You are on page 1of 49

European Financial Management, Vol. 22, No.

4, 2016, 697–745
doi: 10.1111/eufm.12086

Multiple Large Shareholders and


Corporate Risk-taking: Evidence
from French Family Firms
Sabri Boubaker
Champagne School of Management, Groupe ESC Troyes en Champagne, 217, Av. Pierre Brossolette,
CS 20710, 10002 Troyes CEDEX, France; IRG, Universite Paris Est, Creteil, France
E-mail: sabri.boubaker@get-mail.fr

Pascal Nguyen
NEOMA Business School, Rouen, France
E-mail: pascal.nguyen@neoma-bs.fr

Wael Rouatbi
Paris School of Business, Paris, France
E-mail: w.rouatbi@psbedu.paris

Abstract
We investigate the role of multiple large shareholders (MLS) in corporate risk-
taking. Using a sample of publicly listed French family firms over the period
20032012, we show that the presence, number and voting power of MLS are
associated with higher risk-taking. Our results suggest that MLS help restrain the
propensity of family owners to undertake low-risk investments. This effect is much
stronger in firms that are more susceptible to agency conflicts. The results highlight

We thank Yakov Amihud, Narjess Boubakri, Jerry Bowman, John Doukas (Editor), Sadok El
Ghoul, Angelica Gonzalez, Omrane Guedhami, Elaine Hutson, Pierre-Guillaume Meon,
Samir Saadi, Walid Saffar, Christoph Schneider, Ewa Sletten, Ariane Szafarz and two
anonymous referees for their helpful comments and suggestions on earlier versions of
the paper. We are grateful for the comments made by seminar participants at Centre Emile
Bernheim (Solvay Brussels School of Economics and Management), CREM Finance
(IRG, IAE de Rennes), CRCGM (University of Auvergne), Institut de Recherche en Gestion
(University of Paris Est), European Financial Management Association (EFMA), Barcelona,
Spain (2012), Multinational Finance Society (MFS), Krakow, Poland (2012), Financial
Markets and Corporate Governance Conference, Melbourne, Australia (2012), 5th Annual
Research Symposium in Business and Economics, American University of Sharjah, UAE
(2013), 8th Portuguese Finance Network Conference, Vilamoura, Portugal (2014) and Paris
Financial Management Conference, Paris, France (2015). All errors are our own
responsibility.
© 2016 John Wiley & Sons, Ltd.
698 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

the important governance role played by MLS in family firms and may explain why
MLS are associated with higher firm performance.

Keywords: risk-taking, ownership structure, family firms, private benefits of


control, contestability, corporate governance
JEL classification: G30, G32, G34

1. Introduction

A large body of literature shows that the ownership of publicly listed firms tends to be
concentrated in the hands of a few large shareholders (Faccio and Lang, 2002;
Holderness, 2009; La Porta et al., 1999). Besides the largest controlling shareholder
(LCS), other blockholders are often present in a firm’s ownership structure. La Porta
et al. (1999) calculate that 25% of the firms in their multi-country sample have
multiple large shareholders (MLS). Claessens et al. (2000) find that 32.2% of firms in
Asia have more than one large shareholder, while Faccio and Lang (2002) reveal that
39% of firms in Europe have at least two large shareholders. These blockholders play
an important governance role. They monitor the LCS (Bolton and von Thadden, 1998;
Pagano and Roëll, 1998), compete for the firm’s control (Bloch and Hege, 2001), or
form large controlling coalitions that are better aligned with the interests of minority
shareholders (Bennedsen and Wolfenzon, 2000). On the negative side, blockholders
may collude to share divisible private benefits of control (Zwiebel, 1995) or trade on
private information (Kahn and Winton, 1998). However, the balance of evidence
suggests that the presence, number, and voting power of MLS have a positive impact
on corporate governance. Maury and Pajuste (2005), Laeven and Levine (2008) and
Attig et al. (2009) show that MLS are associated with higher firm valuation.
Similarly, Attig et al. (2008) find that the presence of MLS helps decrease a firm’s
cost of equity.
Our purpose is to investigate the role of MLS in corporate risk-taking. This issue is
critical since insufficient risk-taking is likely to result in lower firm value (Low, 2009).
Moreover, corporate risk-taking has been linked to long-term economic growth (Barro,
1991; DeLong and Summers, 1991). We hypothesise that the LCS has strong incentives
to shun risk, particularly in the case of family owners, because the LCS tends to be under-
diversified and may gain more from diverting corporate resources than from pursuing
risky investments. For family owners, retaining control is another sensitive issue because
families endeavour to pass the firm on to the next generation (Anderson and Reeb, 2003;
Burkart et al., 2003). This suggests that family firms will be biased toward conservative
strategies. Shleifer and Vishny (1986) underline that risk avoidance is one of the most
significant costs that large undiversified shareholders can impose on a firm. In this
context, the presence of MLS is expected to positively influence a firm’s risk-taking
behaviour. Other blockholders, with significant wealth at stake, have similar reasons to
monitor the firm, but fewer opportunities to extract private benefits. Hence, their main
concern should be the firm’s value creation activities, especially when they are families
or financial institutions. Accordingly, the presence of MLS should help alleviate the
negative influence of the LCS and result in higher risk-taking. Finally, we argue that the
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 699

incentives of the LCS to decrease risk-taking are stronger in poorly-governed firms,


which implies that the influence of MLS will be more significant.
Using a large sample of publicly listed French firms over the period 20032012, we
show that the presence of an LCS is associated with significantly lower risk-taking. This
result is driven by family owners, who lack adequate diversification and have strong
incentives to protect their private benefits of control (Anderson and Reeb, 2003; Faccio
et al., 2011; John et al., 2008). Furthermore, we find that the greater the excess of their
control rights over their cash-flow rights (excess control), the lower the firm’s risk-
taking. Given their prevalence, we focus on family-controlled firms to analyse the
influence of MLS. Our main result is that MLS are associated with higher risk-taking.
Consistent with Adams et al. (2005), we measure risk-taking at each point in time by the
absolute deviation from the firm’s expected performance rather than by the average
squared deviation from the firm’s average performance over the whole period. This
approach preserves the panel structure of the data instead of forcing the panel to collapse
into a single cross-section. In addition, it provides a more accurate measure of risk, since
it takes into account predictable changes in firm performance. Last but not least, a shock
affecting a firm’s performance at a given time is not carried over to the whole sample
period. Accordingly, the relationship between ownership structure and risk-taking can
be evaluated over several quasi-independent periods instead of depending on a single
observation per firm.
Our findings are robust to a number of sensitivity checks. The positive influence of
MLS remains unchanged whether we use lagged MLS variables, different proxies for
MLS (e.g., the Shapley value), or different measures of risk-taking (e.g., return on assets
volatility and research and development (R&D) intensity). Our core evidence remains
unchanged when we include additional firm-level control variables (e.g., an indicator of
group affiliation) and after matching firms based on their propensity to have multiple
blockholders. Furthermore, we show that the presence of MLS is more effective when
the firm’s internal governance is weak. More precisely, MLS appear to deter firms from
reducing risk when larger and less independent boards allow greater deviation from
optimal risk-taking. Finally, we find that blockholder identity matters. In particular, the
presence of a second largest shareholder is associated with higher risk when he/she is a
financial institution or another family.
This study makes three contributions to the literature. First, we extend Mishra’s
(2011) analysis of emerging markets by showing that MLS are associated with higher
risk-taking in a developed economy. Indeed, the French firms in our sample originate
from an economy with sophisticated financial institutions and effective legal systems
(Djankov et al., 2008; La Porta et al., 1998). Estimates of control premiums by Dyck
and Zingales (2004) suggest that, in France, the extraction of private benefits is
restrained. Faccio et al. (2001) show that European firms pay higher dividends, which
is consistent with blockholder monitoring, whereas Asian firms pay lower dividends,
implying collusion among blockholders. This suggests that the effect of MLS might be
different in a European country. Consistent with this argument, Attig et al. (2008) find
that MLS help decrease a firm’s cost of equity in Asia, but not in Europe. La Porta
et al. (1999) and Claessens et al. (2002) underscore the fact that, in Asia, poor
corporate disclosure and weak enforcement of the rule of law greatly facilitate the
expropriation of minority shareholders. However, our results indicate that MLS are
able to play a positive role in a context where the extraction of private benefits is
more constrained.
© 2016 John Wiley & Sons, Ltd.
700 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Second, we show that MLS can enhance the firm’s internal governance. When
internal governance structures are weak and offer many opportunities to deviate from
the optimal policy, MLS contribute effectively by containing the firm’s tendency to
decrease risk-taking. This result is consistent with that of Low (2009), who finds that
weaker monitoring from the takeover market leads to a decrease in risk-taking by US
firms. However, adequate incentives help curb distortions in risk due to management–
shareholder conflicts. Third, our results suggest that governance is better enforced by
having financial institutions and other families as second largest blockholders. In other
words, some types of blockholders appear to be more effective at governance than
others.
Overall, our results confirm that MLS play a major role in a firm’s governance. By
monitoring the LCS and restraining his/her bias toward conservative investments, MLS
help enhance firm performance, consistent with the evidence presented by Attig et al.
(2009), Jara-Bertin et al. (2008), Laeven and Levine (2008), and Maury and Pajuste
(2005). Our study also extends the literature on corporate risk-taking. Prior studies have
highlighted the role of management ownership (Denis et al., 1997), management
compensation (Coles et al., 2006; Guay, 1999; Rajgopal and Shevlin, 2002), board size
(Cheng, 2008; Nakano and Nguyen, 2012), chief executive officer (CEO) power (Adams
et al., 2005), takeover threats (Low, 2009), investor protection (John et al., 2008), and
creditor rights (Acharya et al., 2011). We contribute to this line of research by showing
that ownership structure and, more specifically, the presence of MLS are a major
determinant of a firm’s risk-taking behaviour.
The remainder of the paper proceeds as follows. In section 2, we analyse the
relationship between ownership structures and corporate risk-taking and articulate a
number of testable hypotheses. Section 3 describes the sample and methodology. Section
4 presents the empirical results. Section 5 concludes the paper.

2. Literature Review and Hypotheses

In this section, we review the literature and outline the implications of ownership
structures where the LCS has excessive power over minority shareholders. We then show
the consequences of the presence of other blockholders (or MLS) with sufficient voting
power to curb the influence of the LCS.

2.1. Risk-taking with a single large shareholder


Firms with an LCS are likely to decrease risk-taking for two reasons. The first
reason is that the LCS tends to be under-diversified. Dyck and Zingales (2004)
observe that maintaining a controlling block requires the dominant shareholder
to not be well diversified. Accordingly, Zhang (1998) argues that the largest
shareholder may reject risky projects that investors with fully diversified portfolios
would find desirable. Paligorova (2010) agrees that unless he/she holds substantially
diversified equity stakes in other firms, the largest shareholder has little incentive to
take risk. Faccio et al. (2011) substantiate the fact that firms controlled by under-
diversified large shareholders undertake less risky projects than diversified large
shareholders do. Similarly, Belenzon and Berkovitz (2010) show that the affiliates
of a business group are more innovative since risk can be distributed across the
group.
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 701

The second reason is that the LCS has incentives to extract private benefits (Bebchuk
et al., 2000; Claessens et al., 2002; Shleifer and Vishny, 1997). Bae et al. (2002),
Bertrand et al. (2002), Cheung et al. (2006) and Johnson et al. (2000) provide evidence of
connected party transactions instigated by the LCS. Baek et al. (2006) suggest that
private security offerings can also be used to expropriate minority shareholders.1
Claessens et al. (2002), Cronqvist and Nilsson (2003) and Lemmon and Lins (2003)
indicate that expropriation activities by the LCS induce a significant discount on firm
value. The larger the amount of private benefits controlling shareholders extract, the
more they will strive to protect these benefits and the less they will tend to make risky
investments (John et al., 2008). In a way, the ability to divert corporate resources
stimulates rent seeking  the appropriation of existing wealth  instead of spurring
innovation to produce wealth. This argument is supported by the negative impact of
ownership concentration on a firm’s R&D intensity and patenting outcome (Ortega
Argiles et al., 2005). We thus posit the following hypothesis:
H1: The presence of an LCS is associated with lower risk-taking.

When the LCS is a family, his/her incentives to decrease firm risk are even more
compelling. Because of the undiversified nature of their holdings, families have strong
reasons to prefer lower risk-taking. Villalonga and Amit (2006) argue that families are
inclined to diversify their firms to make up for their lack of personal diversification.
Chandler (1990) and Anderson et al. (2012) show that US family firms prefer to invest in
physical assets rather than riskier R&D projects. Similarly, Croci et al. (2011) show that
family control is associated with lower R&D expenses in Europe. Morck and Yeung
(2004) presume that concerns for the destabilising effect of rapid growth make families
conservative and suspicious of innovation. Luo and Chung (2005) observe that family
firms are more cautious when it comes to investing in industries with which they are not
familiar.
The incentives to extract control benefits are also stronger in family firms because
the benefits are for family members alone, instead of being shared with a large
number of distinct owners (Villalonga and Amit, 2006). Demsetz and Lehn (1985)
emphasise the non-cash benefits that may be appropriated by controlling shareholders.
In this particular case, families have the opportunity to hand out executive positions to
family members. Schulze et al. (2001) discuss favoritism toward family members
through, for example, privileged employment and promotions. Bennedsen et al.
(2007) show that, in Denmark, firstborn males are associated with a higher likelihood
of family succession. These significant control benefits imply that family firms are
less likely to take risks. In addition, Lins et al. (2013) reveal that, in times of financial
crisis, family-controlled firms are biased toward survival-oriented actions that
help preserve the family’s control benefits at the expense of outside shareholders. In
the same context, Attig et al. (2016) show that family firms pay lower dividends and
are more likely to decrease/cut dividends, consistent with the expropriation
hypothesis.

1
A detailed example in France is given by the murky way the Bouygues brothers gained
control of the company that bears their name (The Economist, “French corporate governance:
Creative construction,” November 30, 2006).

© 2016 John Wiley & Sons, Ltd.


702 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Besides, families tend to view their firms as an asset to pass on to their descendants and
are therefore reluctant to relinquish control.2 According to Villalonga and Amit (2006),
control-enhancing mechanisms reflect family resistance to the dilution of their ownership
stake as the firm grows. Similarly, Caprio et al. (2011) argue that families are wary of
acquisitions, which can dilute their voting power, either immediately (if the acquisition is
paid in shares) or later (if the acquisition is paid in cash, because of the higher probability
of future equity issues). Interestingly, their study also highlights the fact that family firms
are less likely to be acquired. Cespedes et al. (2010) find that family firms in Latin
America tend to prefer debt to equity when losing control is an issue. Due to this
reluctance to open up their equity capital, family firms rely mostly on internal cash-flows
to finance their growth, which suggests that they will strive to generate steady cash
inflows by minimising risk. We thus posit the following hypothesis:3
H2: Family control is associated with lower risk-taking.

Focusing again on the LCS, we expect the divergence between the LCS’s cash-flow
and control rights to decrease firm risk-taking because it exacerbates the incentives to
extract private benefits of control. La Porta et al. (1999) and Bebchuk et al. (2000)
indicate that excess control reflects the controlling shareholder’s ability to expropriate
minority shareholders. Faccio et al. (2001) argue that the probability of minority
shareholder expropriation is particularly high if large shareholders hold voting rights in
excess of their cash-flow rights. Consistent with this line of reasoning, Claessens et al.
(2002) show that firm value decreases with the separation of cash-flow and control rights
of the LCS. Bennedsen and Nielsen (2010) demonstrate that firms with disproportional
ownership structures have lower valuations. Moreover, the smaller the controlling
shareholder’s cash-flow rights, the larger the discounts. Belkhir et al. (2014) analyse the
value of cash holdings in listed French firms and show that it dramatically decreases with
the separation of the LCS’s ownership and control rights. Doidge et al. (2009) find that
foreign firms are less likely to list on a US exchange when their controlling shareholders’
control rights exceed their cash-flow rights, because a US listing would curtail their
ability to extract private benefits of control.
Considering the large private benefits associated with disproportional ownership,
dominant shareholders are expected to try hard to protect these benefits by holding down
firm risk. Besides, Villalonga and Amit (2006) argue that excess control reflects a
resistance to the dilution of ownership. This argument similarly implies lower risk-taking

2
Struggling car manufacturer Peugeot offered a revealing case in 2014. Despite the
company’s pressing need to shore up its balance sheet following steep operating losses,
members of the Peugeot family voiced strong opposition to a s3 billion (US$ 4.1 billion)
capital increase, arguing that this would critically dilute the family’s control. The proposed
equity issue to Peugeot’s Chinese partner, Dongfeng Motor, and to the French State would
give all three blockholders an equal 14% stake in the company. Until then, the Peugeot family
had held 25.5% of the shares but controlled 38% of the votes (Bloomberg, “Peugeot to pursue
US$ 4.1 billion capital increase”, 21 January 2014).
3
A theoretical distinction between the different motives for lower risk-taking is beyond the
scope of the present study. The empirical results, however, strongly point to the extraction of
private benefits by the LCS (or family owners) as a major reason for a firm’s lower risk-
taking.

