Classification and the company with the
closest market equity capitalisation.
b Some companies had incomplete data for the five years because they were not listed at the beginning of the
period.
c One company has been excluded because it was an outlier showing an abnormal return of 161% over five years.
assumed to be three months prior to described as one and two years prior to the
the announcement date as reported by the bid, the difference is around 14 percentage
London Stock Exchange. Results are points. Although economically meaningful,
reported on an equally weighted basis. The panel 2 records that this difference is not
t-statistics in the second panel test the significant at the 5% level. The difference
hypothesis that the average abnormal in returns is reduced to less than 5 percentage
returns in the two relevant samples are points when abnormal returns are esti-
equal. The t-statistics are calculated as mated over the five years prior to the bid.
t (M1 M2)/SD, where M stands for When matching the accepted and hostile
the cell mean SD = /(S12 /N1 + S22/N2, with bids with nonmerging firms of equivalent
N the number of observations averaged in a size and industry, nonmerging firms record
cell mean and S the cross-sectional deviation almost identical performance to successful
about the cell mean. hostile bids over the five years prior to a
A comparison of accepted bids with bid and somewhat worse (but not statisti-
successful hostile bids in panel 1 suggests cally significantly so) performance than
that the pre-bid performance of targets of accepted bids. Abnormal returns are higher
successful hostile bids is worse than that of for unsuccessful bids than for successful
accepted bids: in each of the periods bids, although the difference over five years
HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 741
is small and none of the differences are prior to a bid. The dividend behavior of
statistically significant. An analysis of medi- targets of unsuccessful hostile bids is very
ans discloses similar differences between similar to targets of successful bids.
accepted and successful hostile bids on an Differences in dividend policy may
equally weighted basis two years prior to a reflect changes in payout policy rather than
bid and a smaller difference one year differences in underlying performance.
before a bid. Using the Wilcoxon test, the However, the payout ratios of the three
differences in median are significant at samples calculated for the six-year period
the 13% level two years before a bid and 1980–85 (or 1981–86 for 1986 bids) are
at the 14% level one year before a bid. similar: 38.3% for accepted bids, 40.4%
for successful hostile bids, and 43.3% for
3.4.2. Dividend changes unsuccessful hostile bids. The differences in
the payout ratios between the three sam-
The second measure of financial perform- ples are not statistically significant and the
ance reports firms’ dividend decisions in t-statistics are all less than one.The results
each of the two years prior to a bid using suggest that there is little difference in per-
four classifications: omission, increase, formance between the samples.
decrease, or no change. Poor performance This result stands in marked contrast to
should be associated with reductions and the dividend behavior of a sample of 80
omissions in dividends. Panel 1 of Table 28.5 firms in the lowest quintile of shareprice
shows that in the two years prior to a bid a performance reported by LSPD in 1985.
majority of companies in all three samples The sample had an average abnormal
of bids (successful hostile, unsuccessful return of 63%. Dividends were omitted
hostile, and accepted bids) increased their or reduced in 41% of cases compared with
dividends; very few reduced them. The divi- 12% for successful hostile bids. In the
dend was omitted entirely in only six cases. remaining cases, dividends were held con-
A similar pattern emerges in the second stant (29%) or increased (30%) For 80
panel for dividends declared over the one firms with zero abnormal returns, dividends
Table 28.5 Proportion of companies omitting or changing dividends in accepted and hostile bids
(size of sample, n, in parentheses)
were omitted and reduced in 9% of cases successful hostile bids are not statistically
and increased in 77% of the cases. These different from those of unsuccessful bids.
results are remarkably similar to the divi-
dend performance of the sample of hostile
takeover targets reported in Table 28.5. 3.4.4. Tobin’s Q
This evidence suggests that when there is The failure to uncover differences in
poor performance as measured by share pre-bid rates of return between targets of
prices, dividends are more likely to be omit- hostile bids and other firms may be due to
ted or reduced.The results strongly suggest the stock market capitalizing poor per-
that the sample of targets of hostile bids formance at the start of the period. If this
with its very low level of dividend reduc- is the case, then there will be differences in
tions or omissions is not representative of the Tobin’s Q ratios, the ratio of market to
firms with poor share price performance. replacement cost valuations of firms.
