Corporate Financenext Term and State Enterprise Reform in Previous Termchinanext Term

China Economic Review 16 (2005) 118 – 148
Corporate finance and state enterprise reform

in China
Dongwei SU*
Department of Finance, Jinan University, Guangzhou, Guangdong 510632, People’s Republic of China
Accepted 20 September 2004
Abstract
This paper uses a novel approach in addressing two puzzles in the field of corporate finance in
China, where government is a major player. In addition to the traditional approach based on agency
theory and information asymmetry, the paper uses the political costs approach in studying the stock
dividend puzzle and rights issues puzzle. The paper finds that the extent of political interference,
managerial entrenchment, and institutional control affects corporate financing choices and dividend
distribution decisions. The result sheds new light on improving the important corporate governance
aspects of state enterprise reform in China.
D 2004 Elsevier Inc. All rights reserved.
JEL classification: G32; G35; O53; P21

Keywords: Corporate finance; Dividend policy; Agency costs; Political costs; State enterprise reform
1. Introduction
The Chinese Communist government, wishing to avoid the political and economic
turmoil that accompanied the mass privatization of the former Soviet Union and other
Eastern European governments, has chosen, as a cornerstone of its political survival, the
commercialization and partial privatization of claims over assets and profits and of its
* Tel.: +86 20 8522 4798.

E-mail address: tdsu@jnu.edu.cn.
1043-951X/$ - see front matter D 2004 Elsevier Inc. All rights reserved.
doi:10.1016/j.chieco.2004.09.003
D. Su / China Economic Review 16 (2005) 118–148 119
state-owned enterprises (SOEs). Its SOE reform strategy hinges on the modern enterprise
system (MES) characterized by the separation of ownership and control. Ownership of an
SOE’s assets is distributed among the government, institutional investors, managers,
employees, and private investors. Effective control rights are assigned to the management,
which generally has a very small, or even nonexistent, ownership stake.
The separation of ownership and control creates a conflict of interest between
management and newly enfranchised private investors that is the root cause of the
principal-agent problem in traditional corporate finance theory (Jensen & Meckling,
1976). Moreover, because government desires to retain some control, in part, through
partial retained ownership of commercialized SOEs, further conflicts arise between
politicians and firms (Shleifer & Vishny, 1994). In addition to agency costs of managerial
discretion, the newly corporatized SOEs face political costs of government control,
defined as the reduction in firm value due to government administrative interference. The
games played by the government, management, and outside investors become more
complex than those addressed in the traditional corporate finance models. The principle
challenge of this paper is to assess whether and to what extent the unique ownership
conflicts under the MES can serve as a foundation for solving two puzzles in corporate
China.
While stockholders respond positively to stock dividend announcements in the United
States, which can be explained by the signaling theory, stockholders react negatively to
stock dividend distributions in China. If stock dividend does not signal good news, why do
so many Chinese firms bother to distribute stock dividends? While fewer and fewer firms
issue uninsured rights in the United States and elsewhere, Chinese firms predominantly
use uninsured rights in seasoned equity offerings (SEOs). Models of information
asymmetry argue that firms issuing uninsured rights are of better quality, but Chinese
firms using uninsured rights exhibit no evidence of superior investment opportunities and
experience significant drop in their stock prices. The main contribution of this research is
to analyze these apparent deviations from the prediction of the traditional corporate
finance theory in terms of political and agency costs of the government–management–
investor conflicts unique to the state enterprise reform in China. The papers find that
dividend policies and financing choices are affected by ownership conflicts (agency and
political costs) and the effectiveness of monitoring.
The rest of the paper is organized as follows: Section 2 discusses how earlier attempts
to commercialize SOEs without privatizing them created adverse incentives that defeated
the intent to increase profitability and reduce the need for continual government subsidies.
This failure led to the establishment of the MES, with its unique ownership conflicts and
lack of effective monitoring. Section 3 contains theoretical discussions on the dividend
policies and post-IPO equity financing choices of Chinese firms. It explores the rationale
behind the frequent use of stock dividends and the predominant use of uninsured rights in
seasoned equity offerings in China. Three testable hypotheses are then formulated. Section
4 empirically investigates the stock dividend and rights issues puzzles, discusses the long-
term performance of dividend distribution firms, and explains the negative valuation effect
of stock dividends and rights issues announcements in terms of firm characteristics
associated with agency and political costs. Section 5 concludes with a summary of
findings.
120 D. Su / China Economic Review 16 (2005) 118–148
2. Institutional framework surrounding SOE reform
The reform of SOEs in China has been characterized by efforts to decentralize,

commercialize, and, of necessity, to separate ownership and (effective) control. In general,
the government is reluctant to initiate deep institutional reform and is loath to surrender its
control over enterprises because doing so increases the cost of maintaining political
support in the form of attaining worker satisfaction and a bquietQ population. At the same
time, the inefficiencies and waste inherent in socialist economies, in practice, limit the
government’s ability to maintain its base without resorting to totalitarian controls that have
proven impossible to sustain in a few societies. Therefore, the privatization of SOEs has
been adopted only because it is the only viable means to assure the survival of the existing
political order, which, in the case of China, means the survival of the Chinese Communist
Party. The government, by reforming the agricultural industry, restructuring SOEs via
partial privatization, and encouraging rapid growth of Township and Village Enterprises,
seeks to obtain the benefits of a market economy for the mobilization of private savings,
economic growth and an improved standard of living, while retaining political control.
SOE reform began with the Management Responsibility Contract System (MRCS) in
1987, in which the government transferred management authority to the enterprises and
allowed them to retain some of their profits.1 Under the MRCS, managers of SOEs were
given specific control rights in production, investment, sales, profits, personnel manage-
ment, and distribution of fringe benefits via contracts, therefore, managers have partial
incentives to generate cash flows and expand wages in the form of bonuses and fringe
benefits. However, the contractual relationship between managers and the government was
asymmetric and incomplete. A critical defect was that profit retention by management had
no downside. The state, in fact, remained responsible for final losses, mainly through the
state-owned banks, which were not contractual parties in the enterprise-commercialization
process. Employee bonuses and welfare expenditures, including housing, medical and
schooling expenses, and other fringe benefits, were deductible from enterprise revenue in
calculating profits. Therefore, it is not difficult to see why profit maximization was not the
management’s primary objective and why SOEs continued to function as government
agencies or social security institutions providing bcradle-to-graveQ services to employees
rather than as even approximately profit-maximizing businesses.
Grossman and Hart (1982, 1986) and Hart and Moore (1990) argue that residual rights
of control, not specific rights via contracts, are critical determinants of managerial
incentives. Shleifer and Vishny (1994) also argue that commercialization without
allocation of residual rights to management is unlikely to guarantee more productive or
profitable uses of resources. Moreover, when the government can indirectly subsidize the
management to maintain bloated employment levels, employee housing, schooling, and
other political objectives through control over transfers from the treasury, thus creating a
soft budget constraint, privatization alone is unlikely to enhance productivity. Under
MRCS, managers had little residual rights of control and were subject to strong political
influences to maintain SOE’s social responsibilities. Consequently, they had little interest
1
SOEs had to operate at state-controlled prices and remit all profits to the state before MRCS.
in technology investment, profit maximization, and efficient resource allocation, all of