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 703

to reduce the possibility of future equity issues. Chin et al. (2009) provide evidence that
excess control inhibits risk-taking. They find that excess control is negatively related to
patenting activity in Taiwan’s electronics industry. The positive relationship between
excess control and cash holdings documented by Ozkan and Ozkan (2004) in UK firms
also seems to suggest that excess control is detrimental to corporate risk-taking. We thus
propose the following hypothesis and note that it applies well to the case where the LCS
is a family:4

H3: Excess control rights by the LCS are associated with lower risk-taking.

2.2. Risk-taking with MLS


In firms with an LCS, and family firms in particular, the presence of other large
shareholders restrains the influence of the LCS and increases the power of minority
shareholders to contest corporate decisions that are detrimental to their interests. Pagano
and R€ oell (1998) argue that other blockholders limit the expropriation of a firm’s
resources by monitoring the dominant shareholder. Bloch and Hege (2001) suggest that
competition for corporate control requires enlisting the support of minority shareholders,
which results in a reduction in the consumption of private benefits. Bennedsen and
Wolfenzon (2000) assert that balanced ownership structures composed of MLS require
larger coalitions to secure control. Consequently, the winning coalition internalises more
of the costs of expropriation, which also leads to fewer incentives to extract private
benefits. Using a different argument, Edmans and Manso (2011) propose that other
blockholders discipline the controlling shareholder through trading. Should the firm
make suboptimal decisions, they could sell their stake and penalise the controlling
shareholder by pushing down firm value.
An alternative view is that blockholders collude to extract divisible private benefits at
the expense of minority shareholders (Zwiebel, 1995). An example of this might be two
families who share board positions and run the firm in their common interest. Consistent
with the collusion argument, Faccio et al. (2001) find that the presence of multiple
blockholders is associated with lower dividend rates in Asia. Another drawback of
multiple blockholders is that monitoring may be less effective because of the free-rider
problem (Winton, 1993) or because blockholders prefer to trade opportunistically using
their private information (Kahn and Winton, 1998). In addition, disagreement among
blockholders regarding hard-to-evaluate projects could result in the loss of valuable
investment opportunities (Gomes and Novaes, 2005).
However, the balance of evidence suggests that the presence, number and relative
power of large shareholders other than the LCS are beneficial to minority shareholders.
Attig et al. (2009), Jara-Bertin et al. (2008), Laeven and Levine (2008) and Maury and
Pajuste (2005) show that more balanced ownership structures are associated with higher
firm valuation. Attig et al. (2008) reveal that MLS are associated with lower cost of
equity estimates. Lehmann and Weigand (2000) indicate that the existence of a second

4
In this paper, the tests of H3 focus on family owners since they represent the overwhelming
majority of LCSs in our sample of French firms. Nonetheless, we state the hypothesis in the
broader case of firms with an LCS because the arguments are not restricted to family owners.
The results remain qualitatively unchanged if we run regressions using all firms with an LCS.

© 2016 John Wiley & Sons, Ltd.


704 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

large shareholder is associated with higher firm profitability in Germany, while Faccio
et al. (2001) find that the presence of MLS increases dividend rates in Europe. Further,
Attig et al. (2013) observe that the value of cash holdings correlates positively with a
more balanced distribution of voting power among blockholders.
The way other large shareholders can influence firm value positively is by challenging
suboptimal decisions, such as holding excess cash or using too little leverage. These
decisions are beneficial to the LCS because of his/her under-diversified wealth.
However, for outside shareholders, excess cash is associated with a higher probability of
misappropriation and translates into lower firm valuation (Dittmar and Mahrt-Smith,
2007). Similarly, using low leverage involves the loss of valuable tax shields. The other
large shareholders can also disapprove of excessive diversification and ask the firm to
sell unrelated businesses to focus on its core strengths.5 The other large shareholders may
also question the firm’s ability to remain competitive without making risky but critical
investments. By doing so, the other large shareholders are expected to induce higher risk-
taking. In line with this argument, Mishra (2011) shows that firms with MLS are
associated with higher cash-flow volatility in Asia. Similarly, we posit the following
hypothesis:
H4: The presence, number and relative power of blockholders other than the LCS are
associated with higher risk-taking.

The intensity with which large shareholders monitor a firm is likely to vary according
to their type. Almazan et al. (2005) and Chen et al. (2007) show that, even within the
same category of institutional investors, some monitor while others do not. Laeven and
Levine (2008) hypothesise that large shareholders are less likely to collude when they are
of different types. In line with their analysis, we distinguish four types of shareholders:
families, widely held financial institutions, widely held corporations and the State. Their
effect on risk-taking is expected to depend on their ability to monitor the firm and the
actions of its controlling shareholder.
As non-controlling blockholders, families have strong incentives to monitor, since
they have a substantial fraction of their wealth invested in the firm. At the same time,
their scope for entrenchment is limited, unless they collude with the LCS to extract
private benefits at the expense of other shareholders. While this might happen in
emerging markets, it seems less likely to occur in developed economies (Claessens et al.,
2002; Dyck and Zingales, 2004; Faccio et al., 2001). Besides, non-controlling families
should be less emotionally attached to the firm and thus less likely to be obsessed with
implementing survival strategies than if they were in control. Accordingly, they are
expected to positively influence corporate risk-taking by contesting the defensive
strategies instigated by the LCS. Widely held financial institutions are also expected to
have a positive effect on corporate risk-taking. The decision to invest a large proportion
of their funds in a specific company is likely to reflect the intention to exert some form of

5
For example, back in 1998, Bollore, the second largest shareholder in Bouygues, with a
12.7% stake, openly urged Europe’s leading construction firm to put its telecommunications
division up for sale. This division was deemed too small relative to its rivals and was badly in
need of large infrastructure spending that was depressing its parent’s earnings (Business
Week, “Building Euro Empires”, 13 December 1998).

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 705

control over the firm to ensure that their investment performs well. Various examples
show that non-controlling financial institutions are effective monitors.6
In contrast, the ability of the State and widely held corporations to undertake the same
level of monitoring and ensure optimal performance is highly debatable. State ownership
often involves other objectives, such as preserving employment or protecting the
national interest (Alchian, 1977; Kole and Mulherin, 1997). For this reason, government
ownership has been associated with lower profitability (Dewenter and Malatesta, 2001).
Ownership by non-financial firms is akin to cross-shareholdings, which have been
associated with management entrenchment. Drago et al. (2015) argue that they facilitate
the expropriation of minority shareholders. On the other hand, Guth et al. (2007) insist
that vertical cross-shareholdings create value by enabling trade between buyers and
sellers. In fact, the role of corporate ownership is to strengthen a business relationship
and elicit buyer-specific investments on the part of the investee.7 We recognise the
heterogeneity among non-controlling blockholders in the following hypothesis:
H5: As non-controlling blockholders, families and widely held financial institutions are
associated with higher risk-taking, while the State and widely held corporations are not.

A firm’s internal governance structure is likely to shape the behaviour of the LCS. In
firms with strong governance mechanisms, the LCS has little scope for extracting private
benefits. As a result, his/her negative impact on firm valuation is expected to be small.
The LCS’s ability to decrease the firm’s risk-taking should also be trivial. MLS are
therefore likely to exert only a minor influence. On the other hand, in firms with poor
governance, the opportunities for the LCS to extract private benefits should be numerous
and easier to conceal. Accordingly, MLS are expected to play a more prominent role in
disciplining the LCS and preventing the latter from lessening the firm’s risk-taking.
Consistent with this reasoning, Attig et al. (2008) show that MLS contribute to
decreasing the cost of equity capital in East Asia, where agency problems are more
severe, but not in Europe, where the potential for expropriation is less pronounced
(Faccio and Lang, 2002).
Regarding the effect on risk-taking, Low (2009) demonstrates that the weaker market
discipline resulting from the introduction of stricter takeover rules in the US is associated

6
For example, when Colony Capital invested s1 billion in the French hotel group Accor in
March 2005, it also received two seats on the board. The investment consisted of bonds
redeemable in shares and convertible bonds. So until they were redeemed or exercised, the
largest shareholder was Caisse des Dep^ots, a State-controlled investment company.
Nonetheless, Accor’s strategy took a radical turn with the sale of non-core assets and the spin-
off of its highly profitable services business. In addition, the hotel group stepped up its policy
of selling and leasing back its properties. While no public statements were made linking the
firm’s decisions to its new shareholder, there was little doubt that Colony was the driving
force behind Accor’s revised strategy (Le Monde, “Le fond Colony Capital investit 1 milliard
d’euros chez Accor”, 10 March 2005).
7
For instance, when Dassault Aviation, the maker of the Rafale fighter jet, bought a 20.8%
stake in Thales, the move was seen as ensuring that the European leader in electronic defence
systems would devote its full attention to the interests of its new parent, despite the fact that
the latter accounted for less than 5% of its sales (Les Echos, “Dassault-Thales: les raisons
d’un mariage”, 12 January 2009).

© 2016 John Wiley & Sons, Ltd.


706 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

with a significant decrease in corporate risk-taking. However, managers incentivised to


create shareholder value did not materially decrease firm risk, despite the opportunity to
do so. In other words, better governance helped to mitigate the incentives to deviate from
the optimal policy and led to unchanged risk-taking, while weak governance allowed
managers to look after their own interests and cut firm risk. Following the same line of
thought, the presence of MLS is expected to have a stronger effect in weak governance
environments because the LCS can more easily decrease firm risk to suit his/her risk
preferences. Hence, we propose the following hypothesis:

H6: In firms with weak governance structures, the presence of MLS is associated with
higher risk-taking than in firms with strong governance structures.

3. Data and Methodology

In this section, we describe the sample selection process and data sources. We then
present the method adopted to construct the ultimate ownership and control variables
used in the analysis. The main characteristics of the sample are provided next. Finally, we
describe the method for estimating firm risk.

3.1. Sample
We start by considering all French listed firms appearing in the Worldscope database
over the period 20032012. From that sample, we exclude (1) financial firms (with a
Standard Industrial Classification (SIC) code between 6000 and 6999), (2) firms with
fewer than two usable observations for the whole sample period, (3) and firm–year
observations with missing or incomplete ownership, return, or financial data. These
restrictions result in a final sample of 4,501 firm–year observations, of which 3,446
correspond to family firms (where the LCS with more than 10% of the voting rights is an
individual or a family). Ownership and voting data are from the firms’ annual reports.
Financial data are sourced from Worldscope, while stock returns are from Datastream.
Table 1 shows the distribution of the sample by industry and by year. The industry
classification is based on that of Campbell (1996). Four industries stand out: services
(27.55%), consumer durables (16.46%), capital goods (10.6%) and textile and trade
(9.98%). These industries account for 64.59% of the sample. Each year is evenly
represented with about 1/10 of all observations.

3.2. Ultimate ownership and control rights of the LCS


For each firm in our sample, we compute the ultimate cash-flow rights (UCF) and the
ultimate control rights (UCO) of the LCS as follows. First, we determine the shareholder
who controls the largest block of direct voting rights. Second, we identify that
shareholder’s largest direct shareholder and we repeat this procedure until we reach the
ultimate LCS of each firm. LCSs are classified into four types, namely, families, the
State, widely held corporations and financial institutions (Claessens et al., 2002). Finally,
we use all ownership and control chains to compute the ultimate owner’s UCF and UCO.
Following Claessens et al. (2002), we calculate UCO by summing the weakest links
along each control chain and using a 10% threshold. UCF are obtained by summing the

© 2016 John Wiley & Sons, Ltd.


Table 1
Distribution of the sample across industries and years

This table shows the distribution of the 4,501 sample firm–year observations across industries (based on Campbell’s (1996) industrial classification) and years. The
sample consists of 602 non-financial French listed firms over the period 20032012. Financial firms (SIC 6069) are excluded. The industries are petroleum (SIC
13, 29), consumer durables (SIC 25, 30, 36, 37, 50, 55, 57), basic industry (SIC 10, 12, 14, 24, 26, 28, 33), food and tobacco (SIC 1, 2, 9, 20, 21, 54), construction
(SIC 15, 16, 17, 32, 52), capital goods (SIC 34, 35, 38), transportation (SIC 40, 41, 42, 44, 45, 47), utilities (SIC 46, 48, 49), textiles and trade (SIC 22, 23, 31, 51, 53,

© 2016 John Wiley & Sons, Ltd.


56, 59), services (SIC 72, 73, 75, 76, 80, 82, 87, 89) and leisure (SIC 27, 58, 70, 78, 79).

Industry (SIC codes) 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 Total % sample
Petroleum (13, 29) 3 3 4 4 4 4 4 4 5 5 40 0.89
Consumer durables (25, 30, 36, 37, 50, 55, 57) 70 78 78 81 79 76 73 72 69 65 741 16.46
Basic industry (10, 12, 14, 24, 26, 28, 33) 42 45 43 48 43 41 41 39 42 42 426 9.46
Food and tobacco (1, 2, 9, 20, 21, 54) 30 30 30 29 29 29 28 27 26 23 281 6.24
Construction (15, 16, 17, 32, 52) 23 23 22 21 19 19 18 18 18 18 199 4.42
Capital goods (34, 35, 38) 48 50 46 48 47 45 46 47 50 50 477 10.60
Transportation (40, 41, 42, 44, 45, 47) 9 11 11 14 13 13 13 13 13 12 122 2.71
Utilities (46, 48, 49) 19 21 24 30 28 27 26 26 24 24 249 5.53
Textiles and trade (22, 23, 31, 51, 53, 56, 59) 55 56 52 50 42 40 39 40 38 37 449 9.98
Services (72, 73, 75, 76, 80, 82, 87, 89) 112 118 126 135 128 124 123 124 127 123 1,240 27.55
Leisure (27, 58, 70, 78, 79) 30 31 30 27 28 28 27 26 25 25 277 6.15
Total 441 466 466 487 460 446 438 436 437 424 4,501 100
% sample 9.80 10.35 10.35 10.82 10.22 9.91 9.73 9.69 9.71 9.42 100 –
Multiple Large Shareholders and Corporate Risk-taking
707
708 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Fig. 1. Example of ultimate cash-flow (UCF) rights and ultimate control (UCF) rights calculations
in a three-tier pyramid. Oi,j(Cij) indicates the direct cash-flow (control) rights of entity i in entity j.
In this figure, the family is the largest (ultimate) controlling shareholder of firm B. Its ultimate cash-flow
rights UCFFamily,B equals the sum of products of direct cash-flow rights along the different ownership
chains; i.e., UCFFamily,B ¼ (OFamily,A  OA,B) þ OFamily,B ¼ 35%. Its ultimate control rights UCOFamily,
B is the sum of weakest links along each control chain; i.e., UCOFamily,B ¼ min (CFamily,A; CA,B) þ
CFamily,B ¼ 70%. The excess control of the family is the difference between UCFFamily,B and
UCOFamily,B, all divided by UCOFamily,B; i.e., ((UCOFamily,B  UCFFamily,B) / UCOFamily,B) ¼ 50%.

products of direct cash-flow rights along the different ownership chains. To illustrate
this point, we consider a firm B directly owned by another firm A that holds 60% of its
cash-flow rights and control rights; that is, OA,B ¼ CA,B ¼ 60% (see Figure 1). Firm A is
itself controlled by a family that directly owns 50% of its cash-flow rights and 70% of
its control rights; that is, OFamily,A ¼ 50% and CFamily,A ¼ 70%. The family also directly
owns 5% (10%) of firm B’s cash-flow (control) rights; that is, OFamily,B ¼ 5% and CFamily,
B ¼ 10%. The family is the LCS of firm B. Its UCF, UCFFamily,B, equals the sum of the
products of the direct cash-flow rights along the different ownership chains; that is,
UCFFamily,B ¼ (OFamily,A  OA,B) þ OFamily,B ¼ 35%. Its UCO, UCOFamily,B, is the sum
of the weakest links along each control chain; that is, UCOFamily,B ¼ min (CFamily,A; CA,
B) þ CFamily,B ¼ 70%. The excess control of the family, EXCESS CONTROLFamily,B, is
the difference between UCOFamily,B and UCFFamily,B, all divided by UCOFamily,B; that is,
EXCESS CONTROLFamily,B ¼ (UCOFamily,B  UCFFamily,B)/UCOFamily,B ¼ 50%.