Table 28.7 shows that the Q ratios of tar-
3.4.3. Pre-bid cash flow rates of return gets of successful and unsuccessful hostile
bids in 1980 are similar to those of
Table 28.6 reports cash flow rates of return accepted bids.Thus the difference in abnor-
on capital for 1980 and 1985/86. Rates of mal returns between hostile and accepted
return for targets for successful hostile bids bids revealed in Table 28.4 is not a conse-
(17.82%) are similar to those of accepted quence of differences in initial market val-
bids (20.25%) as well as to their matched ues. However, the nonmerging sample has a
sample of nonmerging companies (17.15%). significantly higher Tobin’s Q than the suc-
None of the differences is statistically cessful hostile bids. There is, therefore,
significant. Rates of return for targets for some evidence that poor performance is
Table 28.6 Cash flow to asset ratios in 1980 and 1985/1986 for various samples of merging and
nonmerging firms (size of sample, t-statistic, and proportion positive in parentheses)
Table 28.7 Tobin’s Q for 1980 (or first year of listing) and 1985/1986 (or last year of
listing/annual report) for various samples of merging and nonmerging firms (size of
sample, t-statistic, and proportion with Tobin’s Q 1 in parentheses)
Panel 1 Tobin’s Q
Matched Successful Matched Unsuccessful
Accepted nonmerging hostile nonmerging hostile
bids samplea bids samplea bids
Source: Datastream, LSPD, own calculations. t-statistic indicates whether Tobin’s Q is different from one.
Definition: Tobin’s Q (Market value of equity Book value of debt)/Replacement cost of assets.
a Based upon the same industry using the Stock Exchange Industrial Classification and the company with the
closest market equity capitalization.
b Significant at the 10% level.
c Significant at the 1% level.
d Significant at the 5% level.
being capitalized in the targets of hostile of targets of hostile bids is not easily dis-
bids when the comparison is made with tinguished from that of targets of accepted
firms in the same industry. bids. The performance of targets of hostile
For 1985 and 1986 the Tobin’s Q ratios bids is not inferior to that of a nonmerging
of targets of hostile bids are lower than sample of firms matched on the basis of
those of accepted bids in the year of the size and industry; the only exception is for
merger (1.04 versus 1.35); the t-statistic Q ratios. This evidence is consistent with
for the difference is 2.95. In addition, Martin and McConnell (1991) who find no
the Q ratio of successful hostile bids is difference in pre-bid performance of
lower than that for the targets of unsuc- targets of accepted and rejected bids.
cessful hostile bids (1.04 versus 1.22, These findings raise the question of
t-statistic 2.2). Thus, the performance whether poor pre-bid performance is more
of targets of hostile takeover bids is signif- closely associated with managerial control
icantly worse than that for the other two changes than with the hostility of bids. We
samples. However, the average Q ratios of examine this proposition by repeating the
the targets of accepted and hostile bids are above pre-performance tests on the sample
both greater than one, providing evidence of of all takeovers partitioned by control
poor relative, rather than poor absolute, changes. A very similar picture to that
performance. described for the hestile and accepted
partition emerges. (Tables relating to this
3.5. Summary of findings section are available from the authors on
request.)
To summarize, three different measures indi- Over five and two years prior to the bid,
cate that the pre-bid financial performance takeovers with control changes display
744 ON TAKEOVER AS DISCIPLINARY MECHANISM
superior abnormal share price performance Martin and McConnell’s (1991) finding of
to those without control changes, the oppo- worse pre-bid performance in takeovers
site of the result predicted by evidence of where there are high executive changes
managerial failure. Only in the year prior to stands in contrast to our results for the U.K.
the bid is there worse performance although
the difference is not statistically significant.