which inevitably led to high political costs of government control.
Groves, Hong, McMillan, and Naughton (1994) provide evidence that MRCS
strengthened workers’ incentives and increased productivity and workers’ compensation,
but neither reduced government subsidies nor increased overall profits. With the adverse
incentive structure as described above, SOEs were able to rely on easy bank credit to pay
production and other boperatingQ expenses. Thus, as of 1994, over 40% of SOEs were
unprofitable, and while SOEs accounted for 34% of GDP, they absorbed three quarters of
domestic credit. Their ever-increasing triangular debt (unpaid bills between state
enterprises, state banks, and the government) had accumulated to over 4900 billion
Renminbi Yuan (or 95% of GDP), and bad debt was estimated to be as much as 25% of
state bank assets in 1996. Many troubled SOEs were unable to pay back these accrued
loans, thus threatening the collapse of China’s banking system. MRCS did not achieve the
goal of effectively reforming SOEs and was terminated nationwide in 1994. The process of
partial privatization through the creation of joint-stock companies (corporatization) began
through the establishment of Modern Enterprise System (MES), or Cooperative Share-
holding System (CSS).2
Under the MES, there are five types of shares. Those shares not retained by the
government, state enterprises, managers, and employees are transferred to outside
investors through initial public offerings (IPOs) and seasoned equity offerings (SEOs).3
Each share type is entitled to the same voting rights and dividend. After corporatization,
all shareholders have residual claims on the company’s assets. Boards of directors are
established to represent the interests of all shareholders. A major characteristic of Chinese
joint-stock companies is that share ownership is dominated by the state or legal person
shareholders. The State Assets Management Bureau (SAMB) retains majority ownership
for about one-third of the listed firms, and legal person shareholders have controlling
shares for the other two-thirds.
To facilitate the privatization process and to wean doomed firms from eternal subsidies,
the government set up an annual reserve fund to support mergers and bankruptcies, to train
surplus workers for new jobs, to free enterprises from some of their responsibility to
provide social services, and thus, to promote a system in which enterprise survival depends
only on market-oriented performance. By the end of 1996, 675 SOEs had declared
bankruptcy, 1801 had been merged or downsized, 9200 SOEs had been corporatized, and
2
In November 1993, the 14th Central Communist Party of China issued the Decision on Issues Concerning the
Establishment of a Socialist Market Economic Structure, which formally introduced the modern corporate system
to the SOEs. The major reform objectives in the system include transforming the SOEs into corporations by the
separation of ownership and control and the establishment of efficient corporate governance structure.
3
Among the five share types, state shares are retained by the State Assets Management Bureau (SAMB) for
central government or local government. Legal person shares are held by other SOEs and nonbank financial
institutions (including investment companies, finance corporations, and mutual funds). Employee shares are those
sold to managers and workers within a company and cannot be traded within the first 3 years. Domestic individual
shares, or A shares, are those held and traded by private Chinese citizens in the two official exchanges in China.
Foreign shares are those held and traded by foreign investors in security exchanges in China (B shares), in Hong
Kong (H shares), or in NYSE (N shares).
of these, 485 of the most profitable had been listed in the two stock exchanges with a total
market capitalization of about 400 billion Renminbi Yuan (equivalent to approximately
U.S. $50 billion).
The extent to which SOE partial privatization leads to benefits in the form of more rapid
productivity and GDP growth is limited by the political- and agency-cost problems. The
games played by government, management, and investors become more and more
complex than those addressed in traditional corporate finance theory.
In traditional corporate finance theory, private firms are viewed as a nexus of contracts
between various economic entities, particularly owners (principals) and managers (agents).
In the absence of complete and fully enforceable contracts, principal-agent problems arise
because owners want to maximize firm value while managers maximize their own utility,
which, in general, reduces the value of the firm. When the government is also an actor, the
traditional agency approach will not be the whole story because the government is
interested in its survival, which may require the use of resources that might otherwise
contribute to the management’s utility or the private investors’ wealth (firm value). These
interrelationships are complicated by the fact that the government is by no means
monolithic and constitutes a collection of agents with conflicting goals: politicians, civil
employees, and citizens themselves (Hart, 1995). An exacerbating characteristic of the
Chinese situation is that the vast majority of private shares are held by small investors.
Dispersed public ownership creates a classic free-rider problem: Small shareholders do not
have the incentive to monitor management because the benefit is enjoyed by all while the
cost is totally borne by a few, active investors.
Several factors can mitigate the agency costs of managerial discretion, including debt
(in conjunction with appropriate bankruptcy procedures), takeover threats, legal protection
of investors, product market competition, etc. Jensen (1986) and Stulz (1990) show that
debt with the threat of bankruptcy imposes a hard budget constraint on managers and
limits the management’s control over firm’s free cash flows. When managers are faced
with choices of reducing empire building and lavish perquisites or going bankrupt, they
are likely to choose the first option. Therefore, leverage reduces agency costs of
managerial discretion. Shleifer and Vishny (1986) show that takeover threats discipline the
management in the presence of many small shareholders. The reason is that if a company
is badly managed, then there is an incentive for someone to acquire a large stake of the
company, improve its performance, and profit on the shares purchased. The threat of such
action can persuade the management to act in the interest of shareholders. Weisbach
(1988) shows that the level of investor protection, including the extent of board
independence, the ability of a board to dismiss managers following poor performance, and
the degree of shareholder activism, is negatively related to agency costs and positively
related to firm performance. Maksimovic and Titman (1991) argue that product market
competition makes profits more sensitive to managerial effort, reduces agency costs, and
enhances investment efficiency. The reason is that temptation to retain cash and engage in
less productive activities is less severe for firms in a more competitive environment.
Unfortunately, because of the political nature of the privatization process itself, bank-
debt financing of corporate investments is not a viable means to assure a hard budget
constraint in transitional economies, such as China (Shleifer & Vishny, 1994). Bank loans
have traditionally been viewed as grants from the state designed to bail out failing firms.
State-owned banks retain a monopoly in the Chinese banking sector, and profit is not their
overriding objective. If political favor is deemed appropriate, subsidized loans,
rescheduling of overdue debt, or even outright transfer of funds can be arranged with
SOEs (soft budget constraints). A banking system that is so heavily controlled by the
government’s political objectives can hardly exert any real influence on management.
Moreover, a market for private, nonbank debt has yet to be established. There is no active
merger or takeover activity in stock markets to discipline management. Information
available in the capital markets is insufficient to keep at arm’s length of the managerial
decisions. Despite the existence of a bankruptcy law since 1986, the rate of bankruptcies
has remained at a very low level. In light of the above peculiarities, direct control by
(large) shareholders through the board of directors appears to be the only means for
shareholders to monitor managers.
With good corporate governance structure, boards of directors would consist of
delegates of large institutional shareholders and minority investors, representing all
shareholders’ interests. Such boards would make critical investment and reorganization
decisions, select top managers, and provide managers with incentives and compensation
appropriate to the investors’ interests. However, board members in Chinese joint-stock
companies consist mainly of representatives or officials from the SAMB, local govern-
ment, and other state enterprises, whose interests are not the same as those of outside
investors. On average, 90% of the board members are government officials and delegates
of other state enterprises.4 Moreover, there is, at best, an inadequately functioning
managerial labor market that can discipline managers and help solve incentive problems
caused by the separation of ownership and control. All top managers must be approved by
the Organization Department of the local or central government. The government
maintains ultimate rights for changing managerial positions.5 With state-controlled stock
companies, political considerations still dominate economic performance in the selection
of managers and adversely affect managerial incentives. Managers, no doubt, care more
about carrying out the wishes of the Party and the government, such as avoiding worker
layoffs and maintaining some level of worker social security, than about the concerns of
shareholders. As a result, many management decisions still reflect the old political
concerns (political costs). In addition, managerial and board member compensation
4
For example, in a report in Financial Times, Lin (1999) highlights the continued government interference in
investment decisions of PetroChina, which has just gone public: bGiven the oil industry’s strategic importance to
the Chinese economy, it is particularly likely that political factors will play a major role in the future governance
of PetroChina. In its role as controlling shareholder, the Chinese government will face many potential conflicts of
interest inherent in its role as owner, regulator, tax authority and primary borrower. PetroChina’s board of
directors appears not to be structured so as to act as a check on the power and interests of the controlling
shareholder and the Chinese government. The new board of directors will have 12 members, all have close ties
with the governmentQ.
5
For example, on January 19, 2000, the Wall Street Journal reported that Li Yizhong, the President of Sinopec,
China’s second major oil company listed in Shanghai Securities Exchange, had been removed from his post and
appointed as the communist party secretary of the Gansu province in northwest China. The Journal commented:
bThe managerial overhaul sends a confusing signal to the oil industry. The moves appear to be motivated by
domestic political considerations and are seen as part of the government’s strategy to accelerate the development
of northwestern ChinaQ.
packages are determined by the government and are seldom directly linked to profits.
Management is not severely constrained from self-dealing behavior, such as diverting SOE
assets outside the state sector for private use, such as charging meals and travel expenses to
company accounts, obtaining large apartments, and purchasing fancy cars and furniture for
personal consumption. Spending cash flow on acquisitions or other unprofitable projects
undertaken with the aid of unrealistically high forecasts of future profitability gives the
management a bigger company to run, thereby increasing their power and prestige in the
community (agency costs).
The allocation of effective control rights to management without clearly defined
corporate laws and bankruptcy procedures gives managers a chance to enrich themselves
by such means as undervaluing state assets and spinning off subsidiaries to be controlled
by the manager’s relatives. In a study conducted by Kernen (1997), some managers of
SOEs admitted to have created a number of collective or private enterprises for their
relatives using SOEs’ assets. Furthermore, China’s embryonic regulatory regime and
rudimentary accounting standards provide very little protection for investors’ rights
because they limit the boards’ ability to monitor abuses of governmental influence and
managerial discretion and thus limit political and agency costs that diminish the
productivity-enhancing effects of privatization and reduce firm value.
It should be noted, however, that agency and political costs do not imply that
managers are totally insensitive to the performance of their firms. Boycko, Shleifer, and
Vishny (1996) argue that managers are more concerned with profits than politicians are.
Bai, Li, Tao, and Wang (2000) posit that, during transition, SOEs are charged with the
multitasks of social welfare provision and efficient production. One implication is that
SOEs are given low profit incentive, and the government’s concern for income
distribution and wage regulation is the root cause for the imperfection in managerial
market. Another implication is that, depending on the speed of emergence of supporting
institutions, managers may be able to escape government monitoring and divert efforts
from providing social services to seeking higher profits. Chang and Wong (2003) find
that the decision-making power of local Party committees relative to the largest
shareholders is positively associated with firm performance, indicating that political
control may help improve firm performance through the mitigation of large shareholders’
agency costs. On the other hand, the decision-making power of local Party committees
relative to managers is negatively associated with firm performance, suggesting that the
existing level of political control is excessive when viewed from the perspective of
managers. On net, they find that reducing political control tends to improve the
performance of listed firms.
In summary, the partial privatization and separation of ownership and control have
transferred residual rights of control from government to managers but have not provided
effective monitoring mechanisms. The traditional government goals characterized by the
biron rice bowlQ concept still exert some control through the selection of managers and
board members. Profit is not the overriding objective for the managers, and the
management has considerable leeway to seek its own self-interest at the expense of
shareholders. In the next two sections, we present theories and empirical evidence bearing
on the influence of agency and political costs revealed in the dividend distributions and
corporate financing choices of Chinese joint-stock companies.
3. Dividend policy and equity financing choices
3.1. Agency costs, political costs, and stock dividend puzzle
The separation of ownership from control inevitably leads to agency problem among all
Chinese SOEs. Dividends can play a role in reducing agency costs because they remove
corporate wealth from insider control. Jensen (1986) suggests that managers, motivated by
compensation and human capital considerations, have incentives to overinvest free cash
flows, even in the absence of profitable growth opportunities. Dividend payout policy can
be a vehicle for monitoring the use of funds by the management. Thus, the observed positive
stock-market reaction on dividend increases is consistent with a reduction in agency costs.
La Porta, Lopez-DeSilanes, Shleifer, and Vishny (2000) also argue that corporate
dividend policies reveal agency problems and vary across legal regimes, with different
levels of shareholder rights protection. Under an effective corporate governance system,
shareholders have the legal power to prevent insiders (managers) from using a large
proportion of the company’s free cash to benefit themselves. This implies that cash
dividends are an outcome of less serious agency conflicts. Cash dividends are paid because
shareholders are able to pressure corporate insiders to disgorge cash.
Lu and Wang (1999) and Wei (2000) find that both cash and stock dividends are
important forms of dividend distribution in China. This empirical result is puzzling: While
fewer and fewer firms distribute stock dividends in the United States, why are stock
dividends widely used in China?
Taken at their face value, stock dividends are small-scale stock splits and bfiner slicing
of a given cake—the total market value of a firmQ. They do not generate funds for the firm
nor alter the proportional ownership of the firm on the part of the existing shareholders.
But if stock dividends are merely cosmetic changes in a firm’s balance sheet, why do a lot
of firms continue to engage in such financial manipulation, particularly when there are
costs of doing so? Grinblatt, Masulis, and Titman (1984) propose the bsignaling
hypothesisQ (or bretained earnings hypothesisQ). They argue that accounting principles
require that stock dividend distributions be accompanied by a reduction in retained
earnings in the balance sheet. In the presence of information asymmetry, i.e., when
managers (insiders) know more about the future prospects of the firm than investors
(outsiders) do, managers of high-quality firms can use stock dividend distributions to
convey favorable information to the investors. These managers can afford to signal
because they do not expect the reduction in the balance of retained earnings to constrain
future cash dividend payments. On the other hand, managers who anticipate poor future
earnings will find it costly to mimic the signal of a high-quality firm. An empirical
implication is that stock prices, on average, react positively to stock dividend announce-
ments, ceteris paribus. Grinblatt et al. (1984) document significantly positive abnormal
returns on and around the announcement day of stock dividends. Their results are further
substantiated in a recent article by Rankine and Stice (1997), who find additional evidence
of positive relationship between abnormal returns around stock dividend distributions and
subsequent earnings growth.
In China, information asymmetry may not be the entire story because dividend policy is
severely influenced by agency problem and political interference. Managers of a number
of firms own a very small percentage of shares. They have virtually no incentive to
increase the stock prices of their firms. In addition, institutional investors have little voice
in some of the boards, bank debt does not provide monitoring on management, the market
for public debt is yet to be developed, and there is no active merger or takeover activity in
the stock market to discipline managers. Moreover, in the process of privatizing SOEs, the
government cannot withhold the temptation to control managers. Political interference by
the government leads to political costs of government failure. Under these circumstances,
managers could and would hold on cash as much as possible for the purpose of covering
their own mistakes, consuming perks, doing favors for politicians, or having financial
flexibility. Because cash dividends reduce the free cash level in the firm and add
disciplines to the management, managers will try to avoid distributing cash dividends to
shareholders as much as possible.
Therefore, we propose the bagency/political cost hypothesisQ—dividend policies in
corporate China are direct consequences of ownership conflicts and monitoring (as
opposed to consequences of differential investment opportunities).6 Firms with very high
agency and political costs and without effective monitoring have neither the pressure nor
the incentive to distribute any dividend. On the contrary, firms with low agency and
political costs and effective institutional monitoring are more likely to pay cash dividends.
Firms with mild agency and political costs may use stock dividend as a substitute for cash
dividend. They are willing to incur costs associated with stock dividends because they
have the incentive to separate themselves from the worst firms (firms plagued with high
agency/political costs).7
The alternative to the agency/political costs hypothesis is the bliquidity/investment
opportunities hypothesisQ. Because firms with good investment opportunities have less
free cash at hand, they would distribute stock dividend or no dividend rather than paying
cash dividend. Furthermore, firms with the best investment opportunities will distribute
stock dividends to separate themselves from those that pay no dividends.
Therefore, we hypothesize that:
H3.1. Managers will have the incentive to distribute cash dividends when the agency and
political costs are small. Between the choices of cash dividend and no dividend, or
between the choices of cash dividend and stock dividend, the probability that a firm will
distribute cash dividend is positively related to management and institutional ownership
and negatively related to government ownership and nonessential expenditures.
6
We acknowledge that the impacts of agency and political costs on firm’s dividend policies and financing
decisions can be quite different. However, we do not delineate agency costs from political costs in our
econometric investigations because we do not have good enough proxies. We plan to conduct further research on
this topic in the future.
7
The reason that stock dividends can also serve as credible signals is as follows: Given information asymmetry
between managers and investors, stock dividends are costly signals that convey the management’s private
information about the firm’s level of agency/political costs. Specifically, in China, firms must maintain a
minimum level of retained earnings to qualify for future share issues. For instance, prior to 1998, the CSRC
requires firms to have at least 10% return on assets for three consecutive years to issue new shares. Therefore,
managers would transfer retained earnings to common stock (and issue free shares) only if they expect future
earnings to increase and do not need to keep large amount of free cash. Investors, therefore, may interpret stock
dividends as good news, although not as good as they do for cash dividends.
The alternative hypothesis is that managers issue cash dividends because they do not
have very profitable net present value projects. The probability of cash dividend
distribution is negatively related to the firm’s investment opportunities, such as market-to-
book ratios, and long-term investment to total asset ratios, and positively related to its
liquidity measured by leverage ratios.
H3.2. Firms with mild agency/political costs distribute stock dividends (compared with
doing nothing) to separate themselves from firms with high agency/political costs.
Between the choices of stock dividend and no dividend, the probability that a firm will
distribute stock dividend is positively related to management and institutional ownership
and negatively related to government ownership and nonessential expenditures.
The alternative hypothesis is that stock dividends are signals of better investment
opportunities. The probability of stock dividend distributions is positively related to the
firm’s investment opportunities, but negatively related to its liquidity.
3.2. Agency costs, political costs and rights issues puzzle
Myers and Majluf (1984) argue that when managers know more about the future
prospects of the firm than outside investors do, managers may convey positive news to the
investors by means of the financing method chosen, with internal financing being the
highest in the pecking order, riskless debt the next highest, followed by equity issues (the
so-called bpecking order paradigmQ). Because new bank debt is likely to be a signal of bad
news in the Chinese situation (and in transitional economies with soft budget constraints,
in general), and because markets for private nonbank debt have not yet emerged, loan
financing of new, profitable investment projects does not appear in the pecking order for
newly privatized SOEs. Consistent with this argument, we observe no corporate debt issues
over the period covered in this study. Therefore, we focus on equity financing choices.
When a firm wishes to raise capital through outside equities, it can choose a number of
flotation methods, including uninsured rights issues, rights issues with standby underwriters,
and fully underwritten offers. The direct flotation costs (e.g., registration fees, mailing costs,
and underwriting fees) are generally smaller for rights issues than they are for firm
commitment underwritten offers. However, managers have to set a sufficiently low
subscription price to guarantee the full subscription of rights, which may lead to
prohibitively high indirect flotation costs for rights issues. When choosing a flotation
method in the absence of information asymmetry, managers typically weigh the costs of
underwriter certification and the costs of a rights offer subscription discount. Under
information asymmetry, the true value of a firm is unobservable to the market. Heinkel and
Schwartz (1986) show that managers who know of profitable investment projects and are
willing to incur investigation costs will ensure the success of their rights issues with
underwriter’s standby offer; managers who have profitable investment projects but are
unwilling to incur additional investigation costs will signal their true value in the choice of a
subscription price in an uninsured rights offer; while managers with lower-quality projects
make fully underwritten offers. Their signaling model implies that the quality of a firm is
revealed through the financing method chosen, with firms using standby rights being the
highest, those using uninsured rights the next, and those using firm commitment the lowest.
Eckbo and Masulis (1992) argue that issuers expecting low shareholder take-up of
rights offers will turn to underwriters for quality certification to avoid the large adverse-
selection costs of a deep subscription discount. As a result, a deep discount in an uninsured
rights issue signals more positive information about firm quality than standby rights do
and should be followed by significantly positive announcement day abnormal returns.
Bohren, Eckbo, and Michalsen (1997) find empirical evidence consistent with the adverse-
selection argument using data on rights offers on the Norwegian stock market.
One common assumption among models of equity financing choices under information
asymmetry is that managers of an issuing firm maximize the value of the firm, which may
not be valid in the Chinese case. In fact, a number of studies have shown that Chinese
firms predominantly use uninsured rights in floating SEOs and that their announcements
are, on average, followed by significantly negative market reactions (Han & Li, 2002;
Yuan, 2003; Zhang & Wang, 2001). Therefore, we conjecture the possibility that, because
of high agency and political costs, uninsured rights offerings in China (having low direct
flotation costs but large indirect costs of share depreciation) represent an attempt by the
management to exploit existing shareholders. That is, uninsured rights are not the outcome
of firms’ value-maximization behavior, but rather a result of ownership conflicts and low
monitoring. Managers of firms that are subject to low agency and political costs will act
upon the shareholders’ interests. When these managers foresee profitable investment
projects but are unable to use internal financing, they will maximize the value of the firm
by choosing the least expensive (in terms of direct and indirect costs) flotation method,
i.e., firm commitment underwritten offers under the Chinese situation. We hypothesize that
H3.3. In the existence of agency conflicts and political interference, equity financing
choices reveal the extent of agency and political problems and the effectiveness of
shareholder monitoring. The probability that a firm will issue uninsured rights offers is
negatively related to managerial and institutional ownership and positively related to
government ownership and nonessential expenditures.
The alternative to the agency/political costs hypothesis is the binvestment opportunities
hypothesisQ—that firms employ underwriters for public offerings of seasoned equities
because they have more profitable investment projects than do those that issue uninsured
rights. Therefore, the probability that a firm will use rights offer is negatively related to
proxies of investment opportunities but positively related to leverage ratios.
In the next section, we discuss empirical irregularities surrounding announcements of
dividend distributions and SEOs and conduct empirical tests of the agency/political costs
null against the investment opportunities alternative.
4. Empirical evidence
4.1. Data
We examine dividend policies and post-IPO equity financing choices for a sample of
508 SOEs that went public before December 31, 1996. Over the period from January 1996
through July 2000, 377 of these firms recorded a total of 1095 annual dividend
announcements, including 513 stock dividends and 582 cash dividends.8 Each cash
dividend announcement is compared with the previous dividend (if there is any) and is
then classified as an increase, a decrease, or a no-change in dividends. Forty-three cash
dividend announcements that have no previous dividend announcements for comparison
are eliminated from the sample. Of the total 539 cash dividend announcements remained
in the sample, 226 are dividend increases, 191 are dividend decreases, and 122 have no
changes. The sample classification is detailed in Table 1.
Over the period from January 1996 through July 2000, 348 of our sample firms
recorded a total of 547 announcements of seasoned equity offerings (SEOs) after their
IPOs, including announcements of 469 uninsured rights issues and 78 firm commitment
underwritten offers.9 The sample classification presented in Table 2 is striking. While there
is a trend toward using underwriter certification in issuing SEOs across countries (Bohren
et al., 1997), uninsured rights continue to be very popular (representing about 87% of all
SEOs) in corporate China.
We use the following variables to proxy agency costs, political costs, liquidity, and
investment opportunities:
Because managers will be more likely to strip assets and transfer resources from SOEs
to themselves if the managers’ fractional investment is small, we use the number of shares
held by managers and board members as a fraction of total shares outstanding
(MANAGER) as one of the proxies for agency costs of managerial discretion.
Because large shareholders may have more incentive to monitor management, which
may, in turn, reduce agency costs, we use the percentage of shares held by the top 10
shareholders (TOPTEN), including institutional and private investors but excluding state
government, as the second proxy for agency costs of managerial discretion.
With large government ownership and heavy presence of government officials in the
board, managers are more likely to try to please politicians by maintaining high fringe
benefits and other social responsibility programs rather than to maximize firm’s residuals
(Chang & Wong, 2003). We use the percentage of shares retained by the local, state, and
8
There are three principal reasons for choosing the January 1996 to July 2000 sample period. First, information
on corporate fundamentals, such as earnings and dividends, may not be reliable in the early stage of development
of an emerging market. Second, the traditional Chinese accounting system was brought from the former Soviet
Union in early 1950s for the highly centralized economy and was quite different from the generally accepted
accounting principles (GAAP) in the market economies. Different industries have different accounting regulations
enforced by their own ministries. On July 1, 1993, the new bEnterprise Accounting StandardsQ and bEnterprise
Financial Accounting PrinciplesQ came into effect and brought the Chinese accounting practices more close to the
international standards. The January 1996 cut-off allows us to compute financial variables from the new
standardized financial statements. Third, the stock price data end in July 2002. To calculate buy-and-hold returns
over a 2-year period, the July 2000 cut-off is necessary. All data are compiled from China Securities Market and
Accounting Research (CSMAR) data base produced by Shenzhen Guo Tai An Information and verified with
information from www.cninfo.com.
9
We note that a Chinese joint-stock company, subject to certain regulations and approval, can issue equities in
the form of A shares (available only to domestic investors and traded in Chinese exchanges), B shares (available
only to foreign investors and traded in Chinese exchanges), H shares (listed and traded in Hong Kong), N shares
(listed and traded in New York), or a combination of these share types. Because of the small number of
observations associated with the issuance of B-, H-, and N-shares through alternative flotation methods, we focus
only on A-share decisions in this paper.
Table 1
Sample classification for dividend distributions
Year Types of distribution
Stock dividend Cash dividend All
Increases No changes Decreases
1996 121 30 45 26 222
1997 98 52 41 38 229
1998 135 74 28 19 256
1999 104 53 54 28 239
2000 (first half) 55 17 23 11 106
Total 513 226 191 122 1052
The sample consists of 508 SOEs that went public before December 31, 1996. Over the period from January 1996
through July 2000, 377 of these firms recorded a total of 1095 annual dividend announcements including 513
stock dividends and 582 cash dividends. Each cash dividend announcement is compared with the previous
dividend (if there is any) and is then classified as an increase, a decrease, or a no change in dividends. Forty-three
cash dividend announcements that have no previous dividend announcements for comparison are eliminated. Of
the total 539 cash dividend announcements remained in the sample, 226 are dividend increases, 191 are dividend
decreases, and 122 have no changes.
central government (GOVNT) as a proxy for the political costs of government interference
on enterprises.
We use welfare expense ratio (WELFARE) as a common proxy for agency and political
costs.10 WELFARE measures the extent of perquisite consumption, employee welfare
expenses (such as housing subsidies and schooling), and other nonessential expenses.
Welfare expenses provide managers opportunities to divert firm’s assets to their private use
and to politicians’ social responsibility programs.
We use two proxies for a firm’s investment opportunities: market-to-book ratio (MTB),
calculated as the average market value of equity 3 months prior to the event day plus book
value of assets in the previous year minus book value of equity in the previous year
divided by the book value of total assets in the previous year, and long-term investment to
total assets ratio (LONGINV).
We use two proxies for a firm’s liquidity: cash flow to total assets ratio (CASHTA),
defined as the operating income before depreciation minus total taxes adjusted for changes
in deferred taxes minus gross interest expenses and dividends and divided by total assets,
and leverage (LEVERAGE), defined as the ratio of long-term debt and the book value of
total assets.
4.2. The determinants of dividend distribution decisions
In the spirit of Grinblatt et al. (1984), we compare the valuation effect of stock
dividends and cash dividends on and around the day of the China Security Daily
announcement of dividend distributions. If shareholders react positively to the announce-
10
Welfare expenses are total expenses minus costs of goods sold, wage expenses, and interest expenses. Welfare
expense is listed as a separate item on the balance sheet for a Chinese firm. Welfare expense ratio is welfare
expenses divided by total expenses.
Table 2
Sample classification for corporate financing choices
Year Types of offering
Uninsured rights Underwritten All
1996 82 11 93
1997 66 24 90
1998 114 16 130
1999 159 18 177
2000 (first half) 48 9 57
Total 469 78 547
The sample consists of 508 SOEs that went public before December 31, 1995. Over the period from January 1996
through July 2000, 348 of these firms recorded a total of 547 announcements of seasoned equity offerings (SEOs)
after their IPOs, including announcements of 469 uninsured rights issues and 78 firm commitment underwritten
offers.
ments of stock dividends, then the signaling theory can potentially explain the stock
dividend puzzle in China. Appendix A details the method used in computing the abnormal
return (AR) and cumulative abnormal return (CAR) in this paper. Table 3 presents average
AR and CAR on and around the stock dividend and cash dividend distributions.
The examination of the table reveals a striking result. The mean 2-day abnormal return
is significantly negative at the 5% level for stock dividend announcements, significantly
Table 3
Average abnormal and cumulative abnormal returns on and around dividend announcement days
Event day Stock dividends Cash dividends
AR CAR t-statistic Increases No change Decreases
(%) (%)
AR CAR t-statistic AR CAR t-statistic AR CAR t-statistic
(%) (%) (%) (%) (%) (%)
10 0.13 0.13 1.21 0.19 0.19 1.17 0.08 0.08 0.16 0.14 0.14 0.26
9 0.11 0.02 0.23 1.07 0.26 1.29 0.10 0.18 0.69 0.06 0.20 0.42
8 0.25 0.27 1.69 0.21 0.05 0.21 0.12 0.30 1.16 0.11 0.31 1.06
7 0.12 0.15 1.26 0.13 0.18 1.15 0.13 0.17 0.59 0.08 0.39 1.19
6 0.30 0.15 1.29 0.03 0.21 1.22 0.08 0.09 0.24 0.20 0.19 0.38
5 0.39 0.54 2.25 0.08 0.29 1.35 0.11 0.20 0.73 0.07 0.26 0.65
4 0.21 0.75 2.68 0.10 0.39 1.46 0.17 0.37 1.21 0.08 0.18 0.34
3 0.12 0.87 2.87 0.19 0.20 1.21 0.26 0.11 0.32 0.13 0.05 0.12
2 0.16 0.71 2.63 0.13 0.07 0.25 0.08 0.19 0.73 0.04 0.09 0.20
1 0.15 0.56 2.27 0.14 0.21 1.22 0.11 0.30 1.24 0.16 0.07 0.16
0 1.02 0.46 2.16 0.93 1.14 3.55 0.28 0.58 1.56 0.15 0.22 0.51
1 0.18 0.64 2.52 0.27 1.41 3.82 0.19 0.39 1.32 0.06 0.28 0.73
2 0.16 0.80 2.79 0.24 1.17 3.59 0.12 0.51 1.49 0.11 0.17 0.35
3 0.29 1.09 3.46 0.11 1.06 3.47 0.09 0.60 1.61 0.08 0.09 0.21
4 0.18 0.91 2.95 0.26 1.32 3.73 0.16 0.44 1.38 0.10 0.19 0.38
5 0.24 0.67 2.57 0.19 1.13 3.54 0.13 0.31 1.19 0.16 0.03 0.10
6 0.17 0.50 2.20 0.16 1.29 3.68 0.14 0.45 1.40 0.04 0.01 0.06
7 0.10 0.40 2.01 0.20 1.49 3.90 0.16 0.29 1.22 0.14 0.13 0.21
8 0.15 0.55 2.26 0.25 1.24 3.62 0.12 0.41 1.35 0.09 0.22 0.52
9 0.18 0.37 1.95 0.08 1.16 3.57 0.17 0.24 0.78 0.12 0.10 0.26
10 0.11 0.26 1.65 0.12 1.04 3.44 0.05 0.19 0.65 0.13 0.23 0.56
positive at the 5% level for dividend increases, insignificantly positive for no-change in
dividends, and insignificantly negative for dividend reductions. Fig. 1 portrays the
behavior of the daily average cumulative abnormal returns around dividend announce-
ments. A large and pronounced decline in CAR is observed for stock dividends, and a
large and pronounced increase is observed for cash dividend increases during the event
window. On average, Chinese investors react negatively to stock dividend distributions but
positively to cash dividend distributions. Therefore, the bsignaling/retained earning
hypothesisQ of Grinblatt et al. (1984) cannot explain the predominant use of stock
dividends in China. This begs the question: If stock dividends cannot signal good news on
the future prospects, why do firms in China bother to distribute stock dividends?
To shed some light on the answer to this question, we first examine sample statistics for
proxies of firms’ agency and political costs, liquidity, and investment opportunities. The
results are presented in Table 4. Two important regularities emerge from the table: (1)
Firms distributing cash dividends have, on average, larger managerial and institutional
ownership of shares, smaller fractional government ownership, and less nonessential
expense ratio than do firms issuing stock dividends or no dividend. (2) Firms distributing
stock dividends appear to have, on average, larger managerial and institutional share
ownership, smaller government control, and less welfare expenses than do firms that pay
no dividend.
We then proceed to test the agency/political costs versus the liquidity/investment
opportunities hypotheses by estimating multinomial logit regressions predicting whether a
firm will distribute cash dividend, stock dividend, or no dividend. These regressions allow
us to observe important nonlinear effects while simultaneously controlling for a variety of
factors, which are hypothesized to affect corporate dividend policies in China. Table 5
presents empirical results from the multinomial logit regressions. The first two sets of
regressions in the table estimate the determinants of cash dividend distributions, while the
Fig. 1. Cumulative abnormal returns on and around the dividend announcement day.
Table 4
Sample statistics for variables used in studying corporate dividend policy
Cash dividend (N=539) Stock dividend (N=513) No dividend (N=54)
Mean S.D. Median S.D. Mean S.D. Median S.D. Mean S.D. Median S.D.
MANAGER 1.44 1.20 1.53 1.64 0.57 1.25 0.69 1.85 0.10 0.35 0.08 0.21
TOPTEN 24.54 14.15 27.64 19.22 16.25 14.49 17.88 20.07 11.48 10.96 13.10 15.94
GOVNT 13.39 17.21 15.60 20.85 19.24 21.93 18.66 24.64 31.03 19.74 28.41 22.80
WELFARE 1.26 0.85 1.44 2.06 3.35 1.57 2.96 3.55 4.84 1.77 4.29 2.90
MTB 1.68 1.14 2.04 2.96 1.29 0.83 1.55 1.18 1.10 0.76 1.26 1.14
CASHTA 9.27 2.14 11.13 8.62 8.54 2.33 9.30 7.26 8.32 2.09 9.14 7.51
LEVERAGE 10.17 4.35 13.02 8.44 11.30 3.69 10.80 6.58 12.16 5.77 11.04 7.33
LONGINV 10.22 1.72 9.09 5.27 9.07 1.55 8.25 5.29 9.29 2.16 7.92 4.51
MANAGER is the number of shares held by managers and board members divided by the total number of shares
outstanding. TOPTEN is the percentage of shares held by the top 10 shareholders, including institutional and
private investors but excluding the state government. GOVNT is the percentage of shares retained by the state
government. WELFARE is the nonessential expenditures (total expenses minus costs of goods sold, wage
expenses, and interest expenses) as a fraction of total expenses. CASHTA is cash flow to total assets ratio,
calculated as the operating income before depreciation minus total taxes adjusted for changes in deferred taxes
minus gross interest expenses and dividends and divided by total assets. LEVERAGE is the long-term debt as a
fraction of the book value of total assets. MTB is the market-to-book ratio, calculated as the average market value
of equity 3 months prior to the event day plus the book value of assets in the previous year minus book value of
equity in the previous year divided by the book value of total assets in the previous year. LONGINV is long-term
investment expenditure to total asset ratio.
last set of regressions estimates the determinants of stock dividends. To account for
possible multicollinearities among independent variables, each set of regressions is
estimated with and without MTB and LEVERAGE.11
In regression sets 1 and 2, the coefficient estimates for MANAGER are significantly
positive at the 5% and 10% level, respectively, no matter whether MTB and LEVERAGE
are included in the regressions. The coefficient estimates for TOPTEN are significantly
positive at the 5% level in regression set 1 but are only significantly positive at the 10%
when all variables are included in regression set 2. If either MTB or LEVERAGE, or both,
is dropped from the regression, TOPTEN is insignificant, although it has the right sign.
The coefficient estimates for GOVNT and WELFARE are significantly negative, all at the
5% level. These results, by and large, are consistent with the prediction of the bagency/
political costs hypothesisQ. Firms with higher managerial and institutional stock owner-
ship, smaller government control, and lower nonessential expenditures are more likely to
distribute cash dividends. The coefficient estimates for MTB, LONGINV, CASHTA, and
LEVERAGE have the same signs as predicted by the alternative liquidity/investment
opportunities hypothesis, but none is statistically significant. There is little evidence that
firms with better investment opportunities withhold cash dividends.
In regression set 3, the coefficient estimates for MANAGER is significantly positive at
the 10% level when only MTB is omitted or when all variables are included in the
regression. The coefficient estimates for GOVNT are significantly negative at the 5% level
11
Xu and Wang (1999) documents a significantly positive relation between market-to-book ratios and
ownership concentration. Leverage ratios are also suspected to be correlated to both ownership structures and
market-to-book ratios.
134
Table 5
Multinomial logit regression on corporate dividend decisions
(1) Cash dividend vs. no dividend (2) Cash dividend vs. stock dividend (3) Stock dividend vs. no dividend
Intercept 2.261* 1.637* 1.908* 1.162* 2.600* 3.288* 2.039* 1.735* 2.112* 2.590* 2.863* 1.304*
(5.250) (3.966) (4.607) (3.438) (5.861) (6.522) (1.925) (4.193) (4.627) (5.240) (5.903) (3.610)
MANAGER 0.840* 0.868* 1.207* 0.923* 0.720y 0.615y 0.795y 0.866* 0.428 0.613y 0.464 0.715y
D. Su / China Economic Review 16 (2005) 118–148