3.3. Definitions of other variables


To proxy for the degree of separation between the ultimate control and cash-flow rights
of the LCS, we use the variable EXCESS CONTROL. This variable is defined as the
difference between the LCS’s ultimate control and cash-flow rights, divided by his/her
ultimate control rights (i.e., EXCESS CONTROL ¼ (UCO  UCF)/UCO). Consistent
with the work of Attig et al. (2008), we define several variables reflecting the presence,
number and voting power of MLS. The first variable, MLSD, takes the value of one if the
firm has at least two large shareholders, and zero otherwise. A large shareholder is a legal
entity that controls, directly or indirectly, at least 10% of the firm’s voting rights
(La Porta et al., 2002). We also consider a second variable, MLSN, measuring the number

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 709

of large shareholders other than the LCS, up to the fourth largest. To measure control
contestability, we use the sum of the voting rights of the second, third and fourth largest
blockholders (VR234) and the ratio of this sum to the voting rights of the LCS
(VRRATIO). To proxy for control dispersion, we use the Herfindahl index
(HERFINDAHL), calculated as follows:

HERFINDAHL ¼ ðV R1  V R2Þ2 þ ðV R2  V R3Þ2 þ ðV R3  V R4Þ2 ð1Þ

Where VR1, VR2, VR3 and VR4 are the voting rights of the first, second, third and fourth
largest shareholders, respectively. Higher values for the index suggest lower control
contestability of the LCS.
For each firm, we also compute the following variables: (i) firm size (SIZE) is
measured by the natural logarithm of total assets, (ii) capital expenditures divided by
sales (CAPEX) are used as a proxy for growth opportunities, (iii) firm age (AGE) is equal
to the number of years since the firm’s first date of incorporation, (iv) financial leverage
(LEVERAGE) is measured by the ratio of total debt over total assets, and (v)
diversification (DIVERSIFICATION) is equal to the number of business segments in
which the firm operates (using two-digit SIC codes). The Appendix provides the
definitions and data sources for the variables used in this study.

3.4. Measure and determinants of firm risk


Consistent with Adams et al. (2005), we measure risk-taking by the absolute deviation of
the firm’s performance relative to its expected value. This procedure, known as the
Glejser (1969) heteroskedasticity test, has the advantage of preserving the panel
structure of the data, since it generates an indicator of risk at each point in time. In
contrast, the standard deviation of performance used in most studies on risk-taking
collapses the panel data into a single cross-section. Second, the deviation is not taken
with respect to the sample period average, which is constant, but with respect to the
firm’s expected performance, which is time-varying, and takes into account the
predictable changes in a firm’s performance (through the effect of changes in
the determinants of performance detailed below). Third, the difference to the predicted
value is not squared, which reduces the influence of outliers. Furthermore, three
measures of performance are used: (1) return on assets (ROA), defined as the ratio of
earnings before interest and taxes to the book value of assets at the beginning of the
year; (2) Tobin’s Q, proxied by the market-to-book value of assets at the end of the year;
and (3) stock returns observed on a monthly basis.
Running Glejser heteroskedasticity tests involves two steps. The first step requires a
model of performance to determine the firm’s expected performance at each point in
time. For ROA and Tobin’s Q, we use the following models involving firm–year
observations (the subscripts are dropped for notational convenience):

ROA ¼ a0 þ a1 MLSV AR þ a2 EX CESS CONTROL þ a3 SIZE


þ a4 CAPEX þ a5 AGE þ a6 LEV ERAGE
þ a7 DIV ERSIFICATION þ h INDUSTRY þ f Y EAR þ u ð2Þ

© 2016 John Wiley & Sons, Ltd.


710 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Q ¼ a0 þ a1 MLSV AR þ a2 EX CESS CONTROL þ a3 SIZE þ a4 CAPEX


þ a5 AGE þ a6 LEV ERAGE þ a7 DIV ERSIFICATION þ a8 ROA
þ a9 ROAt1 þ h INDUSTRY þ f Y EAR þ u ð3Þ

where MLSVAR is one of the variables MLSD, MLSN, VR234, VRRATIO or


HERFINDAHL, while INDUSTRY (YEAR) denotes a vector of two-digit SIC industry
(year) dummies. To predict monthly stock returns, we use the standard market model,

R ¼ b RM þ u ð4Þ

where RM is the return on the market portfolio proxied by the Societe des Bourses
FranSc aises (SBF) 250 index.
The second step consists of running the following regression:

jb
u j ¼ g0 þ g1 MLSV AR þ g2 EX CESS CONTROL þ g3 SIZE þ g4 CAPEX
þ g5 AGE þ g6 LEV ERAGE þ g7 DIV ERSIFICATION
þ h INDUSTRY þ f Y M þ e ð5Þ

where the dependent variable, |^u|, is the absolute value of the residuals from equations (2)
to (4). Note that, when we use the residuals from equations (2) and (3) (equation (4)), the
variable YM is a vector of year (month) dummies. Equation (5) is estimated using ordinary
least squares regressions with cluster effects at the firm level. We do not use fixed firm
effects because, for most firms, ownership structure changes slowly over time. Therefore,
the fixed effects estimator would fail to detect the influence of excess control and MLS on
corporate risk-taking, even if it actually exists (Adams et al., 2005; Zhou, 2001).

3.5. Summary statistics


Table 2 provides descriptive statistics for the ownership variables and firm character-
istics. Panel A indicates that most French firms (92.1%) have an LCS. Laeven and Levine
(2008) observe that only 4.2% of French firms have no controlling shareholder, which is
low in comparison with other European countries. Consistent with the work of Boubaker
and Labegorre (2008), the LCS is typically a family and his/her average voting rights are
almost 20% higher than his/her cash-flow rights.8 Due to the 2008 financial crisis, the
performance indicators turn out to be uncharacteristically weak. Panel B indicates that
the average ROA is less than 2.5%. Average stock returns are about 1.2% per month, but
only half of the returns are positive.
We then focus on family firms. Panel C of Table 2 reveals that 40.2% of French
family firms have MLS. This proportion is very close to the proportion of 38.2%
indicated by Laeven and Levine (2008). Similarly, Faccio and Lang (2002) observe that
39% of Western European firms have more than one large shareholder at the 10%
threshold. The average voting rights of the three largest shareholders beyond the LCS is
27.1%. Taken together, these voting rights represent only 41.6% of the LCS’s voting

8
Note the presence of negative values for excess control. This is due to the fact that, in some
firms, voting rights are capped to prevent a potential takeover (e.g., Danone) or to cope with
specific industry laws (e.g., M6 Television).

© 2016 John Wiley & Sons, Ltd.


Table 2
Summary statistics for ownership variables and firm characteristics

This table provides summary statistics for the whole sample (Panels A and B) and for family firms (Panels C and D). LCS equals 1 if the firm has a large controlling
shareholder at the 10% threshold, and 0 otherwise. LCS_FAMILY, LCS_STATE, LCS_CORPORATE and LCS_FINANCIAL equal 1 if the controlling owner is a
family, the State, a widely held corporation or a widely held financial institution, respectively, and 0 otherwise. MLSD equals 1 if the firm has at least two large
shareholders (at the 10% threshold), and 0 otherwise. EXCESS CONTROL is the difference between the ultimate control and cash-flow rights divided by the ultimate

© 2016 John Wiley & Sons, Ltd.


control rights of the LCS. MLSN is the number of large shareholders, other than the LCS, up to the fourth. VR234 is the sum of voting rights of the second, third and
fourth largest shareholders. VRRATIO is the sum of voting rights of the second, third and fourth largest blockholders divided by the voting rights of the LCS.
DISPERSION equals the sum of squared differences between the voting rights of the four largest shareholders, that is, (VR1  VR2)2 þ (VR2  VR3)2 þ (VR3 
VR4)2; where VR1, VR2, VR3 and VR4 equal the voting rights of the first, second, third and fourth largest shareholders. ROA is earnings before interest and taxes divided
by the book value of assets at the beginning of the year. Tobin’s Q is the ratio of market to book value of assets. AGE is the number of years since the firm’s first date of
incorporation. LEVERAGE is total debt over total assets. CAPEX is capital expenditures divided by sales. DIVERSIFICATION is the number of business segments.
FAMILY_2, STATE_2, CORPORATE_2 and FINANCIAL_2 equal 1 if the second largest shareholder is a family, the State, a widely held corporation or a widely held
financial institution, respectively, and 0 otherwise. VR2_FAMILY, VR2_STATE, VR2_CORPORATE and VR2_FINANCIAL equal the voting rights of the second
largest shareholder if he/she is a family, the State, a widely held firm or a widely held financial institution, respectively, and 0 otherwise.

Number of 25th 75th


observations Mean S.D. Min percentile Median percentile Max
Panel A: Ownership variables (all firms)
LCS 4,501 0.921 0.270 0.000 1.000 1.000 1.000 1.000
LCS_FAMLY 4,501 0.766 0.423 0.000 1.000 1.000 1.000 1.000
LCS_STATE 4,501 0.051 0.218 0.000 0.000 0.000 0.000 1.000
LCS_CORPORATE 4,501 0.038 0.193 0.000 0.000 0.000 0.000 1.000
LCS_FINANCIAL 4,501 0.066 0.246 0.000 0.000 0.000 0.000 1.000
Multiple Large Shareholders and Corporate Risk-taking

MLSD 4,501 0.385 0.487 0.000 0.000 0.000 1.000 1.000


EXCESS CONTROL 4,501 0.192 0.213 0.251 0.000 0.159 0.302 0.952
Panel B: Firm characteristics (all firms)
ROA (in %) 4,501 2.482 9.447 42.805 0.635 3.921 6.872 38.360
Tobin’s Q 4,501 1.370 0.677 0.592 0.971 1.184 1.556 4.928
711
Table 2
712

Continued

Number of 25th 75th


observations Mean S.D. Min percentile Median percentile Max
Panel B: Firm characteristics (all firms)

© 2016 John Wiley & Sons, Ltd.


Stock return 49,850 0.012 0.316 1.000 0.049 0.000 0.057 0.412
Market return 49,850 0.003 0.049 0.144 0.026 0.014 0.034 0.131
AGE 4,501 42.920 32.289 1.000 17.099 29.348 72.044 100.000
LEVERAGE 4,501 0.213 0.167 0.000 0.074 0.193 0.317 0.763
Total Assets (s billion) 4,501 2.250 5.592 0.008 0.038 0.145 0.838 246.631
CAPEX 4,501 0.063 0.119 0.000 0.014 0.031 0.059 1.120
DIVERSIFICATION 4,501 2.464 1.462 1.000 1.000 2.000 3.000 8.000
Panel C: Ownership variables (family firms)
EXCESS CONTROL 3,446 0.209 0.198 0.251 0.041 0.185 0.311 0.952
MLSD 3,446 0.402 0.490 0.000 0.000 0.000 1.000 1.000
MLSN 3,446 0.495 0.681 0.000 0.000 0.000 1.000 3.000
VR234 1,387 0.271 0.101 0.100 0.197 0.256 0.343 0.667
VRRATIO 3,446 0.416 0.535 0.000 0.000 0.205 0.612 2.227
DISPERSION 3,446 0.319 0.265 0.002 0.080 0.251 0.519 0.958
Panel D: Second largest shareholder (family firms)
FAMILY_2 1,387 0.670 0.470 0.000 0.000 1.000 1.000 1.000
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

STATE_2 1,387 0.024 0.155 0.000 0.000 0.000 0.000 1.000


CORPORATE_2 1,387 0.067 0.250 0.000 0.000 0.000 0.000 1.000
FINANCIAL_2 1,387 0.239 0.426 0.000 0.000 0.000 0.000 1.000
VR2_FAMILY 1,387 0.133 0.114 0.000 0.000 0.139 0.207 0.404
VR2_STATE 1,387 0.006 0.037 0.000 0.000 0.000 0.000 0.256
VR2_CORPORATE 1,387 0.012 0.049 0.000 0.000 0.000 0.000 0.290
VR2_FINANCIAL 1,387 0.041 0.081 0.000 0.000 0.000 0.000 0.301
Multiple Large Shareholders and Corporate Risk-taking 713

rights, which suggests that in most cases the LCS’s power should be hard to contest. The
LCS’s sizeable proportion of all voting rights stems in part from the control-enhancing
mechanisms that family owners often put in place. As a matter of fact, the mean excess
control of the LCS is about 20.9%. Laeven and Levine (2008) report a similar figure for
the French firms in their sample. This large wedge suggests that French family firms are
susceptible to significant agency conflicts between the LCS and minority shareholders.
Panel D provide some details on the second largest shareholder in French family firms
with MLS. In about two-thirds of the cases, the second largest shareholder is another
family that holds on average 19.85% (¼ 0.133/0.670) of the votes. Widely held financial
institutions are the second largest shareholder in 23.9% of the cases and hold on average
17.2% (¼ 0.041/0.239) of the votes.
Table 3 displays the correlation between the explanatory variables for the full sample.
Unsurprisingly, the number of MLS is positively correlated with their voting power.
These two variables are obviously associated with a lower Herfindahl index. Nonetheless,
the presence and voting power of MLS are positively correlated with the LCS’s excess
control, which might suggest that MLS structures arise to restrain the LCS’s control over
the firm’s decisions. This appears to be the case of younger, smaller and more focused
firms, which exhibit a higher proportion of MLS. Finally, it is useful to note that the
correlation between the independent variables is not particularly high, implying that the
multivariate regressions are unlikely to involve multicollinearity problems.

4. Empirical Results

We start by analysing the influence of the LCS on risk-taking. The LCS’s propensity for
selecting a suboptimal level of risk, particularly in family firms, opens the opportunity
for MLS to correct that inefficiency. We thus focus on family firms and test the influence
of MLS. Our analysis is complemented by a number of sensitivity checks and specific
control for self-selection. Finally, we examine the role of the second largest
shareholder’s identity and the role that MLS play depending on the firm’s internal
governance structures.

4.1. Influence of an LCS on risk-taking


The starting point of our analysis is to establish the LCS’s negative impact on risk-taking.
Table 4 presents the regressions of the different risk measures on the dummy variables
indicating the presence of an LCS and the LCS’s identity. Consistent with H1, columns
(1), (4) and (7) indicate that the presence of an LCS is associated with lower variability in
firm performance. For example, the absolute deviation of the ROAs is about 1.1% lower
in the presence of an LCS. This reduction in risk is substantial, considering that the
average absolute deviation of the ROAs for the whole sample is about 5.6%. The results
confirm that firms with an LCS have a tendency to make more conservative investments.
Many reasons could explain the LCS’s inclination to reduce risk: high risk aversion
due to under-diversified wealth, concerns over losing control, and the preservation of
private benefits. To gain further insight, we break down the LCS dummy to distinguish
four types of large shareholders: family owners, the State, widely held firms and financial
institutions. In columns (2), (5) and (8) of Table 4, the regressions only contain a dummy
for family control to highlight its effect relative to all the other types of controlling
shareholders or the absence of an LCS. The coefficients on the family dummy are found
© 2016 John Wiley & Sons, Ltd.
Table 3
714

Pairwise correlations between independent variables

This table reports Pearson (below the diagonal) and Spearman (above the diagonal) correlation between the main variables used in the analysis. The sample consists
of 4,501 firm–year observations representing 602 French publicly listed firms over the period 20032012. MLSD equals 1 if the firm has at least two large
shareholders; and 0 otherwise. MLSN is the number of large shareholders, other than the LCS, up to the fourth. VR234 is the sum of voting rights of the second, third
and fourth largest shareholders. VRRATIO is the sum of voting rights of the second, third and fourth largest blockholders divided by the voting rights of the LCS.

© 2016 John Wiley & Sons, Ltd.


DISPERSION equals the sum of squared differences between the voting rights of the four largest shareholders, that is, (VR1  VR2)2 þ (VR2  VR3)2 þ (VR3 
VR4)2; where VR1, VR2, VR3 and VR4 equal the voting rights of the first, second, third and fourth largest shareholders. EXCESS CONTROL is the excess control of
the LCS, defined as the difference between the LCS’s ultimate control rights and ultimate cash-flow rights, all divided by her ultimate control rights. SIZE is the
natural logarithm of total assets (in thousands of euro). CAPEX is capital expenditures over sales. AGE is the number of years since the firm’s first date of
incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the number of business segments. All continuous variables are winsorised at the 1st
and 99th percentiles.  ,  and  denote statistical significance at the 1%, 5% and 10% level, respectively.

1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11.
        
1. MLSD 1.000 0.980 0.787 0.739 0.378 0.173 0.194 0.060 0.187 0.085 0.109
2. MLSN 0.885 1.000 0.801 0.751 0.396 0.171 0.209 0.061 0.191 0.083 0.111

3. VR234 0.795 0.819 1.000 0.907 0.568 0.135 0.169 0.051 0.246 0.073 0.106

4. VRRATIO 0.644 0.713 0.813 1.000 0.538 0.234 0.177 0.050 0.272 0.074 0.125
5. DISPERSION 0.467 0.450 0.596 0.562 1.000 0.031 0.046 0.038 0.239 0.016 0.054
6. EXCESS CONTROL 0.191 0.186 0.153 0.217 0.121 1.000 0.019 0.074 0.023 0.010 0.002
7. SIZE 0.172 0.195 0.172 0.145 0.054 0.040 1.000 0.286 0.386 0.333 0.271

8. CAPEX 0.038 0.031 0.033 0.044 0.053 0.062 0.062 1.000 0.174 0.213 0.036

9. AGE 0.143 0.148 0.188 0.224 0.169 0.003 0.322 0.006 1.000 0.188 0.181
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

10. LEVERAGE 0.059 0.045 0.057 0.057 0.017 0.001 0.248 0.145 0.127 1.000 0.106
11. DIVERSIFICATION 0.120 0.108 0.116 0.139 0.067 0.020 0.236 0.059 0.151 0.093 1.000
Table 4
Influence of the existence of a largest controlling shareholder on corporate risk-taking

This table presents the regressions of corporate risk-taking on the presence and identity of the largest controlling shareholder (LCS) using the full sample. The
dependent variables are the absolute deviation from expected ROA (Columns (1)–(3)), the absolute deviation from expected Tobin’s Q (Columns (4)–(6)) and the
absolute deviation from predicted stock returns using the market model (Columns (7)–(9)). LCS equals 1 if the firm has a largest controlling shareholder and 0
otherwise. LCS_FAMILY, LCS_STATE, LCS_CORPORATE and LCS_FINANCIAL equal 1 if the ultimate owner is a family, the State, a widely held corporation or

© 2016 John Wiley & Sons, Ltd.


a widely held financial institution, respectively, and 0 otherwise. SIZE is the natural logarithm of total assets (in thousands of euro). CAPEX is capital expenditures
over sales. AGE is the number of years since the firm’s first date of incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the number of
business segments. All continuous variables are winsorised at the 1st and 99th percentiles. The t-statistics in parentheses are computed with standard errors
clustered by firm.  ,  and  denote statistical significance at the 1%, 5% and 10% level, respectively.