[Kini, Kracaw, and Mian (1993) record an
inverse relation between performance and 4. REVISED BIDS
CEO turnover for insider-dominated boards
and no relation for outsider-dominated The previous section reports a high level of
boards.] The proportion of firms cutting or board turnover in hostile take-overs, associ-
omitting dividends is very small in relation ated with high levels of sales of assets and
to the proportion raising dividends for both high bid premiums. However, there is no
samples. There is slightly more evidence evidence of high bid premiums when
that firms omit dividends prior to control takeovers are partitioned on the basis of
change bids but no evidence that they cut control changes. In addition, neither hostile
them. Cash flow rates of return are lower in take-overs nor control changes are associ-
the control change sample both five years ated with poor pre-bid performance. The
before the bid and in the year of the bid. evidence therefore strongly suggests that
However, the differences are not statisti- hostile takeovers are not associated with a
cally significant. Finally, five years prior to disciplinary role in this sample of U.K.
the bid the Tobin’s Q ratio of control takeovers.
change targets is slightly higher (0.79) The evidence of a high level of post-
than that of noncontrol change targets takeover restructuring in conjunction with
(0.70), although in the year of the bid it is high board turnover is more consistent with
lower (1.17 versus 1.32); the differences board dismissal being a result of disagree-
are not statistically significant. In both ment with the bidder over the future
cases, the Q ratios are greater than one and restructuring of the company. Another
therefore suggest profitable investment explanation, consistent with the high bid
opportunities rather than poor performance. premiums in hostile bids, is that rejection
In summary, there is little evidence that of bids is part of the process of negotiating
takeovers with control changes are the the terms of bids. To evaluate this possibil-
result of poor past performance. Several ity, we collected data on bid revisions for
studies examine the relation between board our sample of successful and unsuccessful
turnover and performance. Weisbach (1988) bids. In Table 28.8, we show that hostile
finds a strong association between prior bids were revised in 25 out of the 35 suc-
performance and the probability of director cessful hostile bids. When the hostile bidder
resignations for companies with outsider- did not revise the terms after rejection by
dominated boards. Warmer, Watts, and the target, there was a high rate of bid
Wruck (1988) only find an inverse relation failure (20 out of 30 cases).
between the probability of a management If rejection of bids were a matter of
change and a firm’s share price performance negotiation over price. we would expect to
when the latter is extremely good or bad. observe bids being rejected, revised, and
Gilson (1989) finds a significantly higher then accepted (if the bid succeeded); alter-
level of executive turnover in financially natively, the bid would fail because other
distressed than nondistressed firms. bidders would emerge in the same or in a
Whereas the last two studies find a subsequent bid. The 13 successful hostile
relation between very poorly performing bids that were finally accepted (and there-
companies and high board turnover. we fore recommended) by target management
show that the targets of hostile take-overs can be interpreted in this way. However, we
are performing neither poorly nor worse also observe 12 successful hostile bids
than the targets of accepted bids. However, being rejected to the end, even after the
HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 745
Revised Unrevised
Accepted 0 35
Successful hostile: finally accepted 13b 0
Successful hostile: contested to the end 12c 10
Unsuccessful hostile bidsa 15d 20
bids were revised. For this sample, rejection return are higher in ultimately accepted
may reflect the breakdown of the bargain- than in rejected bids in 1980 but lower in
ing process or outright hostility towards 1985/86.
the particular bidder. The latter is consis- In sum, there is no evidence of poor pre-
tent with a disciplinary role for takeovers bid performance in ultimately rejected bids.
or disagreement over the redeployment of Continuing opposition to the bid appears to
assets after the takeovers. reflect a breakdown over negotiations or
A further test of the disciplining function disagreement over the redeployment of
involves examining whether a sample of assets after the takeover.The fact that post-
hostile bids that were ultimately rejected takeover restructuring is significantly
had worse performance than a sample of higher in hostile than accepted takeovers
revised bids that were ultimately accepted. points to the latter as being an important
Table 28.9 reports three measures of per- consideration.