(2.044) (2.183) (2.642) (2.349) (1.825) (1.870) (1.863) (2.106) (1.388) (1.801) (1.437) (1.842)
TOPTEN 0.507* 0.581* 0.484* 0.410* 0.109 0.241 0.200 0.323y 0.085 0.069 0.121 0.264
(2.182) (2.609) (2.366) (2.204) (1.085) (1.482) (1.399) (1.858) (0.863) (1.082) (1.307) (1.526)
GOVNT 1.307* 1.622* 0.874* 0.791* 1.382* 1.490* 0.925* 0.724* 1.096* 1.507* 1.124* 0.903*
(3.014) (3.960) (2.533) (2.362) (2.961) (3.225) (2.530) (2.141) (2.306) (3.531) (2.807) (2.755)
WELFARE 1.911* 2.240* 1.828* 1.465* 2.019* 1.258* 1.833* 1.503* 0.704 0.925 0.836 1.172y
(2.860) (3.244) (2.715) (2.165) (3.007) (2.013) (2.450) (2.220) (1.169) (1.330) (1.414) (1.874)
MTB 1.119 0.719 1.065 0.834 0.741 0.520
(1.299) (1.189) (1.220) (1.305) (1.144) (0.862)
LONGINV 0.229 0.304 0.105 0.141 0.344 0.120 0.086 0.195 0.240 0.219 0.088 0.163
(0.797) (0.824) (0.400) (0.584) (0.975) (0.522) (0.603) (0.733) (0.755) (0.680) (0.288) (0.659)
CASHTA 0.144 0.285 0.160 0.206 0.315 0.466 0.207 0.293 0.241 0.352 0.262 0.151
(0.650) (1.232) (0.750) (0.932) (1.264) (1.380) (0.965) (1.114) (1.180) (1.277) (0.964) (0.896)
LEVERAGE 0.404 0.247 0.370 0.318 0.325 0.206
(0.922) (0.719) (0.950) (0.802) (0.826) (0.675)
Pseudo R 2 0.220 0.247 0.195 0.163 0.207 0.225 0.210 0.180 0.183 0.196 0.188 0.159
The dependent variable is zero for firms that distribute no dividend in a given year, one for firms that distribute stock dividends, and two for firms that distribute cash
dividends. The independent variables consist of the proportion of shares held by managers and directors (MANAGER), the proportion of shares held by the top 10 largest
shareholders, excluding the government (TOPTEN), the proportion of shares held by the government (GOVNT), the ratio of nonessential expenses and total assets
(WELFARE), market-to-book ratio (MTB), long-term investment expenditure to total asset ratio (LONGINV), cash flow to total assets ratio (CASHTA), and long-term
debt to total asset ratio (LEVERAGE). Figures in parentheses are t-statistics.
* Level of significance: 5%.
y
Level of significance: 10%.
for all regressions, while the coefficient estimates for WELFARE are significant when all
variables are included in the regression. Although the coefficient estimates for TOPTEN
are of the right sign, they are not statistically significant. We also interpret this result as, by
and large, consistent with the prediction of the agency/political costs hypothesis. Firms
with higher insider share ownership, smaller government control, and lower nonessential
expenses are more likely to distribute stock dividends. Again, the coefficient estimates for
MTB, LONGINV, CASHTA, and LEVERAGE have the same sign as predicted by the
liquidity/investment opportunities hypothesis, but are not statistically significant. There is
sufficient evidence that stock dividends are not signals of better investment opportunities,
but rather signals of mild agency problem and political influence in the Chinese case.
4.3. Long-term performance of dividend distribution firms
The agency/political costs hypothesis predicts that firms with low agency and political
costs and effective shareholders’ monitoring will distribute cash dividend, and those with
moderate agency and political costs may use stock dividends in lieu of cash dividends to
separate themselves from firms having high agency and political costs. This indicates that
cash dividend firms will face better long-term stock-market valuation of their shares than
stock dividend firms will.
Under the alternative liquidity/investment opportunities hypothesis, firms that have the
best investment opportunities signal their quality through stock dividends. The short-term
negative market reaction to stock dividends’ announcements simply reflects under-
valuation of firms’ shares by investors. After the firms have successfully implemented
their projects, their stock prices will eventually go up. A reversal of their fortune will be
expected in the market in the long run. Therefore, firms issuing stock dividends will have
better long-term stock performance than will those distributing cash as dividends.
We measure a firm’s long-term stock performance using the average and median buy-
and-hold abnormal return during 6-, 12-, 18-, and 24-month horizons after the dividend
announcement date (BHARs ). Appendix A details the methodology used in computing
BHARs . The null hypothesis of no abnormal buy-and-hold return is tested using
skewness-adjusted t-statistics. A nonparametric Wilcoxon signed-rank test is also
constructed to test the null hypothesis that the median BHARs is zero.
As shown in Table 6, the average and median buy-and-hold abnormal returns are
significantly positive for firms paying cash dividends at the 6-, 18-, and 24-month holding
periods, while they are significantly negative for firms paying stock dividends at the 12-,
18-, and 24-month holding periods. The average and median buy-and-hold abnormal
returns are negative for firms with neither cash dividends nor stock dividends but are only
statistically significant for the 18-month holding period. We view the empirical evidence
described here as consistent with the bagency/political costs hypothesisQ.12
12
According to the agency/political cost hypotheses, firms distributing stock dividends have milder agency/
political costs than do firms distributing no dividends. Table 6 also shows that firms paying no dividends have
negative mean abnormal returns, although they are often not statistically significant and the magnitudes are often
smaller than that for the firms paying stock dividends. These kinds of results do not seem to be totally consistent
with the implication of the agency/political costs theory.
Table 6
Long-term market-based performance of dividend distribution firms
Period Average buy-and-hold abnormal return (%)
(months) No dividend firms Cash dividend firms Stock dividend firms
Mean Median Mean Median Mean Median
6 1.21 (1.09) 1.08 [1.16] 2.93* (3.05) 2.28* [2.12] 1.79 (1.51) 1.39 [1.27]
12 1.42 (1.15) 1.33 [1.18] 1.48 (1.36) 1.19 [1.04] 2.47* (2.75) 1.83y [1.82]
18 2.18* (2.18) 1.80y [1.86] 2.36* (2.58) 1.95y [1.88] 3.12* (3.36) 2.16* [2.37]
24 0.78 (0.91) 0.63 [0.84] 3.29* (3.61) 2.45* [2.31] 2.61* (3.18) 2.08* [2.24]
PN
The
s average
buy-and-hold
abnormal
return (BHAR) for s months is calculated as BHARs ¼ N1 i¼1
s
jt¼1 1 þ ri ; t jt¼1 1 þ rc ; i ; t , where r i ,t is the monthly stock return for announcement firm i and r c,i,t is
the monthly return on control portfolio for firm i. The test statistic for the null hypothesis of no average buy-and-
hold abnormal return (figure in parentheses)pffiffiffiis the skewness-adjusted t -statistic calculated as
adjusted tBHARs ¼ tBHARs þ 3p1ffiffiffi ðt
N BHARs
Þ2 skews þ 6N skews , where t BHARs is the usual t-statistic for BHARs
and skews is the skewness of the BHARt series (t=1,2,. . .,s). The test statistic for the null hypothesis of the
median buy-and-hold abnormal return (figure in brackets) is the nonparametric Wilcoxon signed-rank test
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
statistic, calculated as sign ¼ ðps 0:5Þ= ps ð1 ps ÞN , where p s is the percentage of positive abnormal buy-
and-hold returns during s months.
* Statistically significant at the 5% level.
y
Statistically significant at the 10% level.
Some researchers may be skeptical about the use of market-based performance