Absolute deviation from expected ROA Absolute deviation from expected Tobin’s Q Absolute deviation from expected stock return

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9)
LCS 1.07839 0.05376 0.00553
(2.40) (1.83) (1.67)
LCS_FAMILY 1.65214 1.59527 0.04761 0.06234 0.00738 0.00729
(4.15) (3.42) (2.43) (2.09) (3.08) (2.13)
LCS_STATE 0.36015 0.06191 0.00114
(0.48) (1.51) (0.23)
LCS_CORPORATE 0.08780 0.01360 0.00018
(0.09) (0.30) (0.03)
LCS_FINANCIAL 0.13399 0.01989 0.00029
(0.16) (0.52) (0.06)
SIZE 0.78568 0.84697 0.85619 0.02800 0.02826 0.02791 0.00703 0.00727 0.00730
Multiple Large Shareholders and Corporate Risk-taking

(9.36) (9.81) (9.65) (6.35) (6.16) (5.96) (12.68) (13.49) (12.97)


CAPEX 0.05061 0.04579 0.04551 0.00494 0.00444 0.00450 0.00002 0.00001 0.00001
(2.96) (2.72) (2.68) (2.79) (2.48) (2.51) (0.29) (0.11) (0.10)
AGE 0.01716 0.01657 0.01736 0.00042 0.00039 0.00041 0.00009 0.00009 0.00009
(3.04) (2.96) (3.04) (1.40) (1.28) (1.36) (2.33) (2.20) (2.17)
LEVERAGE 2.63153 2.97021 2.99147 0.10179 0.09718 0.09873 0.04958 0.05022 0.05034
715
Table 4
716

Continued

Absolute deviation from expected ROA Absolute deviation from expected Tobin’s Q Absolute deviation from expected stock return

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9)
(2.29) (2.58) (2.60) (1.49) (1.41) (1.42) (6.83) (6.93) (6.95)

© 2016 John Wiley & Sons, Ltd.


DIVERSIFICATION 0.30077 0.25249 0.24733 0.01109 0.01009 0.01009 0.00081 0.00062 0.00061
(2.93) (2.40) (2.34) (2.14) (1.93) (1.94) (1.19) (0.91) (0.90)
Intercept 16.81864 17.47002 17.41767 0.75832 0.72955 0.73636 0.16772 0.16907 0.16904
(10.76) (12.04) (11.41) (9.57) (9.94) (9.78) (14.78) (16.07) (14.78)
Year/month Yes Yes Yes Yes Yes Yes Yes Yes Yes
dummies
Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes
Number of 4,501 4,501 4,501 4,501 4,501 4,501 49,850 49,850 49,850
observations
Adjusted R2 0.135 0.145 0.146 0.129 0.130 0.130 0.040 0.041 0.041
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi
Multiple Large Shareholders and Corporate Risk-taking 717

to be negative and highly significant. In addition, the effect appears to be stronger


compared to the results with the LCS dummy. For instance, the reduction in the absolute
deviation of the ROAs is about 1.65% compared to a reduction of approximately 1.1% in
the presence of an LCS. These results indicate that family-controlled firms are
particularly prone to cut risk relative to other firms.
In columns (3), (6) and (9) of Table 4, the four dummies representing the different types
of LCS are included together in the regression to highlight their effect relative to the case in
which the firm has no LCS. Interestingly, only the variable representing family ownership
loads with a significantly negative coefficient. For the other types of LCS, the coefficient
tends to be negative but is not statistically significant. These results appear to make sense.
State-owned firms have no specific reasons to cut back risk. In fact, some of these firms
have embarked on high-risk ventures on the back of the French State’s financial support. In
firms where the LCS is a widely held firm or a widely held financial institution, the
incentives for cutting back risk appear to be similarly weak. In particular, widely held
financial institutions are usually well diversified and consequently have no reason to push
the firm to reduce risk below the optimal level consistent with value maximisation.
Overall, the results suggest that the incentives to decrease firm risk apply primarily to
family owners, which is consistent with H2. Other types of LCSs do not seem to have the
same motivations and, accordingly, do not reduce the level of risk-taking in firms under
their control. The reason that LCSs are found to be associated with lower firm risk is
because family owners represent the overwhelming majority of LCSs.
The control variables have generally the effect predicted in theory and verified in
practice. Consistent with the work of Adams et al. (2005), larger firms are characterised
by significantly lower deviations from expected performance, regardless of the way
performance is measured. Older firms also tend to be associated with more stable
performance. On the other hand, firms with higher greater growth opportunities exhibit
less stable performance, which can be explained by the high level of uncertainty due to
the implementation of their investments. Diversification is found to have a negative
impact on the variability of operating returns but not on the variability of stock returns.
Finally, the results show that leverage is associated with a higher variability of operating
and stock returns.
Since risk reduction occurs primarily in family firms, we focus on these to investigate
whether excess control rights by the LCS exacerbate his/her incentives to decrease risk.
The regression results are presented in Table 5. For each performance measure, the larger
the wedge between the control and cash-flow rights of the family owners (EXCESS
CONTROL), the lower the level of risk-taking. In line with H3, a larger wedge should
make the extraction of private benefits more valuable to the LCS and their potential loss
all the more undesirable. It follows that the LCS should have a strong incentive to reduce
risk-taking. Furthermore, the wedge may signal that the family is seeking to maintain or
increase its control while being unable to put up the required funds.9 Family owners with
limited funds should, accordingly, prefer to limit the firm’s risk-taking activities.

9
Almeida and Wolfenzon (2006) explain that family firms create pyramidal structures
because their cash-flows and/or assets are difficult to pledge to outside investors. The fact that
they need to tap outside investors probably signals the family’s poor diversification and/or
limited capital. In other words, the wedge should be associated with the family’s inability to
fund the firm’s new projects.

© 2016 John Wiley & Sons, Ltd.


718 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Table 5
Effect of the LCS’s excess control on corporate risk-taking in family firms

This table presents the regressions of corporate risk-taking on the excess control of LCS using the
subsample of family firms (the LCS thus represents the controlling family). The dependent variables are
the absolute deviation from expected ROA (Column (1)), the absolute deviation from expected Tobin’s Q
(Column (2)) and the absolute deviation from predicted stock returns using the market model (Column
(3)). EXCESS CONTROL is the excess control of the LCS, defined as the difference between the LCS’s
ultimate control rights and ultimate cash-flow rights, all divided by her ultimate control rights. SIZE is the
natural logarithm of total assets (in thousands of euros). CAPEX is capital expenditures over sales. AGE is
the number of years since the firm’s first date of incorporation. LEVERAGE is total debt over total assets.
DIVERSIFICATION is the number of business segments. All continuous variables are winsorised at the
1st and 99th percentiles. The t-statistics in parentheses are computed with standard errors clustered by
firm.  ,  and  denote statistical significance at the 1%, 5% and 10% level, respectively.

Absolute deviation Absolute deviation Absolute deviation from


from expected ROA from expected Tobin’s Q expected stock return

Variable (1) (2) (3)


 
EXCESS 1.44774 0.08511 0.00766
CONTROL
(2.01) (2.08) (1.72)
SIZE 0.69703 0.02424 0.00643
(7.21) (4.38) (11.22)
CAPEX 0.02691 0.00507 0.00005
(1.72) (2.25) (0.63)
AGE 0.01908 0.00027 0.00010
(3.44) (0.82) (2.51)
LEVERAGE 3.72855 0.04928 0.04587
(3.01) (0.62) (6.67)
DIVERSIFICATION 0.19631 0.00772 0.00109
(1.86) (1.40) (1.70)
Intercept 12.71830 0.55700 0.13668
(9.52) (6.94) (17.51)
Year/month Yes Yes Yes
dummies
Industry dummies Yes Yes Yes
Number of 3,446 3,446 38,243
observations
Adjusted R2 0.140 0.116 0.046

4.2. Influence of MLS on risk-taking


To investigate the role of MLS, we focus on family firms because of their prevalence
among firms with an LCS and because the results from the previous section clearly
indicate that only family owners are associated with lower risk-taking. Before turning to
multivariate regressions, we briefly present some univariate results. In Table 6, we
compare the absolute deviation from the firm’s expected performance (Panels A–C) and

© 2016 John Wiley & Sons, Ltd.


Table 6
Univariate analysis

This table compares the mean and median of different risk-taking proxies by the degree of contestability of the LCS (controlling family). MLSD equals 1 if the firm has
at least two large shareholders; and 0 otherwise. MLSN is the number of large shareholders, other than the LCS, up to the fourth. VR234 is the sum of voting rights of the
second, third and fourth largest shareholders. VRRATIO is the sum of voting rights of the second, third and fourth largest blockholders divided by the voting rights of the
LCS. DISPERSION equals the sum of squared differences between the voting rights of the four largest shareholders, that is, (VR1  VR2)2 þ (VR2  VR3)2 þ

© 2016 John Wiley & Sons, Ltd.


(VR3  VR4)2; where VR1, VR2, VR3 and VR4 equal the voting rights of the first, second, third and fourth largest shareholders. The risk-taking proxies are the absolute
deviation from expected ROA (Panel A), the absolute deviation from expected Tobin’s Q (Panel B) the absolute deviation from predicted stock returns using the
market model (Panel C), the standard deviation of industry-adjusted ROA (Panel D), the standard deviation of industry-adjusted Tobin’s Q (Panel E), and the standard
deviation of market-adjusted monthly stock return (Panel F).  and  denote statistical significance at the 1% and 5% level, respectively.

Mean Median

t-stat z-stat
Low High (Low–High) Low High (Low–High)
Panel A: Absolute deviation from expected ROA
MLSD 4.622 6.160 7.525 3.082 3.933 6.937
MLSN 4.615 6.161 7.560 3.071 3.952 7.041
VR234 4.672 5.981 6.465 2.993 3.830 6.467
VRRATIO 4.675 6.304 7.712 2.958 3.846 6.848
DISPERSION 5.878 4.426 7.210 3.770 3.106 5.491
Panel B: Absolute deviation from expected Tobin’s Q
MLSD 0.291 0.356 6.874 0.225 0.294 7.133
MLSN 0.294 0.354 6.369 0.232 0.290 6.054
Multiple Large Shareholders and Corporate Risk-taking

VR234 0.297 0.344 5.060 0.225 0.283 6.084


VRRATIO 0.299 0.354 5.662 0.221 0.284 6.013
DISPERSION 0.345 0.281 6.937 0.288 0.224 7.013
Panel C: Absolute deviation from expected stock return
MLSD 0.072 0.080 9.520 0.050 0.054 7.515
MLSN 0.072 0.080 9.520 0.050 0.054 7.515
719
Table 6
720

Continued

Mean Median

t-stat z-stat
Low High (Low–High) Low High (Low–High)

© 2016 John Wiley & Sons, Ltd.


Panel C: Absolute deviation from expected stock return
VR234 0.072 0.079 9.192 0.049 0.053 7.983
VRRATIO 0.072 0.081 9.900 0.050 0.053 6.522
DISPERSION 0.078 0.072 6.138 0.053 0.049 6.581
Panel D Standard deviation of industry-adjusted ROA
MLSD 4.298 6.172 3.355 2.814 4.030 4.032
MLSN 4.359 6.192 3.255 2.808 4.059 4.119
VR234 4.450 6.031 2.804 2.934 3.882 3.099
VRRATIO 4.230 6.859 4.547 2.855 3.979 3.769
DISPERSION 5.811 4.274 2.742 3.948 2.929 2.880
Panel E: Standard deviation of industry-adjusted Tobin’s Q
MLSD 0.232 0.291 2.822 0.166 0.221 2.881
MLSN 0.233 0.293 2.857 0.165 0.222 3.016
VR234 0.238 0.284 2.159 0.171 0.212 2.932
VRRATIO 0.231 0.309 3.570 0.174 0.220 2.739
DISPERSION 0.286 0.224 2.990 0.221 0.174 2.519
Panel F: Standard deviation of market-adjusted monthly stock return
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

MLSD 0.104 0.120 3.256 0.091 0.117 3.964


MLSN 0.105 0.120 2.908 0.091 0.117 3.991
VR234 0.107 0.117 1.967 0.094 0.108 2.544
VRRATIO 0.106 0.121 3.048 0.093 0.110 2.741
DISPERSION 0.116 0.105 2.233 0.108 0.094 2.152
Multiple Large Shareholders and Corporate Risk-taking 721

the standard deviation of industry-adjusted performance (Panels D–F) across different


ownership structures.10 The results clearly indicate that firms with MLS are associated
with higher risk-taking. For instance, Panel A shows that the mean absolute deviation
from expected ROAs is about 4.6% when there are no other blockholders but increases to
about 6.2% when other blockholders are present. Interestingly, Panel D shows that the
standard deviation of the industry-adjusted ROAs is also about 6.2% when other
blockholders are present, but only about 4.3% when there are no other blockholders.
These results appear to support H4. It is also useful to point out that, in Panels A–C, a
measure of risk is available for each firm in each year, while, in Panels D–F, only one
measure of risk is available for each firm over the whole sample period. This finding
explains that the tests of difference, which are not adjusted for persistence in firm
performance, are more significant using the absolute deviation of performance than the
standard deviation of performance.11

4.2.1. Main results. The main regressions are presented in Table 7, with the results for
each indicator of risk-taking in a separate panel. The results in Panel A show that the
absolute deviation of the ROA relative to its expected value is about 1% higher and
significant at the 1% level when firms have more than one blockholder (Column (1)). The
magnitude of the coefficient indicates that the presence of MLS is able to largely
neutralise the risk reduction associated with family control. The other regressions
confirm the role of MLS in promoting corporate risk-taking. The number of blockholders
besides the LCS (Column (2)) has a positive effect on the absolute deviation of the ROA.
Similarly, the cumulative votes of the other blockholders (up to the fourth largest) and
their relative power (Columns (3) and (4)) are associated with greater variability in the
firm’s operating performance. In contrast, when the concentration of votes is relatively
high (indicating strong control by family owners), the level of risk-taking measured at the
firm’s operational level appears to be significantly lower.
Panel B of Table 7 displays the regression results using Tobin’s Q as the performance
indicator. The coefficients of the MLS variables exhibit similar levels of statistical
significance and have the anticipated signs. Column (1) shows that the presence of MLS
is associated with higher risk-taking. The positive difference of about 4.76% appears to
partially offset the reduction in risk associated with family control (indicated in
Column (5) of Table (4)). Consistent with H3, greater control by the LCS relative to
his/her ownership is associated with a significant reduction in risk-taking. Thus, the
lower variability in operating performance apparent in the previous panel is confirmed by
lower variability in the firm’s market valuation. This suggests that investors are sensibly
factoring in the incentives for the LCS to decrease the firm’s risk profile as well as the
actual reduction in the firm’s earnings variability. However, the presence of other
blockholders besides the LCS contributes to increasing the variability in the firm’s
market value. This is consistent with H4, that MLS encourage firms to take more risk.
The other results confirm the positive influence of MLS on corporate risk-taking. A
larger number of other blockholders (Column (2)) and a higher percentage of voting

10
The results remain qualitatively unchanged when we use the standard deviation of the
unadjusted ROA, Tobin’s Q and stock return in the univariate analysis.
11
This issue is dealt with in the multivariate regressions by using standard errors corrected for
clustering at the firm level.

© 2016 John Wiley & Sons, Ltd.