formance Tobin’s Q, abnormal share price
returns, and cash flow rates of return. If
initial rejection but ultimate acceptance of 5. CONCLUSION
bids has to do with bargaining over price,
then we would predict (i) higher Q ratios This paper examines whether hostile
for bids that are ultimately accepted (S.2 takeovers act as a discipline on manage-
in table 28.9), (ii) larger abnormal returns ment. Evidence comes from the perform-
and (iii) higher cash flow rates of return. ance of targets of hostile takeovers before,
The sample sizes are small, between 10 during, and after acquisition. We report
and 15 for each partition of hostile bids. that hostile takeovers are associated with
Table 28.9 reports that the pattern of high levels of boardroom changes, higher in
results is not consistent for any of the three hostile than accepted bids and higher in
measures, except that the Q ratios for con- targets of successful than unsuccessful
tested bids that are ultimately accepted are hostile bids. There are also larger asset
greater (0.87) than for the sample of bids disposals and bid premiums in hostile
that are ultimately rejected (0.73) in takeovers. However, targets of hostile
1980. In 1985/86, the Q ratios are almost takeovers do not perform worse prior to
identical. Abnormal returns are smaller for bids than either targets of other acquisi-
ultimately accepted bids over five years, tions or nonmerging firms. By only one
though the difference in averages is not measure (Tobin’s Q) is there any evidence
statistically significant. Cash flow rates of of statistically significant worse performance
Table 28.9 Tobin’s Q, abnormal returns, and cash flow rates of return for (i) revised successful hostile bids which are ultimately rejected by the incumbent
management (S.1), and (ii) revised successful hostile bids which are ultimately accepted by the incumbent management (S.2) (size of sample,
t-statistic, and proportion with Tobin’s Q 1 or abnormal returns and cash flow rates of return 0 in parentheses)
S.1: Ultimately
rejected bids 0.73% 1.02% 0.50% 5.31% 0.85% 18.72% 21.60%
(10, 3.1, 10) (11,0.2,45) (12, 0.28,47) (12, 0.58, 46) (12, 0.09, 31) (10, 7.07, 100) (11, 4.79, 100)
S.2: Ultimately
accepted bids 0.87% 1.08% 4.20% 3.00% 8.33% 27.12% 13.92%
(10, 1.0, 30) (11, 1.2, 64) (13, 1.40, 47) (13, 0.28, 47) (13, 0.87, 40) (10, 3.06, 100) (11, 7.82. 100)
746 ON TAKEOVER AS DISCIPLINARY MECHANISM
Panel 2 t-statistic
Tobin’s Q Abnormal returns Cash flow rates of return
1980 1985/86 5 years 2 years 1 year 1980 1985/86
S.1 less S.2 0.87 0.64 0.98 0.14 0.69 0.91 1.59
Source: Datastream, LSPD, own calculations. Abnormal returns are based on a market model with parameter values 0 and 1.
Definitions: Tobin’s Q (Market value of equiry Book value of debt). Replacement cost of assets. Cash flow Profit after tax Depreciation Interest payments. Book
value of debt ST loans LT loans Cash balances Deferred tax. Cash flow rates of return Cash flow (Market value of equity Book value of debt).
HOSTILE TAKEOVERS AND THE CORRECTION OF MANAGERIAL FAILURE 747
of targets of hostile bids. There is no evi- The definition of hostility is similar to that in
dence of either high bid premiums or poor this paper, where the target management initially
pre-bid performance when takeovers expressed opposition to the bid.
involve managerial control changes. The
market for corporate control does not REFERENCES
therefore function as a disciplinary device
for poorly performing companies. Bhagat, Sanjay, Andrei Shleifer, and Robert W.
Rejection of bids is more consistent with Vishny, 1990. Hostile takeovers in the 80s:
opposition to the anticipated redeployment The return to corporate specialization,
of assets by the bidder and negotiation over Brookings Papers on Economic Activity,
the terms of bids. We find that there is a 1–72.
high level of bid revisions and large bid pre- Bhide, Amar, 1989, The causes and conse-
miums in hostile takeovers out no evidence quences of hostile takeovers, Journal of
of poor performance where negotiations Applied Corporate Finance 2, 36–59.
over the terms of the bid break down.Thus, Bradley, Michael, Anand Desai, and E. Han Kim,
our results provide a picture of a market 1988, Synergistic gains from corporate
for corporate control in the U.K. that is acquisitions and their division between the
more consistent with dissatisfaction over stockholders of target and acquiring firms,
the bid terms and the redeployment of Journal of Financial Economics 21,
assets after the takeover than with a disci- 183–206.
plinary role. Dann, Larry and Harry DeAngelo, 1988,
Corporate financial policy and corporate con-
trol: A study of defensive adjustments in asset
NOTES and ownership structure, Journal of Financial
Economics 20, 87–127.
* Corresponding author. Franks, Julian R. and Robert Harris, 1989.