measures because Chinese stock markets are often plagued with speculative activities. To
illustrate, at the end of 1999, more than 99% of the participants in the Shanghai Securities
Exchange consisted of small individual investors rather than institutional investors. In
addition, share turnover (defined as trading volume divided by the total number of
outstanding shares, a measure for trading activities) exceeded 400%; that is, on average,
each stock changed hands more than four times per year, which is more than 10 times as
high as that of international markets. Therefore, it is of no surprise that Chinese stock
markets have been described as a bcasinoQ because of the tremendous short-term
speculation that takes place within it (Chang & Wong, 2003). For this reason, we
complement market-based performance measures with accounting-based ratios including
return on asset (ROA), return on equity (ROE), and return on sales (ROS).
Table 7 presents the mean and standard deviation of ROA, ROE, and ROS for cash
dividend, stock dividend, and no-dividend firms in 1997, 1998, and 1999, respectively. As
shown in the table, the mean ROA, ROE, and ROS for cash dividend firms is higher than
those for stock dividend and no-dividend firms each year, suggesting that, on average,
firms paying cash dividends have better performance than do firms issuing stock dividends
or no dividends. However, the mean ROA, ROE, and ROS for stock dividend firms are
similar to those for no-dividend firms. The overall evidence seem to support the view that
dividend policies are value relevant in China.
4.4. The determinants of equity financing decisions
To explore in greater depth the financial characteristics of firms that use rights issues,
we first compare the liquidity and investment opportunities of firms that issue uninsured
Table 7
Accounting-based performance measures for dividend distribution firms
Accounting ratio No-dividend firms Cash dividend firms Stock dividend firms
Mean S.D. Mean S.D. Mean S.D.
ROA1997 0.046 (0.049) 0.060 (0.071) 0.044 (0.058)
ROA1998 0.038 (0.041) 0.065 (0.083) 0.031 (0.042)
ROA1999 0.044 (0.058) 0.055 (0.070) 0.049 (0.066)
ROE1997 0.074 (0.095) 0.082 (0.127) 0.069 (0.106)
ROE1998 0.072 (0.108) 0.086 (0.133) 0.073 (0.115)
ROE1999 0.068 (0.096) 0.080 (0.109) 0.058 (0.098)
ROS1997 0.087 (1.262) 0.104 (0.173) 0.091 (0.188)
ROS1998 0.092 (1.437) 0.115 (0.205) 0.086 (0.167)
ROS1999 0.085 (1.394) 0.108 (0.190) 0.082 (0.159)
Definitions of the performance variables are as follows: ROA (return on assets) is after-tax profits divided by the
book value of total assets. ROE (return on equity) is after-tax profits divided by the book value of shareholders’
equity. ROS (return on sales) is after-tax profits divided by gross sales revenue. Figures in parentheses are
standard deviations.
rights with those that use firm commitment underwritten offers.13 We calculate the average
of CASHTA and LEVERAGE ratios, which represent a firm’s liquidity and debt capacity,
and MTB and LONGINV, which represent its investment opportunities. If a firm issues
new shares in 1996, we include it in computing the financial ratios of 1995, and so on and
so forth.
As shown in Table 8, cash flow as a fraction of total assets is slightly higher for firms
that use underwritten offers than it is for those that issue uninsured rights in every year
from 1995 through 1999, with the gap growing from one percentage point in 1995 to about
two percentage points in 1999. The average long-term debt as a fraction of total assets is
lower for firms that use underwritten offers than it is for those that issue uninsured rights
over the period covered. The market-to-book ratio, a measure of investors’ evaluation of a
firm’s investment opportunities and future profitability, and the expenditures on long-term
investment projects as a fraction of total assets are also higher for firms using underwritten
offers than they are for firms issuing uninsured rights. These comparative statistics are
inconsistent with the predictions of the signaling model.
We then compute the abnormal returns and cumulative abnormal returns on and around
the announcements of SEOs. Table 9 presents average AR and CAR, and Fig. 2 portrays
the behavior of the average CAR during the event window. A large and pronounced
decline in CAR is observed for uninsured rights issues, and a large and pronounced
increase is observed for underwritten offers. In fact, the mean abnormal returns over a 2-
day period that includes the day of the China Security Daily announcement and the day
preceding the announcement for underwritten offers and rights issues (t-statistics) are
4.40% (6.87) and 2.57% (4.36), respectively. On the average, investors react favorably
to firm commitment underwritten offers but negatively to uninsured rights issues. The
results are inconsistent with the predictions of the adverse-selection model.
13
Note that no standby rights issues are observed on Chinese stock markets over the period covered in this
study.
Table 8
Liquidity and investment opportunities for firms classified by SEO flotation choices
Year Uninsured rights Underwritten offers
Mean S.D. Median S.D. Mean S.D. Median S.D.
CASHTA 1995 7.48 5.21 6.95 8.22 8.41 7.20 8.86 10.13
1996 7.33 5.10 7.08 8.29 9.11 7.86 9.39 11.20
1997 9.04 7.25 8.27 11.31 11.25 9.82 10.81 14.25
1998 9.58 6.84 8.92 11.26 11.60 9.00 10.73 13.51
1999 8.80 6.92 8.33 10.10 10.69 7.88 11.05 13.62
LEVERAGE 1995 11.84 6.02 12.07 7.55 8.43 5.31 7.91 5.10
1996 9.16 6.05 8.93 7.14 9.12 5.00 8.66 6.03
1997 9.55 5.39 8.82 5.80 6.18 3.95 5.83 4.41
1998 9.08 5.23 9.63 6.10 8.64 5.27 8.95 6.38
1999 10.14 6.38 10.52 7.33 8.94 6.12 8.50 6.60
MTB 1995 1.09 0.95 1.22 1.60 1.38 1.18 1.26 1.72
1996 1.28 1.04 1.31 1.66 1.48 1.23 1.44 1.70
1997 1.41 1.15 1.37 1.81 1.64 1.20 1.58 1.93
1998 1.31 0.97 1.28 1.70 1.35 1.06 1.39 1.82
1999 1.12 0.88 1.09 1.31 1.26 1.06 1.30 1.44
LONGINV 1995 8.02 2.37 6.91 4.55 10.93 3.60 9.84 6.11
1996 8.28 3.30 7.53 4.31 10.45 4.26 9.36 4.90
1997 9.54 4.02 9.18 5.33 11.93 3.89 11.08 6.00
1998 7.36 2.74 7.19 4.01 9.44 3.65 9.06 5.14
1999 8.62 3.43 8.95 5.06 10.25 3.11 9.68 4.62
CASHTA is the cash flow to total assets ratio, calculated as the operating income before depreciation minus total
taxes adjusted for changes in deferred taxes minus gross interest expenses and dividends and divided by total
assets. LEVERAGE is the long-term debt as a fraction of the book value of the total assets. MTB is the market-to-
book ratio, calculated as the average market value of equity 3 months prior to the event day plus the book value of
assets in the previous year minus the book value of equity in the previous year divided by the book value of total
assets in the previous year. LONGINV is the long-term investment expenditure to total asset ratio.
We believe that agency/political costs consideration can help explain why rights issues
and underwritten offers are followed by significantly negative and positive stock-market
valuations, respectively.
This leaves us with the following puzzle: If uninsured rights do not signal good news to
the market, why are they predominantly used by Chinese firms in floating SEOs?
To enhance the search for the underlying causes of the rights issue anomaly, Table 10 is
constructed to present the mean and median statistics for fractional managerial,
institutional, and government ownership, as well as the ratio of nonessential expenses
to total expenses, categorized by SEO flotation methods. An beyeball analysisQ of Table 10
yields comparisons that are important regularities predicted by agency/political costs
paradigm: Firms using underwriter certification have, on average, larger managerial and
institutional ownership of shares than do firms using uninsured rights. The only exception
is that rights issuers have a slightly larger managerial ownership in 1997. Another
distinguishing feature of firms that issue uninsured rights is that the variable GOVNT, the
proportion of shares owned by the state government, is far larger than for firms using
underwritten offers. Moreover, firms using underwritten offers have, on average, less
nonessential expense ratio than do those issuing uninsured rights in every year during the
sample period.
Table 9
Average abnormal returns (ARt ) and cumulative abnormal returns (CARK,L ) on and around SEO announcement
days
Event day Rights issues Underwritten offers
ARt CARK,L t-statistic ARt CARK,L t-statistic
10 0.14 0.05 0.16 0.17 0.17 0.52
9 0.07 0.12 0.38 0.13 0.30 0.74
8 0.15 0.03 0.11 0.22 0.52 1.59
7 0.13 0.16 0.44 0.39 0.13 0.41
6 0.10 0.06 0.19 0.15 0.02 0.11
5 0.15 0.09 0.29 0.20 0.18 0.55
4 0.18 0.09 0.30 0.14 0.32 0.79
3 0.11 0.20 0.57 0.11 0.43 1.37
2 0.26 0.06 0.20 0.18 0.05 0.22
1 0.74 0.68 1.71 2.93 2.98 4.58
0 1.83 2.51 5.75 1.47 4.45 6.95
1 0.10 2.41 5.46 0.33 4.12 6.44
2 0.09 2.32 5.14 0.64 3.48 5.43
3 0.31 2.01 4.08 0.28 3.76 5.87
4 0.44 1.57 2.61 0.30 4.06 6.34
5 0.27 1.84 3.40 0.16 4.22 6.59
6 0.11 1.95 3.82 0.15 4.07 6.36
7 0.15 1.80 3.27 0.11 3.96 6.18
8 0.33 1.47 2.32 0.52 4.48 7.02
9 0.08 1.39 2.04 0.36 4.84 7.56
10 0.09 1.48 2.35 0.21 5.05 7.89
The abnormal return for firm i on day t is ARi,t =R i ,t R c,i,t , where R i,t is the return for firm i on day t, and
R c,i,t is the returnPfor the control portfolio of firm i on day t. The average abnormal returns for each trading
N
day is ARt ¼ N1 i¼1 ARi;t , wherePN is the number of stocks. The average cumulative abnormal returns
N
from day K to L is CARK;L ¼ N1 i¼1 CARi;K;L .
Based on the preliminary analysis on statistics reported in Tables 10 and 11, the
more favorable characteristics belong to firms with underwritten offers, which exhibit
more liquidity, less leverage ratio, better investment opportunities, and evidence of
lower agency/political costs than do firms issuing uninsured rights. Perhaps, the most
striking feature is that the average abnormal return is distinctly positive for the
issuers of firm commitment public offerings but negative for the issuers of uninsured
rights.
To test the above hypotheses, we test a logit model of a firm’s choice of alternative
SEO flotation methods. The dependent variable equals to 1 if a firm issues uninsured
rights, and 0 otherwise. The independent variables consist of proxies for agency costs,
political costs and investment opportunities, leverage ratio, the initial IPO return (IPOR),
and the logarithm of the proceeds from SEOs (SEOVALUE). IPOR is included in the
logit regression to proxy the quality of a firm. Su and Fleisher (1999) show that Chinese
firms of better quality tend to underprice more and return to the secondary market to
issue SEOs in greater amount more quickly. SEOVALUE is included to control for the
size effect. Perhaps, sheer size has an informational advantage in that economies of scale
increase the likelihood that investors will know something about these firms, enabling
Fig. 2. Cumulative abnormal returns on and around the SEO announcement day.
them to make better informed investment decisions when opportunities to purchase new
shares arise.
Empirical results from three logit estimations of the probability of rights issues are in
Table 12. The coefficient estimates on the MANAGER variable have the correct sign but
are insignificant in all three regressions. That is, the probability of rights issues does not
decrease directly with a firm’s managerial ownership. However, the coefficient estimates
for TOPTEN are significantly negative at the 5% level, as predicted by the agency/political
costs hypothesis. Firms that are more closely monitored by large institutional shareholders
are less likely to issue uninsured rights. The coefficient estimates for GOVNT are
significantly positive at the 5% level, indicating that firms with larger government
ownership and more political intervention are more likely to use uninsured rights in
issuing SEOs. In addition, the probability of rights issues are positively related to the
nonessential expenses, and the relationship is statistically significant at the 10% level. This
renders further support to the agency/political cost approach to explain rights issues puzzle
in China.
The coefficient estimates for LONGINV and LEVERAGE are of the correct signs, as
predicted by the alternative investment opportunities hypothesis, but they are not
statistically significant. The coefficient estimates for MTB is not only of incorrect sign
but also statistically insignificant. The evidence presented in Table 12 provides no support
for the investment opportunities hypothesis.
It is interesting to note that the coefficient estimate for IPOR is insignificant, suggesting
that there is no relationship between the probability of using rights issues in floating SEOs
and the degree of IPO underpricing, which is consistent with Su (2003) and Su (2004).
However, the coefficient estimate for SEOVALUE is significantly negative, implying that
the probability of using a underwriter increases in the size of the issue. This is probably
because larger issues have more diverse ownership of shares and less government
interference.
Table 10
Agency and political costs for firms classified by SEO flotation choices
Year Uninsured rights Underwritten offers
Mean S.D. Median S.D. Mean S.D. Median S.D.
MANAGER 1995 0.46 0.51 0.52 0.60 1.63 1.67 1.49 1.73
1996 0.81 0.69 0.70 1.03 1.34 1.15 1.22 1.40
1997 1.09 0.96 1.10 1.22 1.07 0.98 1.05 1.31
1998 1.26 1.00 1.08 1.33 1.55 1.36 1.47 1.75
1999 1.14 0.96 1.07 1.24 1.83 1.47 1.68 2.05
TOPTEN 1995 5.05 6.22 5.28 8.15 14.62 12.30 13.20 15.17
1996 12.64 11.05 11.38 16.31 17.10 15.46 15.83 18.75
1997 9.82 8.61 9.39 13.05 21.44 18.20 19.60 25.13
1998 10.77 8.35 9.62 11.00 19.85 17.26 18.35 23.28
1999 8.83 7.18 8.40 11.33 24.06 20.26 22.39 31.05
GOVNT 1995 34.08 29.55 37.50 51.33 14.24 16.29 13.20 18.41
1996 30.03 27.60 32.85 46.34 12.62 14.08 14.80 19.36
1997 28.25 26.42 29.71 41.66 13.48 15.17 14.63 19.83
1998 31.50 34.00 34.62 52.04 11.60 13.52 12.48 16.00
1999 27.58 24.10 29.01 42.61 10.94 12.87 11.40 15.99
WELFARE 1995 5.72 1.73 5.48 6.91 2.36 1.03 2.08 3.44
1996 5.14 2.30 4.80 6.21 2.52 1.12 2.37 3.10
1997 6.03 3.22 5.64 7.84 1.82 0.90 1.65 2.10
1998 5.33 3.05 4.96 6.32 1.62 0.85 1.47 2.25
1999 4.85 2.68 4.61 5.96 1.73 1.01 1.50 2.24
MANAGER is the number of shares held by managers and board members divided by the total number of shares
outstanding. TOPTEN is the percentage of shares held by the top 10 shareholders, including institutional and
private investors but excluding the state government. GOVNT is the percentage of shares retained by the state
government. WELFARE is the nonessential expenditures (total expenses minus costs of goods sold, wage
expenses, and interest expenses) as a fraction of total expenses.
4.5. Valuation effect of SEO announcements
We now study the stock price reaction to SEOs made by Chinese firms. A larger
number of studies on U.S. firms have convincingly showed that SEOs are associated
with a decrease in announcement firm’s stock price, especially when the flotation
method used is firm commitment underwritten offers. However, some studies (Bohren
et al., 1997; Kang & Stulz, 1996) document positive investors’ reaction to uninsured
rights issues in other countries. As mentioned in Section 4.1, on average, investors
respond negatively to uninsured rights offers but positively to underwritten offers in
China.
To explain the abnormal returns surrounding SEO issue announcements, we estimate
the following cross-section regression using generalized least squares (GLS):
CARi;1;0 ¼ a1 þ a2 SEOVALUEi þ a3 CARi;11;2 þ a4 STDi þ a5 SHAREi
þ a6 MANAGERi þ a7 TOPTENi þ a8 GOVNTi