722 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Table 7
Influence of multiple large shareholders on corporate risk-taking in family firms: Main evidence

This table examines the influence of MLS on corporate risk-taking in family firms. The dependent
variables are the absolute deviation from expected ROA (Panel A), the absolute deviation from
expected Tobin’s Q (Panel B) and the absolute deviation from expected returns using the market model
(Panel C). MLSD equals 1 if the firm has at least two large shareholders, 0 otherwise. MLSN is the
number of large shareholders, other than the LCS, up to the fourth. VR234 is the sum of voting rights of
the second, third and fourth largest shareholders. VRRATIO is the sum of voting rights of the second,
third and fourth largest blockholders divided by the voting rights of the LCS. DISPERSION equals the
sum of squared differences between the voting rights of the four largest shareholders, that is,
(VR1  VR2)2 þ (VR2 – VR3)2 þ (VR3  VR4)2; where VR1, VR2, VR3 and VR4 equal the voting rights
of the first, second, third and fourth largest shareholders. EXCESS CONTROL is the excess control of
the LCS, defined as the difference between the LCS’s ultimate control rights and ultimate cash-flow
rights, all divided by her ultimate control rights. SIZE is the natural logarithm of total assets (in
thousands of euro). CAPEX is capital expenditures over sales. AGE is the number of years since the
firm’s first date of incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the
number of business segments. All continuous variables are winsorised at the 1st and 99th percentiles.
The t-statistics in parentheses are computed with standard errors clustered by firm.  ,  and  denote
statistical significance at the 1%, 5% and 10% level, respectively.

Variable (1) (2) (3) (4) (5)


Panel A: Corporate risk-taking is measured by the deviation from expected ROA
MLSD 0.99439
(3.18)
MLSN 0.48166
(2.22)
VR234 2.79391
(2.46)
VRRATIO 0.83154
(2.76)
DISPERSION 2.20605
(3.32)
EXCESS CONTROL 1.84639 1.74376 1.67106 1.82397 2.29414
(2.51) (2.36) (2.30) (2.45) (3.03)
SIZE 0.66520 0.66956 0.67649 0.67953 0.69404
(6.78) (6.74) (6.90) (6.95) (7.10)
CAPEX 0.02542 0.02639 0.02617 0.02572 0.02298
(1.62) (1.68) (1.66) (1.65) (1.47)
AGE 0.01786 0.01807 0.01734 0.01605 0.01614
(3.25) (3.29) (3.20) (2.98) (2.99)
LEVERAGE 3.80054 3.73426 3.75970 3.81945 3.53874
(3.06) (2.99) (3.02) (3.07) (2.81)
DIVERSIFICATION 0.17258 0.18512 0.18350 0.15821 0.17670
(1.67) (1.78) (1.75) (1.51) (1.68)
Intercept 11.77036 12.06977 12.06698 12.04071 13.75587
(8.56) (8.70) (8.86) (8.90) (9.71)
Year dummies Yes Yes Yes Yes Yes

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 723

Table 7
Continued

Variable (1) (2) (3) (4) (5)


Industry dummies Yes Yes Yes Yes Yes
Number of 3,446 3,446 3,446 3,446 3,446
observations
Adjusted R2 0.146 0.142 0.142 0.141 0.145
Panel B: Corporate risk-taking is measured by the deviation from expected Tobin’s Q
MLSD 0.04758
(2.90)
MLSN 0.02386
(2.13)
VR234 0.12304
(1.92)
VRRATIO 0.03802
(1.99)
DISPERSION 0.08228
(2.43)
EXCESS CONTROL 0.10332 0.09535 0.09700 0.10279 0.10846
(2.49) (2.28) (2.37) (2.45) (2.54)
SIZE 0.02239 0.02294 0.02321 0.02380 0.02417
(4.04) (4.05) (4.17) (4.31) (4.36)
CAPEX 0.00452 0.00490 0.00481 0.00470 0.00468
(2.01) (2.17) (2.13) (2.09) (2.10)
AGE 0.00021 0.00023 0.00024 0.00017 0.00022
(0.65) (0.70) (0.72) (0.51) (0.67)
LEVERAGE 0.03537 0.04070 0.03621 0.03999 0.04609
(0.44) (0.50) (0.45) (0.50) (0.57)
DIVERSIFICATION 0.00580 0.00657 0.00642 0.00630 0.00622
(1.06) (1.21) (1.16) (1.14) (1.14)
Intercept 0.47678 0.50966 0.50976 0.51997 0.55431
(6.01) (6.28) (6.35) (6.51) (7.18)
Year dummies Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes
Number of 3,446 3,446 3,446 3,446 3,446
observations
Adjusted R2 0.119 0.117 0.114 0.117 0.118
Panel C: Corporate risk-taking is measured by the deviation from expected stock returns
MLSD 0.00469
(2.22)
MLSN 0.00497
(3.22)
VR234 0.02276
(2.70)
VRRATIO 0.00662
(3.27)

© 2016 John Wiley & Sons, Ltd.


724 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Table 7
Continued

Variable (1) (2) (3) (4) (5)


Panel C: Corporate risk-taking is measured by the deviation from expected stock returns
DISPERSION 0.01021
(2.10)
EXCESS CONTROL 0.01262 0.01362 0.01288 0.01426 0.01416
(2.51) (2.69) (2.58) (2.82) (2.72)
SIZE 0.00621 0.00607 0.00618 0.00627 0.00638
(9.97) (9.69) (10.05) (10.44) (10.56)
CAPEX 0.00006 0.00006 0.00006 0.00007 0.00007
(0.70) (0.74) (0.76) (0.80) (0.82)
AGE 0.00010 0.00009 0.00009 0.00008 0.00009
(2.26) (2.18) (2.12) (1.93) (2.18)
LEVERAGE 0.04732 0.04684 0.04729 0.04759 0.04705
(6.54) (6.51) (6.62) (6.68) (6.52)
DIVERSIFICATION 0.00121 0.00121 0.00125 0.00112 0.00125
(1.84) (1.85) (1.90) (1.72) (1.91)
Intercept 0.13334 0.13152 0.13280 0.13388 0.14172
(14.79) (14.68) (14.90) (15.56) (17.17)
Month dummies Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes
Number of 38,243 38,243 38,243 38,243 38,243
observations
Adjusted R2 0.046 0.047 0.047 0.047 0.046

rights in their hands (Column (3)), especially relative to the voting rights of the family
(Column (4)), are associated with greater variability in the firm’s market valuation. In
contrast, the lack of power of MLS to contest the family’s control over the firm’s policy
(Column (5)) results in lower variability in the firm’s valuation, indicating that family
owners are successful in reducing firm risk.
Panel C of Table 7 examines the absolute deviation of stock returns relative to the
market’s realised return and the firm’s fitted beta. The results are consistent with those
based on the other performance measures. The absence of other blockholders appears to
allow family owners to decrease firm risk. The incentive to decrease risk is positively
related to the divergence between the control and cash-flow rights of the family owners.
Again, the presence, number and power of MLS are associated with significantly higher
deviations from expected stock returns. This result indicates that stock returns are less
predictable and supports the assumption that the presence of MLS prevents firms from
reducing their risk-taking. Overall, the economic significance of MLS in monitoring the
LCS (family owners) and promoting better governance of the firm is clearly
demonstrated.

4.2.2. Sensitivity tests. We test the robustness of the main results by running several
sensitivity checks. First, we use two alternative proxies for the LCS’s relative power (or
lack of it). One is based on principal component analysis (PCA). The purpose is to
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 725

generate a linear combination of the five MLS variables previously defined that better
captures the general influence of MLS. In our case, PCA generates one factor with an
eigenvalue greater than one that enters positively and significantly in all the risk
regressions. Another proxy for the LCS’s relative power is the Shapley value. This
variable is defined as the solution for the LCS in a four-shareholder voting game where
the four largest blockholders are individual players and the rest are considered an
‘ocean’. The higher this variable, the higher the LCS’s relative power. The results show
that the Shapley value enters negatively and significantly in all the risk regressions. This
finding confirms that the weaker contestability of the LCS’s power is associated with
lower risk-taking. We omit the tabulation of these results to save space.
Second, we use a number of alternative risk measures. Following Cheng (2008), we
compute the standard deviation of the ROA, Tobin’s Q and stock returns over the sample
period and relate these risk measures to the MLS variables and firm characteristics
averaged over the sample period. This approach yields similar results to those obtained
with Glejser heteroskedasticity tests, which corroborates our main conclusions.12 We
omit the tabulation of these results to focus on four different risk measures. Consistent
with Mishra (2011), we calculate the standard deviation of ROAs over the next 5 years.
Faccio et al. (2011) use the difference between the maximum and minimum ROAs,
which we also calculate over the next 5 years. Finally, we use the firm’s R&D intensity
and its volatility. Kothari et al. (2002) and Cheng (2008) argue that R&D expenses
correspond to investments in high-risk projects. As a matter of fact, the rationale for
expensing R&D investments rather than their capitalisation is to reflect their highly
uncertain pay-off. Coles et al. (2006) associate increases in R&D expenditures with
managerial incentives to take risk. The results based on these risk measures are presented
in Table 8. In all the regressions, the coefficient of the MLS dummy is positive and
significant, particularly for the level and volatility of R&D expenditures. Excess control
by the LCS is also confirmed to play a negative role in the firm’s risk-taking behaviour.
Third, we include a number of additional firm-level variables to mitigate potential
concerns relating to omitted variable bias. The results are displayed in Table 9 for the
absolute deviation of the ROA. In Column (1), we add the LCS’s ultimate cash-flow
rights (UCF). In line with Mishra (2011), the coefficient of UCF is positive and
significant at the 10% level. This result suggests that greater alignment of interests and
lower incentives to extract private benefits encourage firms to implement riskier
strategies to create greater value. More importantly, the coefficient of MLSD remains
materially unchanged. In Column (2), we add an indicator for group affiliation
(GROUP). Paligorova (2010) argues that group affiliation may induce greater risk-
taking, since firms benefit from implicit risk-sharing arrangements. Consistent with the
author’s argument, the coefficient of GROUP is found to be positive and significant (at
the 1% level). However, the coefficient of MLSD remains materially unchanged. We also

12
In line with Cheng (2008), we reproduce the results of the cross-sectional regressions using
the industry-adjusted ROA (Tobin’s Q) and market-adjusted stock returns as dependent
variables. The industry-adjusted ROA (Tobin’s Q) is the difference between the firm’s ROA
(Tobin’s Q) and the industry ROA (Tobin’s Q) in the same year. The latter is defined as the
median ROA (Tobin’s Q) of all firms with the same two-digit SIC code. The market-adjusted
stock return is the difference between the firm’s monthly stock return and the return on the
SBF 250 index. The results remain qualitatively unchanged.

© 2016 John Wiley & Sons, Ltd.


726 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Table 8
Regressions with alternative risk-taking measures

This table evaluates the influence of MLS on corporate risk-taking in family firms using four alternative
proxies. ROA volatility is measured by the standard deviation of ROA in the next 5-year period. To
include a firm in our sample, we require at least five observations of ROA volatility over the period
20032012. ROAMax  ROAMin is the difference between the maximum and minimum ROA over the
next 5-year period. R&D intensity is the ratio of R&D expenditures to sales. R&D volatility is measured
by the standard deviation of R&D intensity over the next 5-year period. MLSD equals 1 if the firm has at
least two large shareholders, 0 otherwise. EXCESS CONTROL is the excess control of the LCS, defined
as the difference between the LCS’s ultimate control rights and ultimate cash-flow rights, all divided by
her ultimate control rights. SIZE is the natural logarithm of total assets (in thousands of euro). CAPEX is
capital expenditures over sales. AGE is the number of years since the firm’s first date of incorporation.
LEVERAGE is total debt over total assets. DIVERSIFICATION is the number of business segments. All
continuous variables are winsorised at the 1st and 99th percentiles. The t-statistics in parentheses are
computed with standard errors clustered by firm.  ,  and  denote statistical significance at the 1%,
5% and 10% level, respectively.

ROA volatility ROAMax  ROAMin R&D R&D volatility


in next 5 years in next 5 years intensity in next 5 years

Variable (1) (2) (3) (4)


  
MLSD 0.90086 2.19105 0.02364 0.00897
(1.97) (1.90) (5.02) (3.16)
EXCESS CONTROL 2.53955 6.55254 0.03177 0.01509
(2.24) (2.30) (2.01) (2.32)
SIZE 0.71656 1.77231 0.00326 0.00274
(4.90) (4.80) (2.21) (3.38)
CAPEX 0.04875 0.12805 0.00119 0.00009
(1.95) (1.95) (2.32) (0.71)
AGE 0.00647 0.01487 0.00018 0.00003
(0.83) (0.76) (2.85) (0.55)
LEVERAGE 3.68559 9.02784 0.03090 0.00439
(2.31) (2.24) (2.23) (0.42)
DIVERSIFICATION 0.03330 0.07255 0.00161 0.00057
(0.24) (0.21) (1.20) (0.60)
Intercept 19.91603 50.86130 0.06600 0.12534
(17.96) (18.40) (3.13) (18.56)
Year dummies Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes
Number of 1,959 1,959 3,446 1,959
observations
Adjusted R2 0.158 0.156 0.132 0.131

add the lagged value of ROA under the premise that high-performing firms are likely to
be high risk takers. By clustering around high performers, MLS might thus simply pick
up the effect of lagged firm performance instead of actually inducing higher risk-taking.
Column (3) shows that the coefficient of lagged ROA is positive and highly significant.
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 727

Table 9
Risk-taking regressions with additional control variables

This table presents regression results on the effect of MLS on risk-taking. The dependent variable is the
absolute deviation from expected ROA. MLSD equals 1 if the firm has at least two large shareholders, 0
otherwise. UCF is the ultimate cash-flow rights of the LCS, defined as the sum of products of direct
cash-flow rights along the different ownership chains. GROUP is a dummy variable that equals 1 if the
LCS is affiliated with a business group. EM is a measure of the extent of earnings management,
calculated as in Leuz et al. (2003). EXCESS CONTROL is the excess control (at the 10% threshold) of
the LCS, defined as the difference between the LCS’s ultimate control rights and ultimate cash-flow
rights, all divided by her ultimate control rights. SIZE is the natural logarithm of total assets (in
thousands of euros). CAPEX is capital expenditures over sales. AGE is the number of years since the
firm’s first date of incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the
number of business segments. All continuous variables are winsorised at the 1st and 99th percentiles.
The t-statistics in parentheses are computed with standard errors clustered by firm.  ,  and  denote
statistical significance at the 1%, 5% and 10% level, respectively.

Variable (1) (2) (3) (4)


  
MLSD 1.07737 1.08335 0.87650 0.97568
(3.46) (3.52) (3.12) (3.09)
UCF 0.01119
(1.67)
GROUP 0.01030
(3.84)
ROAt–1 0.96673
(3.16)
EM 0.00486
(0.76)
EXCESS CONTROL 1.35495 3.86315 1.67064 1.83679
(1.66) (4.02) (2.47) (2.50)
SIZE 0.65970 0.66593 0.64201 0.65822
(6.96) (6.89) (7.44) (6.79)
CAPEX 0.02479 0.02681 0.02488 0.02564
(1.57) (1.77) (1.70) (1.64)
AGE 0.01844 0.01802 0.01549 0.01803
(3.48) (3.40) (3.15) (3.29)
LEVERAGE 3.79022 3.80966 5.35853 3.80513
(3.08) (3.08) (4.99) (3.09)
DIVERSIFICATION 0.15794 0.19199 0.11617 0.16782
(1.56) (1.92) (1.25) (1.64)
Intercept 11.24419 11.82180 10.29290 18.81776
(7.89) (8.63) (8.75) (14.05)
Year dummies Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes
Number of observations 3,446 3,446 3,446 3,446
Adjusted R2 0.145 0.155 0.147 0.146

© 2016 John Wiley & Sons, Ltd.