This paper is part of an ESRC funded project,
Shareholder wealth effects of corporate
number W102251003, on ‘Capital Markets,
Corporate Governance, and the Market for takeovers: The UK experience 1955–1985,
Corporate Control’. We are grateful to Nick Journal of Financial Economics 23,
Carrick, Luis Corrcia da Silva, and Marc 225–249.
Goergen for valuable research assistance. We Franks, Julian R. and Colin P. Mayer, 1990,
wish to thank John Chown for financial assis- Capital markets and corporate control: A
tance on this project and for helpful discussions. study of France, Germany and the UK,
The paper has been presented at the 1993 Economic Policy 10, 191–230.
Annual Meeting of the American Finance Franks, Julian P. and Colin P. Mayer,
Association in Anaheim, at the Annual Meeting 1994, Ownership and control in Germany,
of the French Finance Association in Paris, at
Working paper (London Business School,
the Annual Meeting of the European Finance
Association in Athens, at a conference on London).
European Restructuring at INSEAD and at Gilson, Stuart C., 1989, Management turnover
seminars in Essex, Keele, Oxford, Paris, and and financial distress. Journal of Financial
Warwick. We are particularly grateful to discus- Economics 25, 241–267.
sants Patrick Bolton, Stuart Gilson, Kristian Grossman, Sanford J. and Oliver D. Hart, 1980,
Rydgvist, and Paul Seabright for valuable sug- Takeover bids, the free rider problem, and the
gestions. We are also grateful for comments theory of the corporation, Bell Journal of
from Swami Bhaskaran, Michael Brennen, Economics 11, 42–64.
Michel Habib, David Hirsbleifer, Ernst Maug, Herzel, Leo and Richard W. Shepro. 1990.
Narayan Naik, Kjell Nyborg, and Walter Torous.
Bidders and targets: Mergers and acquisi-
We also wish to thank Richard Ruback (the
editor) and John Pound (the referee) for tions in the U.S. (Blackwell, Oxford).
many helpful comments in the revision of the Hirshleifer, David and Anjan Thakor, 1994,
manuscript. Managerial performance, boards of directors
1 Their sample includes hostile takeovers for and takeover bidding, Journal of Corporate
targets with a value greater than $50 million. Finance 1, 63–90.
748 ON TAKEOVER AS DISCIPLINARY MECHANISM
Jensen, Michael, 1988, Takeovers: Their causes Rappaport, Alfred, 1990, The staying power of
and consequences, Journal of Economic the public corporation. Harvard Business
Perspectives 2, 21–48. Review 1. 96–104.
Klein, April and James Rosenfeld, 1988, Shleifer, Andrei and Larry H. Summers,
Targeted share repurchases and top manage- 1988, Breach of trust in hostile takeovers, in:
ment changes. Journal of Financial A.J. Auerbach ed., Corporate takeovers:
Economics 20. 493–506. Causes and consequences (National Bureau
Kini, Omesh,William Kracaw, and Shehzad Mian, of Economic Research, Chicago, IL).
1993. Corporate takeovers, firm performance and Warner, Jerold B., Ross L. Watts, and Karen H.
board composition.Working paper (Pennsylvania Wruck. 1988, Stock prices and top manage-
State University. University Park, PA). ment changes, Journal of Financial
Martin. Kenneth J. and John J. McConnell, Economics 20, 461–492.
1991, Corporate performance, corporate Weisbach, Michael S. 1988, Outside directors
takeovers and management turnover. Journal and CEO turnover, Journal of Financial
of Finance 46. 671–687. Economies 20, 431–460.