þ a9 WELFAREi þ a10 DUMMYi þ ni ð1Þ
Table 11
Logit regression on choices of SEO flotation methods
(i) (ii) (iii) (iv) (v)
Intercept 0.418* (5.530) 0.525* (5.911) 0.292* (3.428) 0.236* (3.166) 0.409* (5.367)
MANAGER 1.850 (1.482) 1.644 (1.417) 1.191 (1.194) 1.265 (1.370)* 1.083 (0.991)
TOPTEN 0.525* (2.138) 0.763* (2.540) 0.612* (2.463) 0.428* (2.072) 0.906* (2.974)
GOVNT 1.805* (3.400) 1.433* (2.835) 0.868* (2.473) 0.735* (2.196) 0.660* (2.007)
WELFARE 3.250* (2.614) 2.871* (2.290) 1.658y (1.853) 0.926y (1.701) 1.202y (1.766)
MTB 0.125 (0.353) 0.075 (0.140) 0.122 (0.169) 0.068 (0.153)
LONGINV 0.169 (0.380) 0.240 (0.452) 0.333 (0.628) 0.464 (0.735) 0.509 (0.711)
LEVERAGE 0.364 (0.764) 0.301 (0.851) 0.493 (0.602)
IPOR 0.037 (1.060)
SEOVALUE 1.026* (2.144)
Pseudo R 2 0.165 0.171 0.183 0.191 0.215
The dependent variable equals to one if a firm issues uninsured rights, and zero otherwise. The independent
variables consist of the proportion of shares held by managers and directors (MANAGER), the proportion of
shares held by the top 10 largest shareholders, excluding the government (TOPTEN), the proportion of shares
held by the government (GOVNT), the ratio of nonessential expenses and total assets (WELFARE), market-to-
book ratio (MTB), long-term investment expenditure to total asset ratio (LONGINV), long-term debt to total asset
ratio (LEVERAGE), the initial IPO return (IPOR), and the logarithm of the proceeds from SEOs (SEOVALUE).
Figures in parentheses are t-statistics.
* Level of significance: 5%.
y
Level of significance: 10%.
where DUMMYi is a dummy variable that takes value one if the issue uses uninsured
rights and zero otherwise; CARi,1,0 is the cumulative abnormal return during the 2-day
announcement window for the ith issue; SHAREi is the ratio of the number of shares
offered to the number of shares outstanding prior to the offer, a proxy for the extent of
share dilution in the case for underwritten offers and the degree of potential shareholder
take-up in the case of rights issues; CARi, 12,2 is the cumulative daily abnormal
returns during a 10-day period, beginning at 12 days before the issue announcement day,
a proxy detecting possible stock price run-up prior to the event day; and STDi is
standard deviation of stock returns calculated over a period from 122 days before the
announcement day and 122 days after the announcement, a proxy for the risk of the
issue.
The results contained in Table 12 indicate that the proportion of shares retained by
state government and held by institutional investors are two important variables to
explain the cross-section variation in abnormal returns around SEO announcements. The
coefficient estimate for GOVNT is significantly negative at the 10% level in regressions
I, II, and IV and significantly negative at the 5% level in regression II when
SEOVALUE is added to the list of independent variables. There is strong evidence that
investor’s reaction more positively to SEOs in which government ownership is smaller.
The coefficient estimate for TOPTEN is significantly positive at the 10% level in
regressions I, II, and IV, although it is insignificant in regression II when SEOVALUE is
added. There is some evidence that institutional ownership is positively related to
announcement day abnormal returns. However, the coefficient estimates for MANAGER
and WELFARE are insignificant, suggesting that the size of share ownership by
Table 12
Valuation effect of SEO announcements
(I) (II) (III) (IV)
Intercept 0.827* (5.830) 0.093* (2.144) 0.174* (2.592) 0.105* (4.550)
MANAGER 0.057 (0.828) 0.126 (1.049) 0.088 (0.735) 0.104 (0.586)
TOPTEN 0.059y (1.722) 0.017 (0.619) 0.074y (1.833) 0.081y (1.861)
GOVNT 0.012y (1.820) 0.032* (2.217) 0.018y (1.891) 0.025y (1.926)
WELFARE 0.007 (0.238) 0.008 (0.274) 0.008 (0.281) 0.008 (0.309)
SEOVALUE 0.048 (1.390)
SHARE 0.174 (1.192)
CARi,11,2 0.159 (1.208)
STD 0.637 (0.828)
DUMMY 0.429* (2.738) 0.260* (2.395) 0.316* (3.091) 0.385* (3.324)
R̄2 0.265 0.280 0.314 0.299
The dependent variable is the cumulative abnormal return during the 2-day announcement window for the ith
issue (CARi,1,0). The independent variables include the logarithm of the proceeds from SEOs (SEOVALUE), the
ratio of the number of shares offered to the number of shares outstanding prior to the offer (SHARE), the
cumulative daily abnormal returns during a 10-day period (CARi,11,2), the standard deviation of stock returns
calculated over a period from 122 days before and 122 days after the SEO announcement (STD), the proportion of
shares held by managers and directors (MANAGER), the proportion of shares held by the top 10 largest
shareholders, excluding the government (TOPTEN), the proportion of shares held by the government (GOVNT),
the ratio of nonessential expenses and total assets (WELFARE), and a dummy variable that takes the value one if
the issue uses uninsured rights, and zero otherwise (DUMMY). Figures in parentheses are t-statistics.
* Statistically significant at the 5% level.
y
Statistically significant at the 10% level.
management and nonessential expenses do not affect announcement day returns in a