728 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

However, this does not affect the positive effect on risk associated with the presence of
MLS. Finally, we add a variable measuring the extent of the firm’s earnings management
(EM). This variable is calculated according to Leuz et al. (2003). Firms that extensively
manage their earnings may appear to take on less risk since their performance tends to be
more predictable. MLS are expected to increase monitoring and thus deter earnings
management. Accordingly, the regression may suggest an increase in firm risk from the
presence of MLS that is actually attributable to a decrease in earnings management (also
caused by their presence). Column (4) shows that the EM variable is insignificant. In
contrast, Mishra (2011) finds that earnings management decreases the volatility of cash-
flows. This could be because Asian firms manage their earnings more aggressively. In
our case, the effect of MLS on firm risk-taking does not appear to arise from a decrease in
earnings management that they may or may not cause.
4.2.3. Control for endogeneity.
We then address the likely endogeneity between ownership structure and risk-taking.
Previous research indicates that ownership structure depends on the firm’s valuation and
contracting environment. In particular, Demsetz and Lehn (1985) argue that the structure
of corporate ownership varies systematically in ways that are consistent with value
maximisation. Since performance and risk-taking are intimately related, we expect
ownership to be similarly dependent on the firm’s risk-taking behaviour. For example,
MLS may prefer to invest in high-risk firms because of the higher average returns they
can expect to earn. In that case, the positive relationship between MLS and risk-taking
would reflect an endogenous selection issue rather than a causal effect.
To tackle this problem, we use a propensity score matching (PSM) approach. In short,
the idea is to control for observable differences in firm characteristics between firms with
and without MLS. More precisely, each firm with MLS is matched with a firm without
MLS based on a propensity score that estimates the likelihood of having MLS. By doing
so, the procedure attempts to mimic the random assignment of subjects in treatment and
control groups (Rosenbaum and Rubin, 1983). Dehejia and Wahba (2002) demonstrate
that PSM can successfully replicate the outcome of a randomised experiment in labour
economics.
To implement the PSM method, we use a probit model where the dependent variable is
MLSD. The explanatory variables are firm-level characteristics that have been found in
previous studies to determine the presence of large blockholders (e.g., Demsetz and
Lehn, 1985; Holderness, 2009; Richter and Weiss, 2013; Villalonga and Amit, 2010).
Specifically, we use (i) firm size, (ii) firm age, (iii) leverage, (iv) free cash-flows, and (v)
asset tangibility. The model also includes industry and year dummies.
Firm size (SIZE) is expected to decrease the presence of MLS since it is more costly for
investors to hold a large fraction of ownership and control rights in a large firm. In
addition, risk aversion should discourage investors from holding a large fraction of their
wealth in a single firm and forgoing the benefit of diversification (Demsetz and Lehn,
1985; Villalonga and Amit, 2010). Similarly, firm age (AGE) is expected to decrease the
presence of MLS because older firms are more likely to have gone through several
rounds of external funding that diluted the ownership of their historical shareholders.
Furthermore, the founders of closely held firms may have incentives to sell their stakes
over time to diversify their wealth, which increases the likelihood of the firms being
widely held (Black and Gilson, 1998; Claessens et al., 2000; Holderness, 2009).
Leverage (LEVERAGE) is expected to be negatively associated with the presence of
MLS, since the disciplining role it plays may already be adequate without the
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 729

involvement of MLS. Free cash-flows (FCF) are expected to increase the presence of
MLS because free cash-flows are associated with a higher risk of opportunistic behaviour
by the LCS. Finally, asset tangibility (TANGIBLE) is expected to decrease the presence
of MLS, since firms with more tangible assets have less discretionary spending and can
be more easily monitored by capital providers (Habib and Ljungqvist, 2005).
After calibrating the probit model, we match, with replacement, each firm with MLS to
the closest firm with no MLS using the nearest-neighbour technique within a maximum
distance of 1% (Dehejia and Wahba, 2002). This procedure yields a propensity score
matched sample that consists of 2,774 firm–year observations (30,670 firm–year
observations when using monthly stock returns). We then verify that the matching
procedure generates a balanced sample of firms with and without MLS. Table 10 shows
that all the variables used in the PSM model display mean values that are nearly
indistinguishable across the two groups, in contrast to the original sample. To further
examine the covariate balance between the two samples, we report the standardised bias,
which is the difference in means between firms with and without MLS, divided by the
average standard deviation of the variable for the two groups (Rosenbaum and Rubin,
1985). The lower this bias is, the more balanced the two groups are in relation to the
variable under consideration. Table 10 shows that the absolute values of the standardised
bias for the PSM sample are less than 9.4%, indicating that the matches are reasonably
balanced with respect to all the relevant variables. In the case of LEVERAGE and FCF,
the bias is less than 2%. Moreover, the reduction in bias relative to the original sample is
sizeable, ranging from 65.3% to 86.3%.
Using the PSM sample, we re-estimate Models (2) to (5). The endogenous covariate is
MLSD. Table 11 displays the estimation of the probit model in Column (1). As expected,
firm size and age have a strong negative impact on the presence of MLS, while FCF has a
strong positive influence. Asset tangibility also has a significant negative influence. On
the other hand, leverage has a negative but insignificant effect on the presence of MLS.
Columns (2) to (4) reveal that MLSD has a positive and significant impact on each
indicator of risk-taking. For example, the absolute deviation in ROA is about 1.64%
higher in the presence of MLS. For each risk measure, the estimated effect is larger in the
PSM sample compared to the effect in the original sample. This suggests that MLS tend
to self-select into firms that decrease their risk the most. Accordingly, their positive
influence on the firm’s risk-taking behaviour is actually stronger than the effect indicated
by the regressions based on the original (unmatched) sample.

4.3. Role of MLS identity and firm-level governance


While MLS appear to have a positive effect on firm risk, all types of blockholders may
not contribute equally to disciplining the LCS. We distinguish four types of
blockholders: family owners, the French State, and widely held non-financial firms
and financial institutions. More specifically, we focus on the second largest shareholder,
since it has the strongest voting power to contest the LCS’s decisions. Shareholder
coalitions opposing the LCS are also likely to include or form around the second largest
shareholder.
Table 12 presents the regressions using the second largest shareholder’s identity and
voting power. Specifically, we replace the variable MLSVAR in Models (2) to (5) by the
dummies FAMILY_2, STATE_2, CORPORATE_2 and FINANCIAL_2, which equal one
if the second largest shareholder is a family, the State, a widely held firm or a widely held
© 2016 John Wiley & Sons, Ltd.
Table 10
730

Comparison of variables determining the presence of multiple large shareholders

This table compares the mean values of the variables considered to determine the presence of MLS in the original and in the matched samples. SIZE is the natural
logarithm of total assets (in thousands of euros). AGE is the number of years since the firm’s first date of incorporation. LEVERAGE is total debt over total assets.
FCF is free cash-flows divided by total assets. TANGIBLE is property, plants and equipment, divided by total assets. All continuous variables are winsorised at the
1st and 99th percentiles. The t-statistics are in parentheses beneath each difference in means.  ,  and  denote statistical significance at the 1%, 5% and 10%

© 2016 John Wiley & Sons, Ltd.


level, respectively.

Original sample Propensity score matched sample

Firms with Firms w/o Diff. in Standardized Firms with Firms w/o Diff. in Standardized Reduction in
Variable MLS MLS means (t-stat) bias (%) MLS MLS means (t-stat) bias (%) bias (%)
SIZE 11.483 12.116 0.633 28.9 11.483 11.396 0.087 4.0 86.3
(8.26) (1.03)
AGE 38.293 47.030 8.737 27.6 38.293 41.269 2.976 9.4 65.9
(8.00) (2.47)
LEVERAGE 0.218 0.324 0.106 6.2 0.218 0.201 0.017 1.0 83.9
(1.64) (1.57)
FCF 0.010 0.003 0.007 7.3 0.010 0.008 0.002 1.9 73.3
(2.15) (0.49)
TANGIBLE 0.152 0.180 0.028 19.8 0.152 0.162 0.010 6.9 65.3
(5.73) (1.77)
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi
Multiple Large Shareholders and Corporate Risk-taking 731

Table 11
Risk-taking regressions using matched firms

This table presents regression results on the effect of MLS on risk-taking using a propensity score
matched sample. In Column (1), the presence of MLS (MLSD ¼ 1) is estimated using a probit regression
and the full sample (3,446 observations). Propensity score is the estimated probability of MLS presence.
Each firm with MLS (MLSD ¼ 1) is then matched with a firm without MLS (MLSD ¼ 0) whose
propensity score is the closest in the same industry. In Columns (2)–(4), the risk-taking regressions are
estimated using this matched sample (2  1,387 ¼ 2,774 firm–year). MLSD equals 1 if the firm has at
least two large shareholders, 0 otherwise. EXCESS CONTROL is the excess control of the LCS, defined
as the difference between the LCS’s ultimate control rights and ultimate cash-flow rights, all divided by
her ultimate control rights. SIZE is the natural logarithm of total assets (in thousands of euros). CAPEX
is capital expenditures over sales. AGE is the number of years since the firm’s first date of incorporation.
LEVERAGE is total debt over total assets. DIVERSIFICATION is the number of business segments.
FCF is free cash-flows divided by total assets. TANGIBLE is property, plants and equipment, divided by
total assets. All continuous variables are winsorised at the 1st and 99th percentiles.  ,  and  denote
statistical significance at the 1%, 5% and 10% level, respectively.

Absolute Absolute Absolute


Probit for deviation deviation in deviation in
MLS presence in ROA Tobin’s Q stock returns

Variable (1) (2) (3) (4)


 
MLSD 1.63939 0.07308 0.00567
(4.82) (3.85) (2.29)
EXCESS CONTROL 1.96326 0.11679 0.01629
(2.37) (2.40) (2.80)
SIZE 0.08141 0.54726 0.01228 0.00475
(7.24) (5.72) (2.67) (6.22)
CAPEX 0.03623 0.00657 0.00006
(1.92) (1.65) (0.45)
AGE 0.00254 0.02040 0.00043 0.00014
(2.91) (3.74) (1.12) (3.12)
LEVERAGE 0.03774 3.69180 0.10906 0.03296
(1.62) (2.76) (1.20) (4.10)
DIVERSIFICATION 0.22613 0.00614 0.00169
(1.82) (0.88) (2.00)
FCF 0.0057940
(2.63)
TANGIBLE 0.0037956
(2.18)
Intercept 1.07230 25.15745 0.23437 0.13823
(6.01) (22.58) (4.35) (12.77)
Year/month dummies Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes
Number of observations 3,446 2,774 2,774 30,670
Chi2 223.75
Pseudo R2 0.048
Adjusted R2 0.188 0.112 0.053

© 2016 John Wiley & Sons, Ltd.


Table 12
732

Role of the second largest shareholder’s identity

This table examines the relationship between the identity of the second largest shareholder and the firm’s risk-taking. The dependent variables are the absolute
deviation from expected ROA (Columns (1)–(3)), the absolute deviation from expected Tobin’s Q (Columns (4)–(6)) and the absolute deviation from expected
returns using the market model (Columns (7)–(9)). FAMILY_2, STATE_2, CORPORATE_2 and FINANCIAL_2 equal 1 if the second largest shareholder is a
family, the State, a widely held corporation or a widely held financial institution, respectively, and 0 otherwise. VR2_FAMILY, VR2_STATE, VR2_CORPORATE

© 2016 John Wiley & Sons, Ltd.


and VR2_FINANCIAL equal the voting rights of the second largest shareholder if he/she is a family, the State, a widely held corporation or a widely held financial
institution, respectively, and 0 otherwise. EXCESS CONTROL is the excess control of the LCS, defined as the difference between the LCS’s ultimate control rights
and ultimate cash-flow rights, all divided by her ultimate control rights. SIZE is the natural logarithm of total assets (in thousands of euro). CAPEX is capital
expenditures over sales. AGE is the number of years since the firm’s first date of incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the
number of business segments. All continuous variables are winsorised at the 1st and 99th percentiles. The t-statistics in parentheses are computed with standard
errors clustered by firm.  ,  and  denote statistical significance at the 1%, 5% and 10% level, respectively.

Absolute deviation in ROA Absolute deviation in Tobin’s Q Absolute deviation in stock return

Variable (1) (2) (3) (4) (5) (6)


  
FAMILY_2 1.00256 0.04696 0.00697
(2.69) (2.59) (2.85)
STATE_2 0.36058 0.03769 0.00705
(0.30) (0.97) (0.48)
CORPORATE_2 0.44236 0.00679 0.00211
(0.83) (0.12) (0.46)
FINANCIAL_2 1.47677 0.07061 0.00131
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

(3.03) (2.79) (0.45)


VR2_FAMILY 4.43132 0.14754 0.03095
(2.62) (1.99) (2.74)
VR2_STATE 3.21557 0.12599 0.04527
(0.63) (0.62) (0.69)
VR2_CORPORATE 3.37979 0.22282 0.01680
Table 12
Continued

Absolute deviation in ROA Absolute deviation in Tobin’s Q Absolute deviation in stock return

Variable (1) (2) (3) (4) (5) (6)


(1.35) (0.70) (0.74)

© 2016 John Wiley & Sons, Ltd.


VR2_FINANCIAL 7.00254 0.31470 0.00936
(2.68) (2.55) (0.43)
EXCESS CONTROL 1.83936 1.64792 0.10051 0.09926 0.01245 0.01250
(2.48) (2.25) (2.41) (2.50) (2.48) (2.54)
SIZE 0.65402 0.67017 0.02185 0.02190 0.00622 0.00621
(6.66) (6.86) (3.92) (4.04) (10.02) (10.29)
CAPEX 0.02498 0.02532 0.00476 0.00482 0.00004 0.00006
(1.59) (1.60) (2.08) (2.09) (0.49) (0.68)
AGE 0.01849 0.01832 0.00022 0.00027 0.00009 0.00009
(3.37) (3.34) (0.68) (0.85) (2.19) (2.14)
LEVERAGE 3.86478 4.00936 0.03522 0.03459 0.04737 0.04790
(3.13) (3.25) (0.44) (0.43) (6.59) (6.70)
DIVERSIFICATION 0.17533 0.17796 0.00645 0.00671 0.00120 0.00123
(1.70) (1.72) (1.19) (1.23) (1.82) (1.86)
Intercept 11.78807 12.06064 0.47021 0.47270 0.13375 0.13368
(8.51) (8.93) (5.80) (6.01) (15.01) (15.48)
Year/Month dummies Yes Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes
Multiple Large Shareholders and Corporate Risk-taking

Number of observations 3,446 3,446 3,446 3,446 38,243 38,243


Adjusted R2 0.147 0.147 0.120 0.113 0.047 0.047
733
734 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

financial institution, respectively, and zero otherwise. These dummy variables are
included together in Columns (1), (3) and (5) to highlight their impact relative to the
absence of MLS. The results indicate that widely held financial institutions have a
significant positive effect on firm risk measured by the absolute deviation of ROA and
Tobin’s Q. Their well-diversified portfolios probably explain their greater ability and
therefore readiness to take on more risk (Faccio et al., 2011). The effect on the absolute
deviation in stock returns turns out to be insignificant. The likely cause for this result
seems to be the 2008 financial crisis, since the estimation for the period preceding the
crisis yields a significant positive coefficient.
As the second largest shareholder, families also appear to encourage higher risk-
taking. While they might arguably be as undiversified as the controlling family, they also
have fewer incentives to shun risk. Most importantly, they should be less concerned
about losing control in the event of financial distress since they have little control over the
firm. Hence, the downside associated with higher risk-taking does not impose a specific
cost on them. In addition, the second largest shareholders may not, in practice, be able to
extract substantial private benefits of control. While this case is possible (Zwiebel, 1995)
and could be significant in Asia, it does not seem to occur extensively in Europe (Faccio
et al., 2001). Consequently, only the LCS may have incentives to restrain risk to protect
his/her consumption of private benefits. This explains how having another family as the
second largest blockholder is beneficial to the firm’s risk-taking activities. In contrast,
widely held firms and the State do not seem to have a systematic influence on firm risk;
their motivations are likely to be different. For instance, other firms may hold a
significant equity stake to alleviate concerns related to the hold-up problem. Since these
concerns are unrelated to the recipient firm’s level of risk, there is no reason to expect the
presence of non-financial firms to be associated with a significant difference in risk-
taking.
We then use the voting rights of each type of second largest shareholder. To do
so, we consider the variables VR2_FAMILY, VR2_STATE, VR2_CORPORATE and
VR2_FINANCIAL, which equal the voting rights of the second largest shareholder if
he/she is a family, the State, a widely held firm or a widely held financial institution,
respectively, and zero otherwise. Since these variables have no common support (i.e., for
any observation, at most one variable can take a non-zero value), they are entered
together in columns (2), (4) and (6). Consistent with the previous results, the voting rights
of widely held financial institutions and other families are associated with significantly
higher firm risk measured by the absolute deviation of the ROA and Tobin’s Q. For these
two risk measures, the coefficients of the voting rights of financial institutions are about
twice as large as those of family owners. This result appears to suggest that concerns over
their under-diversified wealth prevent these family owners from taking on as much risk
as well-diversified financial institutions. Overall, the results in Table 12 are in line with
H5 and confirm that the effect of MLS on corporate risk-taking depends on their type.13
Another interesting issue is whether MLS have the same effect in well-governed firms.
After all, internal governance should be designed to prevent insiders from taking on a
suboptimal level of risk to serve their own objectives. Low (2009) demonstrates that

13
In untabulated regressions, we also use the votes of each type of second largest shareholder
relative to the votes of the LCS (controlling family). The results are broadly in line with those
based on raw voting rights.