Name index
Pfeffer, J. 470 Ruback, R. 43n, 45n, 56, 67, 221, 493, 539,
Pfleiderer, P. 138, 147, 149, 154n 572, 650, 663, 708, 711
Phillips, G. 726 Rydqvist, K. 60, 62, 79, 177, 182, 654, 659n
Pinnell, M. 526, 548–549 Ryngaert, M. 60, 659n
Pistor, K. 105
Pontiff, J. 43n, 70 Sabow, S. 470, 472
Porter, M. 22, 45n, 80, 102, 248n, 517n Sadler, G. 395n, 396n
Posner, R. 56, 344n Sahlman, W. 17, 29, 45n
Poulsen, A. 44n, 45n, 60, 224, 249n, 572, Saly, P. 420
648, 649, 659n Santerre, R. 453
Pound, J. 24, 36, 44n, 45n, 46n, 63, 126, Sapienza, P. 102
224, 493, 517n, 596n, 601, 615, Sarin, A. 200, 249n, 364, 709, 726–727
648, 659n Scharfstein, D. 64, 67, 70–73, 79, 137, 154n,
Pourciau, S. 308n 315, 710
Prabhala, N. 368, 419–448 Scharfstein, D.S. 137, 139
Pratt, J. 400 Schipper, K. 572, 574
Prowse, S. 66, 154n, 225, 653, 657, 659n Schmid, F. 66, 68, 70, 99–100, 453–454,
Pruitt, S. 210 456, 462, 463n, 495–496, 518n, 716
Pugh, W. 553n Schmidt, K. 70, 495
Puia, G. 560, 562 Schmidt, R. 58, 280n, 296n, 339, 364, 389
Scholes, M. 374–376, 381, 383–384, 390,
Rahi, R. 138 395n, 399, 402–404, 406, 409–410, 412,
Rajan, R. 69–70, 74, 101, 107n, 115–116, 414–416, 417n, 419, 422, 428, 432, 447
129n, 571, 705n Schooley, D. 601
Ramseyer, M. 104 Schrand, C. 416, 417n
Rao, R.P. 553n Schumpeter, J. 12
Rappaport, A. 734 Schut, G. 4, 488, 557–566
Rau, P.R. 541 Schwalbach, J. 394n, 453–454, 456, 463n
Raviv, A. 45n, 69, 71, 72, 74, 99 Schwert, R. 25, 60, 552n, 672, 708
Rechner, P. 573 Sebora, T. 453
Rediker, K. 601, 605 Segerx, F. 463n
Rees, W. 620n Sen, P.K. 559–560, 564
Reese, J. 470 Senbet, L. 398, 400
Reishus, D. 288, 340, 364 Senbet, L.W. 249n
Renneboog, L. 2, 133–134, 159, 161, 170, Sengupta, S. 559–560, 564
188n, 191–218, 256, 280n, 285–310, Serrano, J. 66, 663
313–344, 364, 717 Servaes, H. 69, 164, 223, 230, 237, 243,
Rice, E. 37, 44n, 60, 75, 649, 652, 659n 248n, 249n, 258, 260, 269, 339, 344n, 470,
Richman, L. 42 571, 600, 603, 659n
Roberts, D. 452 Seth, A. 601, 605
Roberts, J. 395n, 401, 407 Seward, J. 648, 657–658
Rock, K. 605 Seyhun, H. 553n
Roe, M. 24, 36, 44n, 45n, 46n, 52, 65, 78–80, Shane, S. 562
216n, 517n, 518n, 692 Shapiro, D. 76, 495–496, 518n, 519n
Roëll, A. 133, 157–162, 260, 307n, 704n Shaw, W. 290
Rogoff, K. 61 Sheehan, D. 44n, 45n, 64–65, 154n, 289,
Roll, R. 59 321, 339–340, 572, 673
Romano, R. 52, 64, 79, 223–224, 314 Sheikh, S. 620n
Rose, M. 288 Shepro, R. 734
Rose, N. 396n Shin, H.-H. 494, 535, 553n
Rosen, S. 379, 386, 392, 395n Shivdasani, A. 66, 68, 221, 280n, 364,
Rosenbaum, V. 525–526, 548 572–573, 659n
Rosenfeld, A. 59, 541 Shleifer, A. 2, 7–9, 15, 43n, 44n, 45n, 52–82,
Rosenfeld, J. 364, 737 91–107, 126, 133, 154n, 163, 199, 207,
Rosenstein, S. 364, 573 221, 224, 285–286, 289, 315, 492–494,
Rosett, J. 43n, 70 517n, 518n, 535, 539, 541, 548, 552n,