systematic way.
When SEOVALUE is added to the cross-section regression, it is insignificant. The
result is in contrast to Masulis and Korwar (1986) and Kalay and Shimrat (1986), who find
that larger issues result in more negative abnormal returns in the United States. The
coefficient estimate for SHARE is also insignificant, suggesting that the percentage change
in outstanding shares has no effect on abnormal return. Neither the potential extent of
shareholder take-up in the rights issues nor the possibility of share dilution affects
investors’ reaction to firms’ subsequent equity offerings. Moreover, a firm’s prior
performance in the stock market, as measured by CARi, 12,2, and the company’s specific
risk, as proxied by STDi , are insignificant. There is no evidence that preissue short-term
abnormal return is negatively related to announcement day abnormal return found in
Korajczyk, Lucas, and McDonald (1991).
Finally, the coefficient estimate for the uninsured rights dummy is significantly
negative, indicating that market reacts more negatively to rights offerings. Overall,
the empirical evidence suggests that the choice of SEO flotation methods reveals
ownership conflicts and external monitoring and that market favors underwritten
offers.
In summary, the choice of equity financing methods by Chinese firms reveals the extent
of ownership conflicts and shareholder monitoring, rather than firms’ investment
opportunities.
5. Conclusion
This paper analyzes the corporate dividend policies and post-IPO financing choices of
newly privatized SOEs in China, as they reflect conflicts among government’s desire to
retain political control, private investors’ desire for return on their investments, and
management’s desire to appropriate resources for its own benefit. The paper thus hopes to
shed some light on resolving the stock dividend puzzle and the rights issues puzzle in
corporate China.
While stockholders respond positively to stock dividend announcements in the
United States, which can be explained by the signaling theory, stockholders react
negatively to stock dividend distributions in China. If stock dividend does not signal
good news, why do so many Chinese firms bother to distribute stock dividends? The
paper finds that firms with a relatively high degree of government control and a
relatively less managerial and institutional ownership are more likely to distribute
stock dividends as opposed to cash dividends. However, there is no evidence that
these firms have better investment opportunities (proxied by the market-to-book and
long-term investment to total assets ratios) than do those distributing cash dividends.
As a result, a large and pronounced decline in cumulative abnormal return is observed
for stock dividend announcements during the event window. A comparison of the
average buy-and-hold abnormal returns between cash dividend and stock dividend
firms over different holding periods shows that stock dividend firms perform poorly in
the long run.
While fewer and fewer firms issue uninsured rights in the United States and
elsewhere, uninsured rights are predominantly used as a means of obtaining seasoned
equities in China. The signaling model posits that, under asymmetric information,
managers who know of profitable net present value projects convey positive news to
outside investors via the flotation method chosen, with standby rights the first, uninsured
rights the next, and underwritten offers the last. The adverse-selection model argues that
uninsured rights require higher adverse-selection costs and therefore reveal more positive
information to investors. However, the predictions of both types of model are
inconsistent with the Chinese situation because firms using uninsured rights exhibit no
evidence of superior investment opportunities and experience a significant drop in stock
prices surrounding the announcement day. The paper finds that the key in addressing the
rights issues puzzle is the unique ownership conflicts that arise due to agency problems
and political control. Firms with higher agency and political costs are more likely to issue
uninsured rights. The empirical evidence on the determinants of equity financing choices
and the cross-section pattern of announcement day abnormal returns validates this
argument.
Although the paper obtains some interesting results, two issues remain. One is that the
securities regulatory authority in China requires certain qualifications (e.g., 6% ROE
during the past 3 years) to make SEOs, therefore, some firms may attempt to divert cash to
reduce the size of net assets and to inflate the accounting profitability ratios. The intricate
relation between dividend policies and SEO behavior is left unexplored. Another is that
the empirical analysis does not differentiate political from agency costs, as the main
objective is to differentiate agency/political consideration from the liquidity/investment
opportunities argument. We hope to explore these unresolved issues more fully in future
research.
Acknowledgements
I thank Chun Chang, the coeditor, two anonymous referees, Belton Fleisher, Xigang
Han, Mike Levis, Jun Liu, Melinda Newman, Andrew Saprarovscheko, Steven Yamarik,
Hongmao Zhou, and the seminar participants at Jinan University, Tulane University,
University of Akron, and the CES session during the 2001 ASSA meetings in New
Orleans for their helpful comments and suggestions. I also acknowledge financial support
by the Excellent Young Teachers Program (EYTP) of the Chinese State Education
Ministry, Fok Ying Tong Education Foundation, and Guangdong Natural Science
Foundation (# 031909).
Appendix A. Methodology used in computing abnormal, cumulative abnormal, and