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 735

executive stock options can incentivise insiders to not decrease their firm’s risk, even
when they have the opportunity to do so. To gain a better appreciation of how MLS could
improve the firm’s risk-taking in different governance environments, we interact the
dummy variable indicating the presence of other blockholders (MLSD) with various
internal governance variables considered to affect corporate risk-taking.14 The results are
provided in Table 13.
The first variable is the LCS’s excess control. As discussed in H3, excess control
increases the LCS’s incentives to extract private benefits of control and is likely to be
correlated with the LCS’s financial constraints. Hence, the higher the excess control, the
lower the firm’s risk-taking. In line with this argument, columns (1), (5) and (9) of
Table 13 indicate that the coefficient of excess control is significantly negative. On the
other hand, the coefficient of the interaction with MLSD is positive and significant. This
result shows that the presence of MLS is able to restrain the influence of the LCS.
Consistent with H6, the presence of MLS is most valuable when the firm’s internal
governance is poor but appears to be less significant otherwise. Furthermore, comparison
of the magnitudes of the coefficients of MLSD and its interaction with the LCS’s excess
control suggests that the presence of MLS largely neutralises the LCS’s negative
influence on firm risk.
The next governance mechanism is board size. Yermack (1996) argues that firms with
large boards are poorly governed, which explains their lower operating performance and
market value. Apart from being slow to react to negative shocks, large boards tend to
select conservative investments (Cheng, 2008; Nakano and Nguyen, 2012). The
literature on group decision-making suggests that large groups tend to opt for less
extreme choices. However, this prudence makes them vulnerable to being swept away by
bolder competitors, even though some may fail as a result of their risky strategies. We
repeat the same procedure by interacting the variable BOARD SIZE, proxied by the
natural logarithm of the number of directors on the board, with MLSD. Columns (2), (6)
and (10) of Table 13 indicate that the effect of board size on firm risk is significantly
negative, in line with prior research. However, larger boards could actually have a
positive influence on firm risk-taking in the presence of MLS. This outcome is consistent
with the assertion of Coles et al. (2006), that larger boards are not always detrimental to
firm performance. They can even enhance firm value by encouraging the firm to embrace
an appropriate level of risk. Nonetheless, caution should be applied to this conclusion,
since board size is not independent of the presence of MLS. In fact, the presence of MLS
is likely to be associated with larger boards since large shareholders tend to be
represented on the firm’s board of directors.
The third governance variable that may affect firm risk is board independence. Boards
composed of a high proportion of executives tend to be under the CEO’s influence. This
could provide the CEO with greater freedom to implement his/her preference for lower
risk-taking. In contrast, board independence should be associated with higher risk-
taking. Consistent with this argument, columns (3), (7) and (11) of Table 13 indicate that
the coefficients on BOARD INDEP, defined as the percentage of independent directors
on the firm’s board, are positive and statistically significant (at the 5% level or better). In
contrast, the coefficients of the interactions with MLSD are not statistically significant.

14
The results (unreported) remain qualitatively the same when we repeat the analysis using
alternative proxies for MLS.

© 2016 John Wiley & Sons, Ltd.


Table 13
736

Interactions between MLS presence and firm governance

This table presents regression results on the effect of firm internal governance on the relationship between MLS and risk-taking. The dependent variables are the
absolute deviation from expected ROA (Columns (1)–(4)), the absolute deviation from expected Tobin’s Q (Columns (5)–(8)) and the absolute deviation from
expected stock returns using the market model (Columns (9)–(12)). MLSD equals 1 if the firm has at least two large shareholders, 0 otherwise. EXCESS CONTROL
is the excess control of the LCS, defined as the difference between the LCS’s ultimate control rights and ultimate cash-flow rights, all divided by her ultimate control

© 2016 John Wiley & Sons, Ltd.


rights. BOARD SIZE is the natural logarithm of the number of directors on the board. BOARD INDEP is the percentage of independent directors on the firm’s board.
ONE_TIER equals 1 if the firm has a one-tier board structure, and 0 otherwise. SIZE is the natural logarithm of total assets (in thousands of euros). CAPEX is capital
expenditures over sales. AGE is the number of years since the firm’s first date of incorporation. LEVERAGE is total debt over total assets. DIVERSIFICATION is the
number of business segments. All continuous variables are winsorised at the 1st and 99th percentiles. The t-statistics in parentheses are computed with standard
errors clustered by firm.  ,  and  denote statistical significance at the 1%, 5% and 10% level, respectively.

Absolute deviation in ROA Absolute deviation in Tobin’s Q Absolute deviation in stock returns

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

MLSD 0.96647   0.88123   0.97428   0.99491   0.04615   0.04703   0.05300   0.04658   0.00468  0.00544  0.00459  0.00479 
(3.06) (2.85) (3.12) (3.27) (2.77) (2.87) (3.20) (2.95) (2.19) (2.57) (2.18) (2.26)
EXCESS CONTROL 1.94945  1.32379 2.32299   1.55026  0.10666  0.09008  0.10077  0.08900  0.01386  0.01026  0.01397   0.01146 
(2.50) (1.80) (3.07) (2.12) (2.53) (2.14) (2.38) (2.21) (2.55) (1.98) (2.79) (2.23)
EXCESS 2.12625 0.10282 0.01920 
CONTROL  MLSD
(1.67) (1.72) (2.25)
BOARD SIZE 1.12842   0.04611   0.01146  
(3.94) (2.60) (4.30)
BOARD SIZE  MLSD 1.03007 0.07850  0.01077 
(1.83) (2.42) (2.42)
Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

BOARD INDEP 1.60085  0.11857   0.00968 


(2.34) (3.11) (2.07)
BOARD 1.72715 0.07163 0.00743
INDEP  MLSD
(1.44) (1.22) (0.91)
ONE_TIER 0.74361  0.03340  0.00508 
(2.03) (2.05) (2.36)
Table 13
Continued

Absolute deviation in ROA Absolute deviation in Tobin’s Q Absolute deviation in stock returns

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
ONE_TIER  MLSD 2.98641   0.19144   0.01536  
(4.57) (6.66) (3.57)

© 2016 John Wiley & Sons, Ltd.


SIZE 0.65563   0.54569   0.71328   0.63335   0.02070   0.01762   0.02806   0.02061   0.00614   0.00472   0.00662   0.00615  
(6.68) (5.85) (6.91) (6.53) (3.67) (3.12) (5.31) (3.89) (9.76) (7.03) (10.19) (9.73)
CAPEX 0.02496 0.02263 0.02456 0.02799 0.00463  0.00463  0.00469  0.00496  0.00006 0.00006 0.00005 0.00004
(1.59) (1.45) (1.59) (1.84) (2.06) (2.08) (2.10) (2.23) (0.77) (0.73) (0.66) (0.54)
AGE 0.01786   0.01500   0.01713   0.01943   0.00020 0.00016 0.00018 0.00026 0.00010  0.00008 0.00009  0.00011 
(3.25) (2.72) (3.15) (3.46) (0.63) (0.49) (0.57) (0.85) (2.31) (1.89) (2.18) (2.52)
LEVERAGE 3.78373   3.60702   3.88008   3.43330   0.05677 0.03738 0.01980 0.05785 0.04752   0.04483   0.04794   0.04580  
(3.05) (2.98) (3.11) (2.84) (0.69) (0.48) (0.26) (0.77) (6.50) (6.27) (6.57) (6.30)
DIVERSIFICATION 0.16680 0.14344 0.17966 0.14915 0.00694 0.00563 0.00594 0.00843 0.00117 0.00124 0.00120 0.00115
(1.61) (1.43) (1.76) (1.39) (1.25) (1.05) (1.17) (1.61) (1.78) (1.79) (1.87) (1.69)
Intercept 11.53942   10.74474   12.47770   11.66511   0.46073   0.46760   0.57475   0.48213   0.13040   0.11861   0.13914   0.13322  
(8.50) (7.73) (9.11) (8.14) (5.68) (6.93) (9.56) (6.20) (15.16) (12.35) (15.93) (15.46)
Year/Month dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Number of observations 3,446 3,446 3,446 3,446 3,446 3,446 3,446 3,446 38,243 38,243 38,243 38,243
Adjusted R2 0.146 0.155 0.149 0.159 0.118 0.132 0.133 0.146 0.047 0.050 0.047 0.049
Multiple Large Shareholders and Corporate Risk-taking
737
738 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

This result can be interpreted as suggesting that the governance role of MLS is less
important in the presence of independent boards. However, it is important to keep in
mind that the two interacted variables are probably correlated. For instance, suppose that
the presence of MLS is associated with greater board independence. Accordingly, the
positive effect on firm risk would be jointly determined by these two variables, making it
difficult to dissociate their respective contributions.
The fourth governance variable is an indicator for unitary (one-tier) board structure
(ONE_TIER). This variable constitutes another measure of the board’s ability to monitor
management. Maury (2006) shows that firms with a one-tier board structure are less
likely to replace poorly performing CEOs than firms with a two-tier board structure. As a
matter of fact, two-tier boards clearly separate the responsibilities of managers and those
of their monitors. Firms with two-tier boards are hence expected to be better monitored
and should take on more risk (or should be prevented from taking on less risk). In
contrast, firms with unitary boards could offer their managers greater opportunity to
decrease risk. Columns (4), (8) and (12) of Table 13 confirm this assumption with a
significant negative coefficient for the indicator for a one-tier board. The presence of
MLS changes the sign of the association (ONE_TIER  MLSD) with firm risk. The
coefficient is also highly significant. In other words, unitary boards are able to act in
the best interests of shareholders by taking on the appropriate level of risk only in the
presence of MLS.
Overall, the results in Table 13 indicate that the benefit of MLS lies in their ability to
neutralise the deficiencies of firms’ internal governance structures, which would
otherwise lead to a decrease in risk-taking detrimental to shareholders.

5. Conclusion

The presence of MLS is considered to enhance corporate governance (Bennedsen and


Wolfenzon, 2000; Bloch and Hege, 2001; Bolton and von Thadden, 1998; Pagano and
Roëll, 1998) and has been associated with higher firm valuation (Attig et al., 2009; Jara-
Bertin et al., 2008; Laeven and Levine, 2008; Maury and Pajuste, 2005). The mechanism
through which MLS improve firm performance is, however, not clearly established. The
conventional view is that MLS prevent the LCS from diverting corporate resources for
his/her own benefit. The tunnelling of cash-flows and related party transactions are
typical examples, especially in emerging markets, where the rule of law is often poorly
enforced. Another way the LCS can divert corporate resources from the firm’s best use is
by dissuading the firm from carrying out high-risk projects despite evidence that these
projects tend to create more value. Minority shareholders are short-changed by this
suboptimal allocation of resources. However, the LCS can better protect his/her control
of the firm and the stream of private benefits that can be derived from his/her controlling
position (John et al., 2008; Mishra, 2011).
MLS can contest the LCS’s preference for lower firm risk and their presence is
hypothesised to be associated with higher risk-taking. This hypothesis is supported using
a large sample of French listed firms over the period 20032012. When MLS are absent
or in a weak position to challenge the LCS, operating performance, market value and
stock returns exhibit significantly lower variability, suggesting that firms are selecting
low-risk projects. In contrast, the presence and greater voting power of MLS are found to
result in greater performance variability, consistent with the selection of riskier
investments. The difference in risk is not only statistically significant but also
© 2016 John Wiley & Sons, Ltd.
Multiple Large Shareholders and Corporate Risk-taking 739

economically large. For instance, the average deviation from the firm’s expected ROA is
found to be one-third higher when MLS are present in the firm’s ownership structure.
By opposing the LCS’s preference for low-risk projects and obstructing his/her plans
to tilt the firm toward more conservative policies, MLS play an important role that could
explain why their presence and voting power are associated with higher market value and
why investors are more willing to invest in these firms (Attig et al., 2008). Our results
complement those of Mishra (2011) for Asian firms. However, our context is quite
different: France is a developed economy with a stronger legal system that better protects
small investors against expropriation (Djankov et al., 2008; La Porta et al., 1998). In
addition, extra-legal factors, such as public opinion pressure and media independence,
help curb the extraction of private benefits (Dyck and Zingales, 2004). However, our
results are quite similar, which underlines the powerful influence that MLS can have on a
firm’s decisions.
Using a methodology for measuring risk that allows taking full advantage of the panel
structure of the data, we provide two new results. First we show that the identity of the
second largest shareholder is important. Financial institutions are able to promote greater
risk-taking probably because of their well-diversified portfolios, as Faccio et al. (2011)
have established. Family owners are also able to push firms to take on higher risk. The
fact that wealth diversification does not appear to be the key difference with the LCS
(controlling family) suggests that concerns related to control retention are important
factors in determining the level of firm risk. Second, we show that the presence of MLS is
beneficial due to their ability to counterbalance some of the firm’s governance
deficiencies. More specifically, MLS appear to offset the tendency of firms with large
and less independent boards to make more conservative investments.

Appendix

Variable definitions

Variable name Variable definition Source


ROA The ratio of earnings before interest and Worldscope.
taxes to the book value of assets at the
beginning of the year.
Tobin’s Q The market-to-book value of assets at the Worldscope and
end of the year. authors’ calculations.
LCS Dummy variable that takes the value of Annual reports and
one if the firm has a large controlling authors’ calculations.
shareholder at the 10% threshold, and
zero otherwise.
LCS_FAMILY Dummy variable that takes the value of As above.
one if the firm’s ultimate owner is a
family, and zero otherwise.
LCS_STATE Dummy variable that takes the value of As above.
one if the firm’s ultimate owner is the
State, and zero otherwise.

© 2016 John Wiley & Sons, Ltd.


740 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Continued
Variable name Variable definition Source
LCS_CORPORATE Dummy variable that takes the value of As above.
one if the firm’s ultimate owner is a
widely held corporation, and zero
otherwise.
LCS_FINANCIAL Dummy variable that takes the value of As above.
one if the firm’s ultimate owner is a
widely held financial institution, and
zero otherwise.
EXCESS CONTROL The difference between the LCS’s As above.
ultimate control and cash-flow rights,
divided by her ultimate control rights
(i.e., (UCO  UCF)/UCO)).
MLSD Dummy variable that takes the value of As above.
one if the firm has at least two large
shareholders (at the 10% threshold),
and zero otherwise.
MLSN The number of large shareholders (at the As above.
10% threshold), other than the LCS, up
to the fourth.
VR234 The sum of voting rights of the second, As above.
third and fourth largest blockholders.
VRRATIO The ratio of the sum of voting rights of the As above.
second, third and fourth largest
blockholders to the voting rights of the
LCS.
DISPERSION This variable is calculated as follows: As above.
DISPERSION ¼ ðVR1  VR2Þ2 þ
ðVR2  V R3Þ2 þ ðVR3  VR4Þ2
Where VR1, VR2, VR3 and VR4 equal
the voting rights of the first, second,
third and fourth largest shareholders,
respectively.
FAMILY_2 Dummy variable that equals one if the As above.
second largest shareholder is a family,
and zero otherwise.
STATE_2 Dummy variable that equals one if the As above.
second largest shareholder is the State,
and zero otherwise.
CORPORATE_2 Dummy variable that equals one if the As above.
second largest shareholder is a widely
held corporation, and zero otherwise.
FINANCIAL_2 Dummy variable that equals one if the As above.
second largest shareholder is a widely
held financial institution, and zero
otherwise.

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 741

Continued
Variable name Variable definition Source
VR2_FAMILY The voting rights of the second largest As above.
shareholder if he/she is a family, and
zero otherwise.
VR2_STATE The voting rights of the second largest As above.
shareholder if he/she is the State, and
zero otherwise.
VR2_CORPORATE The voting rights of the second largest As above.
shareholder if he/she is a widely held
corporation, and zero otherwise.
VR2_FINANCIAL The voting rights of the second largest As above.
shareholder if he/she is a widely held
financial institution, and zero
otherwise.
SIZE The natural logarithm of total assets. Worldscope.
CAPEX Capital expenditures divided by sales. As above.
AGE The number of years since the firm’s first Annual reports and
date of incorporation. authors’ calculations.
LEVERAGE The ratio of total debt to total assets Worldscope.
DIVERSIFICATION The number of business segments in Worldscope.
which the firm operates (using two-
digit SIC codes).
UCF The ultimate cash-flow rights of the LCS, Annual reports and
defined as the sum of products of direct authors’ calculations.
cash-flow rights along the different
ownership chains.
GROUP A dummy variable that equals one if the As above.
LCS is affiliated with a business group,
and zero otherwise.
EM A measure of the extent of earnings Worldscope and
management, calculated as in Leuz authors’ calculations.
et al. (2003).
FCF The firm’s free cash-flows divided by Worldscope and
total assets. authors’ calculations.
TANGIBLE The firm’s property, plants and Worldscope.
equipment, divided by total assets.
BOARD SIZE The natural logarithm of the number of As above.
directors on the board.
BOARD INDEP The percentage of independent directors As above.
on the board.
ONE_TIER Dummy variable that equals one if the As above.
firm has a one-tier board structure, and
zero otherwise.

© 2016 John Wiley & Sons, Ltd.