Rozeff, M. 600–601, 603–604 553n, 571, 576, 623–625, 648, 663, 705n,
Rubach, M. 453 710, 712, 734–735, 737
756 CORPORATE GOVERNANCE AND CORPORATE FINANCE
Short, H. 228, 248n, 495, 517n, 518n, 715 Tomlinson, L. 229
Sicherman, N. 44n Torregrosa, P. 600
Siegel, D. 44n, 45n, 297 Towers, P. 472
Siegfried, J. 452 Townsend, R. 71, 138
Singh, H. 74, 221, 560, 562 Travlos, N. 256, 347–365
Sisneros, P. 605 Treisman, D. 104
Skinner, D. 626–627 Trigeorgis, L. 45n
Slatter, S. 571 Tufano, P. 400
Slovin, M. 659n Tyrell, M. 177
Smith, A. 37, 44n, 45n
Smith, C. 68, 71, 398, 399, 400, 401, 404, Udell, G. 115–116, 129n, 130n
408, 410, 474, 572, 574, 600, 605, 659
Smith, M. 224 Venkatraman, N. 559–560, 564
Snell, S. 571 Vermaelen, T. 307, 541, 553n
Song, M. 649 Vetsuypens, M. 420
Spencer, C. 663 Viénot, M. 285, 308n
Stafford, E. 553n Vietor, R. 43n
Stapledon, G. 198, 216n, 225, 228, 248n, Vijh, A. 480, 482, 541
249n, 339 Vishny, R. 2, 7–9, 45n, 52–107, 126, 133,
Starks, L. 364 154n, 163, 199, 207, 221, 224, 285–286,
Stein, J. 45n, 62, 71, 75, 80, 553n 289, 315, 492–494, 517n, 535, 539, 541,
Steiner, I. 45n 548, 552n, 553n, 571, 576, 604, 623–625,
Stephens, S. 44n, 45n 648, 663, 705n, 710, 712, 735
Stevenson, H. 17 Vives, X. 517n
Stewart, G. 35, 44n Volpin, P. 255, 257–283
Stigler, G. 53, 94, 453 Voogd, R. 651
Stiglin, L. 43n
Stiglitz, J. 58, 76, 492 Wahal, S. 223–224, 245, 247
Stomper, A. 307n Wahlen, J. 623, 626
Strangeland, D. 199 Walkling, R. 44n, 57, 59–60, 541, 649, 659n
Strong, J. 308n Walsh, J. 648, 657–658
Strong, N. 566n Ware, R. 519n
Stulz, R. 44n, 45n, 59, 69, 72, 102, 106, Warner, J. 45n, 63, 71, 73, 264, 286,
248n, 398, 400–401, 410, 493–494, 535, 296–297, 308n, 339, 364, 389, 493
541, 553n, 600, 604 Watson, B. 21
Subrahmanyam, A. 544 Watson, R. 394n
Subramanian, G. 539 Watts, R. 45n, 63, 364, 399, 408, 474, 605,
Sudarsanam, P.S. 4, 221, 343n, 488, 568–596 627, 744
Summers, L. 15, 44n, 62, 70, 80, 734 Wei, K.C.J. 553n
Sundaram, R. 420 Weigandz, J. 3, 453, 487, 491–519
Sushka, M. 659n Weinstein, D. 69, 102
Swartz, L. 25 Weisbach, M. 43n, 45n, 59, 63, 264, 286,
297, 302–303, 308n, 314, 339–340, 342n,
Tedlow, R. 44n 343n, 364, 389, 552n, 572–573, 604, 744
Teoh, S. 62, 288 Weiss, L. 64, 519n
Teubner, G. 63 Welch, I. 62
Thadden, E.L. von 72, 107n, 138, 145, 149, Wenger, E. 194, 517n, 666, 720, 730, 731n
154n, 199, 210, 494, 704n Weston, J. 600, 611
Thakor, A. 138, 155n, 737 White, H. 64, 292, 409, 412, 459, 507,
Thaler, R. 553, 596n 512–514, 619
Thomas, J.K. 553n Whitehouse, C. 620n
Thompson, J. 493 Whyte, A. 560
Thonet, P.J. 495 Wiant, K. 560
Tiemann, J. 44n Williams, J. 113, 128, 244n
Tirole, J. 54, 72, 74, 76, 138, 149, 154n, Williamson, O. 54, 56, 72, 154n, 288, 539
155n, 711 Willig, R.D. 76
Titman, S. 74, 553n Winter, R. 604
Todd, R.B. 420 Winton, A. 154n, 224
NAME INDEX 757