742 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

References

Acharya, V., Amihud, Y. and Litov, L., ‘Creditor rights and corporate risk-taking’, Journal of Financial
Economics, Vol. 102, 2011, pp. 150–66.
Adams, R., Almeida, H. and Ferreira, D., ‘Powerful CEOs and their impact on corporate performance’,
Review of Financial Studies, Vol. 18, 2005, pp. 1403–32.
Alchian, A., ‘Some economics of property rights’, in: Alchian, A. (ed.), Economic Forces at Work
(Indianapolis, Liberty Press, 1977), pp. 127–49.
Almazan, A., Hartzell, J.C. and Starks, L.T., ‘Active institutional shareholders and costs of monitoring:
Evidence from executive compensation’, Financial Management, Vol. 34, 2005, pp. 5–34.
Almeida, H. and Wolfenzon, D., ‘A theory of pyramidal ownership and family business groups’, The
Journal of Finance, Vol. 61, 2006, pp. 2637–80.
Anderson, R.C., Duru, A. and Reeb, D., ‘Investment policy in family controlled firms’, Journal of
Banking and Finance, Vol. 36, 2012, pp. 1744–58.
Anderson, R.C. and Reeb, D., ‘Founding-family ownership, corporate diversification, and firm
leverage’, Journal of Law and Economics, Vol. 46, 2003, pp. 653–84.
Attig, N., Boubakri, N., El Ghoul, S. and Guedhami, O., ‘The global financial crisis, family control, and
dividend policy’, Financial Management, forthcoming 2016.
Attig, N., El Ghoul, S. and Guedhami, O., ‘Do multiple large shareholders play a corporate governance
role? Evidence from East Asia’, Journal of Financial Research, Vol. 32, 2009, pp. 395–422.
Attig, N., El Ghoul, S., Guedhami, O. and Rizeanu, S., ‘The governance role of multiple large
shareholders: Evidence from the valuation of cash holdings’, Journal of Management and
Governance, Vol. 17, 2013, pp. 419–51.
Attig, N., Guedhami, O. and Mishra, D., ‘Multiple large shareholders, control contests, and implied cost
of equity’, Journal of Corporate Finance, Vol. 14, 2008, pp. 721–37.
Bae, K-H., Kang, J-K. and Kim, J-M., ‘Tunneling or value added? Evidence from mergers by Korean
business groups’, The Journal of Finance, Vol. 57, 2002, pp. 2695–740.
Baek, J.-S., Kang, J.-K. and Lee, I., ‘Business groups and tunneling: Evidence from private securities
offerings by Korean chaebols’, The Journal of Finance, Vol. 61, 2006, pp. 2415–49.
Barro, R., ‘Economic growth in a cross-section of countries’, Quarterly Journal of Economics,
Vol. 106, 1991, pp. 407–43.
Bebchuk, L., Kraakman, R. and Triantis, G., ‘Stock pyramids, cross-ownership, dual-class equity: the
creation of agency costs of separating control from cash-flow rights’, in Morck, R.K. (ed.),
Concentrated Corporate Ownership (Chicago, University of Chicago Press, 2000), pp. 295–318.
Belenzon, S. and Berkovitz, T., ‘Innovation in business groups’, Management Science, Vol. 56, 2010,
pp. 519–35.
Belkhir, M., Boubaker, S., and Derouiche, I., ‘Control-ownership wedge, board of directors, and the
value of excess cash’, Economic Modelling, Vol. 39, 2014, pp. 110–22.
Bennedsen, M. and Nielsen, K.M., ‘Incentive and entrenchment effects in European ownership’,
Journal of Banking and Finance, Vol. 34, 2010, pp. 2212–29.
Bennedsen, M., Nielsen, K.M., Perez-Gonzalez, F. and Wolfenzon, D., ‘Inside the family firm: the role
of families in succession decisions and performance’, Quarterly Journal of Economics, Vol. 122,
2007, pp. 647–91.
Bennedsen, M., and Wolfenzon, D., ‘The balance of power in closely held corporations’, Journal of
Financial Economics, Vol. 58, 2000, pp. 113–39.
Bertrand, M., Mehta, P. and Mullainathan, S., ‘Ferreting out tunneling: an application to Indian business
groups’, Quarterly Journal of Economics, Vol. 117, 2002, pp. 121–48.
Black, B., and R. Gilson, ‘Venture capital and the structure of capital markets: Banks versus stock
markets’, Journal of Financial Economics, Vol. 47, 1998, pp. 243–77.
Bloch, F. and Hege, U., ‘Multiple shareholders and control contests’, Working Paper Aix-Marseille
University (2001).
Bolton, P. and von Thadden, E.-L., ‘Blocks, liquidity and corporate control’, The Journal of Finance,
Vol. 53, 1998, pp. 1–25.

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 743

Boubaker, S. and Labegorre, F., ‘Ownership structure, corporate governance and analyst following: a
study of French listed firms’, Journal of Banking and Finance, Vol. 32, 2008, pp. 961–76.
Burkart, M., Panunzi, F. and Shleifer, A., ‘Family firms’, The Journal of Finance, Vol. 58, 2003,
pp. 2167–202.
Campbell, J.Y., ‘Understanding risk and return’, Journal of Political Economy, Vol. 104, 1996,
pp. 298–345.
Caprio, L., Croci, E. and Del Giudice, A., ‘Ownership structure, family control, and acquisition
decisions’, Journal of Corporate Finance, Vol. 17, 2011, pp. 1636–57.
Cespedes, J., Gonzalez, M. and Molina, C.A., ‘Ownership and capital structure in Latin America’,
Journal of Business Research, Vol. 63, 2010, pp. 248–54.
Chandler, A.D., Scale and Scope: The Dynamics of Industrial Capitalism (Cambridge, MA, Harvard
University Press, 1990).
Chen, X., Harford, J. and Li, K., ‘Monitoring: Which institutions matter?’ Journal of Financial
Economics, Vol. 86, 2007, pp. 279–305.
Cheng, S., ‘Board size and the variability of corporate performance’, Journal of Financial Economics,
Vol. 87, 2008, pp. 157–76.
Cheung, Y., Rau, P.R. and Stouraitis, A., ‘Tunneling, propping, and expropriation: Evidence from
connected party transactions in Hong Kong’, Journal of Financial Economics, Vol. 82, 2006,
pp. 343–86.
Chin, C., Chen, Y.-J., Kleinman, G. and Lee, P., ‘Corporate ownership structure and innovation:
Evidence from Taiwan’s electronics industry’, Journal of Accounting, Auditing and Finance,
Vol. 24, 2009, pp. 145–75.
Claessens, S., Djankov, S., Fan, J. and Lang, L., ‘Disentangling the incentive and entrenchment effects
of large shareholdings’, The Journal of Finance, Vol. 57, 2002, pp. 2741–71.
Claessens, S., Djankov, S. and Lang, L., ‘The separation of ownership and control in East Asian
corporations’, Journal of Financial Economics, Vol. 58, 2000, pp. 81–112.
Coles, J., Daniel, N. and Naveen, L., ‘Managerial incentives and risk-taking’, Journal of Financial
Economics, Vol. 79, 2006, pp. 431–68.
Croci, E., Doukas, J. and Gonenc, H., ‘Family control and financing decisions’, European Financial
Management, Vol. 17, 2011, pp. 860–97.
Cronqvist, H. and Nilsson, M., ‘Agency costs of controlling minority shareholders’, Journal of
Financial and Quantitative Analysis, Vol. 38, 2003, pp. 695–719.
Dehejia, R.H. and Wahba, S., ‘Propensity score matching methods for nonexperimental causal studies’,
Review of Economics and Statistics, Vol. 84, 2002, pp. 151–61.
DeLong, B. and Summers, L., ‘Equipment investment and economic growth’, Quarterly Journal of
Economics, Vol. 106, 1991, pp. 445–502.
Demsetz, H. and Lehn, K., ‘The structure of corporate ownership: Causes and consequences’, Journal
of Political Economy, Vol. 93, 1985, pp. 1155–77.
Denis, D., Denis, D. and Sarin, A., ‘Agency problems, equity ownership, and corporate diversification’,
The Journal of Finance, Vol. 52, 1997, pp. 135–60.
Dewenter, K.L. and Malatesta, P.H., ‘State-owned and privately owned firms: an empirical analysis
of profitability, leverage, and labor intensity’, American Economic Review, Vol. 91, 2001,
pp. 320–34.
Dittmar A. and Mahrt-Smith J., ‘Corporate governance and the value of cash holdings’, Journal of
Financial Economics, Vol. 83, 2007, pp. 599–634.
Djankov, S., La Porta, R., Lopez-de-Silanes, F. and Shleifer, A., ‘The law and economics of self-
dealing’, Journal of Financial Economics, Vol. 88, 2008, pp. 430–65.
Doidge, C., Karolyi, G.A., Lins, K.V., Miller, D.P. and Stulz, R.M., ‘Private benefits of control,
ownership, and the cross-listing decision’, The Journal of Finance, Vol. 64, 2009, pp. 425–66.
Drago, C., Millo, F., Ricciuti, R. and Santella, P., ‘Corporate governance reforms, interlocking
directorship and company performance in Italy’, International Review of Law and Economics,
Vol. 41, 2015, pp. 38–49.

© 2016 John Wiley & Sons, Ltd.


744 Sabri Boubaker, Pascal Nguyen and Wael Rouatbi

Dyck, A. and Zingales, L., ‘Private benefits of control: an international comparison’, The Journal of
Finance, Vol. 59, 2004, pp. 537–600.
Edmans, A. and Manso, G., ‘Governance through trading and intervention: a theory of multiple
blockholders’, Review of Financial Studies, Vol. 24, 2011, pp. 2395–428.
Faccio, M. and Lang, L., ‘The ultimate ownership of Western European corporations’, Journal of
Financial Economics, Vol. 65, 2002, pp. 365–95.
Faccio, M., Lang, L. and Young, L., ‘Dividends and expropriation’, American Economic Review,
Vol. 91, 2001, pp. 54–78.
Faccio, M., Marchica, M.-T. and Mura, R., ‘Large shareholder diversification and corporate risk-
taking’, Review of Financial Studies, Vol. 24, 2011, pp. 3601–41.
Glejser, H., ‘A new test for heteroskedasticity’, Journal of the American Statistical Association,
Vol. 64, 1969, pp. 316–23.
Gomes, A. and Novaes, W., ‘Sharing of control as a corporate governance mechanism’, Working Paper,
University of Pennsylvania (2005).
Guay, W., ‘The sensitivity of CEO wealth to equity risk: an analysis of the magnitude and
determinants’, Journal of Financial Economics, Vol. 53, 1999, pp. 43–71.
Guth, W., Nikiforakis, N. and Normann, H.-T., ‘Vertical cross-shareholding: Theory and experimental
evidence,’ International Journal of Industrial Organization, Vol. 25, 2007, pp. 69–89.
Habib, M. and Ljungqvist, A., ‘Firm value and managerial incentives: a stochastic frontier approach’,
Journal of Business, Vol. 78, 2005, pp. 2053–94.
Holderness, C., ‘The myth of diffuse ownership in the United States’, Review of Financial Studies,
Vol. 22, 2009, pp. 1377–408.
Jara-Bertin, M., Lopez-Iturriaga, F. and Lopez-de-Foronda, O., ‘The contest to control in European
family firms: How other shareholders affect firm value’, Corporate Governance: an International
Review, Vol. 16, 2008, pp. 146–59.
John, K., Litov, L. and Yeung, B., ‘Corporate governance and risk-taking’, The Journal of Finance,
Vol. 63, 2008, pp. 1679–728.
Johnson, S., La Porta, R., Lopez-de-Silanes, F. and Shleifer, A., ‘Tunneling’, American Economic
Review, Vol. 90, 2000, pp. 22–7.
Kahn, C. and Winton, A., ‘Ownership structure, speculation, and shareholder intervention’, The
Journal of Finance, Vol. 53, 1998, pp. 99–129.
Kole, S.R. and Mulherin, J.H., ‘The government as a shareholder: a case from the United States’,
Journal of Law and Economics, Vol. 40, 1997, pp. 1–22.
Kothari, S.P., Laguerre, T. and Leone, A., ‘Capitalization versus expensing: Evidence on the
uncertainty of future earnings from capital expenditures versus R&D outlays’, Review of Accounting
Studies, Vol. 7, 2002, pp. 355–82.
Laeven, L. and Levine, R., ‘Complex ownership structures and corporate valuations’, Review of
Financial Studies, Vol. 21, 2008, pp. 597–604.
La Porta, R., Lopez-de-Silanes, F. and Shleifer, A., ‘Corporate ownership around the world’, The
Journal of Finance, Vol. 54, 1999, pp. 471–518.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R., ‘Law and finance’, Journal of Political
Economy, Vol. 106, 1998, pp. 1113–55.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R., ‘Investor protection and corporate
valuation’, The Journal of Finance, Vol. 57, 2002, pp. 1147–70.
Lehmann, E. and Weigand, J., ‘Does the governed corporation perform better? Governance structures
and corporate performance in Germany’, European Finance Review, Vol. 4, 2000, pp. 157–95.
Lemmon, M.L. and Lins, K.V., ‘Ownership structure, corporate governance, and firm value: Evidence
from the East Asian financial crisis’, The Journal of Finance, Vol. 58, 2003, pp. 1445–68.
Leuz, C., Nanda, D. and Wysocki, P., ‘Earnings management and investor protection: an international
comparison’, Journal of Financial Economics, Vol. 69, 2003, pp. 505–27.
Lins, K.V., Volpin, P. and Wagner, H.F., ‘Does family control matter? International evidence from the
20082009 financial crisis’, Review of Financial Studies, Vol. 26, 2013, pp. 2583–619.

© 2016 John Wiley & Sons, Ltd.


Multiple Large Shareholders and Corporate Risk-taking 745

Low, A., ‘Managerial risk-taking behavior and equity-based compensation’, Journal of Financial
Economics, Vol. 92, 2009, pp. 470–90.
Luo, X. and Chung, C.N., ‘Keeping it all in the family: the role of particularistic relationships in
business group performance during institutional transition’, Administrative Science Quarterly,
Vol. 50, 2005, pp. 404–39.
Maury, B., ‘Corporate performance, corporate governance, and top executive turnover in Finland’,
European Financial Management, Vol. 12, 2006, pp. 221–48.
Maury, B. and Pajuste, A., ‘Multiple large shareholders and firm value’, Journal of Banking and
Finance, Vol. 29, 2005, pp. 1813–34.
Mishra, D., ‘Multiple large shareholders and corporate risk taking: Evidence from East Asia’,
Corporate Governance: an International Review, Vol. 19, 2011, pp. 507–28.
Morck, R., and Yeung, B., ‘Family control and the rent-seeking society’, Entrepreneurship Theory and
Practice, Vol. 28, 2004, pp. 391–409.
Nakano, M. and Nguyen, P., ‘Board size and corporate risk taking: Further evidence from Japan’,
Corporate Governance: an International Review, Vol. 20, 2012, pp. 369–87.
Ortega Argiles, R., Moreno, R. and Caralt, J.S., ‘Ownership structure and innovation: Is there a real
link?’ Annals of Regional Science, Vol. 39, 2005, pp. 637–62.
Ozkan, A. and Ozkan, N., ‘Corporate cash holdings: an empirical investigation of UK companies’,
Journal of Banking and Finance, Vol. 28, 2004, pp. 2103–34.
Pagano, M. and R€oell, A., ‘The choice of stock ownership structure: Agency costs, monitoring, and the
decision to go public’, Quarterly Journal of Economics, Vol. 113, 1998, pp. 187–226.
Paligorova, T., ‘Corporate risk taking and ownership structure’, Working Paper, Bank of Canada
(2010).
Rajgopal, S. and Shevlin, T., ‘Empirical evidence on the relation between stock option compensation
and risk taking’, Journal of Accounting and Economics, Vol. 33, 2002, pp. 145–71.
Richter, A. and Weiss, C., ‘Determinants of ownership concentration in public firms: the importance of
firm-, industry- and country-level factors’, International Review of Law and Economics, Vol. 33,
2013, pp. 1–14.
Rosenbaum, P. and Rubin, D., ‘The central role of the propensity score in observational studies for
causal effects’, Biometrika, Vol. 70, 1983, pp. 41–55.
Rosenbaum, P. and Rubin, D., ‘Constructing a control group using multivariate matched sampling
methods that incorporate the propensity score’, American Statistician, Vol. 39, 1985, pp. 33–38.
Schulze, W.S., Lubatkin, M.H., Dino, R.N. and Buchholtz, A.K., ‘Agency relationships in family firms:
Theory and evidence’, Organization Science, Vol. 12, 2001, pp. 99–116.
Shleifer, A. and Vishny, R.W., ‘Large shareholders and corporate control’, Journal of Political
Economy, Vol. 94, 1986, pp. 461–88.
Shleifer, A. and Vishny, R.W., ‘A survey of corporate governance’, The Journal of Finance, Vol. 52,
1997, pp. 737–83.
Villalonga, B. and Amit, A., ‘How do family ownership, control, and management affect firm value?’
Journal of Financial Economics, Vol. 80, 2006, pp. 385–417.
Villalonga, B. and Amit, R., ‘Family control of firms and industries’, Financial Management, Vol. 39,
2010, pp. 863–904.
Winton, A., ‘Limitation of liability and the ownership structure of the firm’, The Journal of Finance,
Vol. 48, 1993, pp. 487–512.
Yermack, D., ‘Higher market valuation of companies with a small board of directors’, Journal of
Financial Economics, Vol. 40, 1996, pp. 185–211.
Zhang, G., ‘Ownership concentration, risk aversion and the effect of financial structure on investment
decisions’, European Economic Review, Vol. 42, 1998, pp. 1751–78.
Zhou, X., ‘Understanding the determinants of managerial ownership and its relationship to firm
performance’, Journal of Financial Economics, Vol. 62, 2001, pp. 559–71.
Zwiebel, J., ‘Block investment and partial benefits of corporate control’, Review of Economic Studies,
Vol. 62, 1995, pp. 161–85.

© 2016 John Wiley & Sons, Ltd.

You might also like