buy-and-hold abnormal returns
To study the valuation effect of dividend and seasoned equity announcements, two
sets of control portfolio of firms not subject to the event are constructed to match the
performance of firms making the announcements. One set of control portfolio consists
of six firms making no dividend announcement within a period between 12 months
prior to and 12 months after the event day for each dividend announcement firm.
Another set of control portfolio consists of six firms making no SEO announcement
within a period between 12 months prior to and 12 months after the event day for
each SEO announcement firm. In the spirit of Barber and Lyon (1997), we use the
size and market-to-book ratio to select firms included in the control portfolio. In
particular, we minimize the global distance between the firm making the announce-
ment and the firm in the control portfolio according to the following criterion:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ðSIZEi þ SIZEc Þ2 ðMTBi MTBc Þ2
di ¼ 2
þ
rSIZE r2MTB
where d i is the Euclidian distance between the announcement firm i and control firm
c. SIZEi and SIZEc are average market values for announcement firm i and control
firm c 3 months prior to the event day, respectively. MTBi and MTBc are the market-
to-book ratios for announcement firm i and control firm c, respectively. Market-to-
book ratio is calculated as the average market value of equity 3 months prior to the
event day plus the book value of assets in the previous year minus the book value of
equity in the previous year divided by the book value of assets in the previous year.
r SIZE2 and r MTB2 are the cross-section variances of the average market value and
market-to-book ratio for all firms, respectively.
The abnormal and cumulative abnormal returns are calculated as follows:

ARi;t ¼ Ri;t Rc;i;t ;
where ARi,t is the abnormal return for firm i on day t, R i,t is the return for firm i on day t,
and R c,i,t is the return for the control portfolio of firm i on day t. Average abnormal returns
for each trading day within the event window are calculated by
1 XN
ARt ¼ ARi;t ;
N i¼1
where N is the number of stocks with abnormal returns during event day t. The cumulative
abnormal returns for each stock i, CARi , are formed by summing average abnormal
returns over time as follows:
X
L
CARi;K;L ¼ ARi;t ;
t¼K
where the CARi,K,L is for the period from t=day K until L. The average cumulative
abnormal returns over the event window from day K until L are calculated by
1 XN
CARK;L ¼ CARi;K;L :
N i¼1
Following Ritter (1991), the t-statistics are calculated for CARK,L as follows:
pffiffiffiffi
N CARK;L
tCARK;L ¼
SD CARK;L ;
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where SD CARK;L ¼ T VarðARt Þ þ 2ðT 1Þ CovðARt ; ARtþ1 Þ, T=KL+1, is
the standard deviation of CARK,L . Var(ARt ) and Cov(ARt , ARt+1) are estimated during
a period from 122 to 11 days before the announcement day plus during a period from
11 days to 122 days after the announcement day. The covariance term adjusts for
possible first-order autocorrelation between the abnormal returns due to nonsynchronous
trading.
To study the long-term performance of firms announcing dividends or SEOs, the
average buy-and-hold abnormal return (BHAR) for s months are computed as
follows:
N
Ys s
1 X Y
BHARs ¼ ð1 þ ri;t Þ ð1 þ rc;i;t Þ ;
N i¼1 t¼1 t¼1
where r i,t is the monthly stock return for announcement firm i and r c,i,t is the monthly return
on control portfolio for firm i. Following Barber, Lyon, and Tsai (1999), the test statistic for
buy-and-hold abnormal return is the skewness-adjusted t-statistic:
pffiffiffiffi
1 2 N
adjusted tBHARs ¼ tBHARs þ pffiffiffiffi ðtBHARs Þ skews þ skews ;
3 N 6
where t BHARs is the usual t-statistic for BHARs ,

pffiffiffiffi
N BHARs
tBHARs ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
s VarðBHARt Þ þ 2ðs 1Þ CovðBHARt ; BHARtþ1 Þ;
and skews is the skewness of the BHARt series (t=1,2, . . .,s).
References
Bai, C. E., Li, D. D., Tao, Z., & Wang, Y. (2000). A multitask theory of state enterprise reform. Journal of
Comparative Economics, 28, 716 – 738.
Barber, B. M., & Lyon, J. D. (1997). Detecting long-run abnormal stock returns: The empirical power and
specification of test statistics. Journal of Financial Economics, 43, 341 – 372.
Barber, B. M., Lyon, J. D., & Tsai, C. L. (1999). Holding size while improving power in tests of long-run
abnormal stock returns. Journal of Finance, 54, 165 – 202.
Bohren, O., Eckbo, B. E., & Michalsen, D. (1997). Why underwrite rights offerings? Some new evidence.
Journal of Financial Economics, 46, 223 – 261.
Boycko, M., Shleifer, A., & Vishny, R. W. (1996). A theory of privatization. Economic Journal, 106,
309 – 319.
Chang, E. C., & Wong, S. M. (2003). Political control and performance in China’s listed companies. University
of Hong Kong Working Paper.
Eckbo, B., & Masulis, R. (1992). Adverse selection and the rights offer paradox. Journal of Financial
Economics, 32, 293 – 332.
Grinblatt, M., Masulis, R., & Titman, S. (1984). The valuation effect of stock splits and stock dividends. Journal
of Financial Economics, 13, 461 – 490.
Grossman, S., & Hart, O. (1982). Corporate financial structure and managerial incentives. In J. McCall (Ed.), The
economics of information and uncertainty. Chicago, IL7 University Press.
Grossman, S., & Hart, O. (1986). The costs and benefits of ownership: A theory of vertical and lateral integration.
Journal of Political Economy, 94, 691 – 719.
Groves, T., Hong, Y., McMillan, J., & Naughton, B. (1994). Autonomy and incentives in Chinese state
enterprises. Quarterly Journal of Economics, 109, 183 – 209.
Han, D. Z., & Li, Y. R. (2002). An empirical study on the financing choices of Chinese listed companies. Essays
on Finance and Economics (Chinese), 21, 52 – 57.
Hart, O. (1995). Firms, contracts, and financial structure. London7 Oxford University Press.
Hart, O., & Moore, J. (1990). Property rights and the nature of the firm. Journal of Political Economy, 98,
1119 – 1158.
Heinkel, R., & Schwartz, E. (1986). Rights versus underwritten offerings: An asymmetric information approach.
Journal of Finance, 41, 1 – 18.
Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance and takeovers. American Economic
Review, 76, 323 – 329.
Jensen, M. C., & Meckling, W. H. (1976). Theory of firm: Managerial behavior, agency cost and capital structure.
Journal of Financial Economics, 3, 305 – 360.
Kalay, A., & Shimrat, A. (1986). Firm value and seasoned equity issues: Price pressure, wealth redistribution, or
negative information. Journal of Financial Economics, 19, 109 – 126.
Kang, J. K., & Stulz, R. M. (1996). How different is Japanese corporate finance? An investigation of the
information content of new equity issues. Review of Financial Studies, 9, 109 – 139.
Kernen, A. (1997). State enterprises in Shenyang: Actors and victims in the transition. China Perspectives,
26 – 32.
Korajczyk, R. A., Lucas, D. J., & McDonald, R. L. (1991). The effect of information releases on the pricing and
timing of equity issues. Review of Financial Studies, 4, 685 – 708.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A., & Vishny, R. W. (2000). Agency problems and dividend policies
around the world. Journal of Finance, 55, 1 – 33.
Lin, H. W. (1999, December 23). Investors to tread warily in China oilfields. Financial Times, 8, 6 – 7.
Lu, C. J., & Wang, K. M. (1999). An empirical investigation of dividend policies of Chinese listed firms.
Economic Research Journal, 12, 42 – 51.
Maksimovic, V., & Titman, S. (1991). Financial policy and a firm’s reputation for product quality. Review of
Financial Studies, 4, 175 – 200.
Masulis, R. W., & Korwar, A. N. (1986). Seasoned equity offerings: An empirical investigation. Journal of
Financial Economics, 15, 91 – 118.
Myers, S. C., & Majluf, N. (1984). Corporate financing and investment decisions when firms have information
that investors do not have. Journal of Financial Economics, 13, 187 – 221.
Rankine, G., & Stice, E. K. (1997). The market reaction to the choice of accounting method for stock splits and
large stock dividends. Journal of Financial and Quantitative Analysis, 32, 161 – 182.
Ritter, J. R. (1991). The long-run performance of initial public offerings. Journal of Finance, 42, 365 – 394.
Shleifer, A., & Vishny, R. (1986). Large shareholders and corporate control. Journal of Political Economy, 94,
461 – 488.
Shleifer, A., & Vishny, R. (1994). Politicians and firms. Quarterly Journal of Economics, 109, 995 – 1025.
Stulz, R. (1990). Managerial discretion and optimal financing policies. Journal of Financial Economics, 26,
3 – 27.
Su, D. (2003). Chinese stock markets: A research handbook. Singapore7 World Scientific (UK) Publisher.
Su, D. (2004). Leverage, insider ownership, and the underpricing of Chinese IPOs. Journal of International
Financial Markets, Institutions, and Money, 14, 37 – 54.
Su, D., & Fleisher, B. (1999). An empirical investigation of underpricing in Chinese IPOs. Pacific-Basin Finance
Journal, 7, 173 – 202.
Wei, G. (2000). Dividend policy under asymmetric information. Economic Science (Chinese), 2, 42 – 46.
Weisbach, M. S. (1988). Outside directors and CEO turnover. Journal of Financial Economics, 20, 431 – 460.
Xu, X., & Wang, Y. (1999). Ownership structure and corporate governance in Chinese stock companies. China
Economic Review, 10, 75 – 98.
Yuan, H. Q. (2003). Information asymmetry and signalling: Evidence from rights issues in China. Accounting
Research (Chinese), 5, 16 – 22.
Zhang, W. D., & Wang, Q. (2001). Large shareholders and equity financing choices after IPOs in China.
Accounting Studies (Chinese), 10, 21 – 28.

Corporate Financenext Term and State Enterprise Reform in Previous Termchinanext Term

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Corporate Financenext Term and State Enterprise Reform in Previous Termchinanext Term

Uploaded by

Copyright:

Available Formats

China Economic Review 16 (2005) 118 – 148

Corporate finance and state enterprise reform

Accepted 20 September 2004

JEL classification: G32; G35; O53; P21

* Tel.: +86 20 8522 4798.

2. Institutional framework surrounding SOE reform

The reform of SOEs in China has been characterized by efforts to decentralize,

in technology investment, profit maximization, and efficient resource allocation, all of

3. Dividend policy and equity financing choices

3.1. Agency costs, political costs, and stock dividend puzzle

3.2. Agency costs, political costs and rights issues puzzle

4.2. The determinants of dividend distribution decisions

D. Su / China Economic Review 16 (2005) 118–148

4.3. Long-term performance of dividend distribution firms

Some researchers may be skeptical about the use of market-based performance

4.4. The determinants of equity financing decisions

4.5. Valuation effect of SEO announcements

CARi;1;0 ¼ a1 þ a2 SEOVALUEi þ a3 CARi;11;2 þ a4 STDi þ a5 SHAREi

þ a6 MANAGERi þ a7 TOPTENi þ a8 GOVNTi

management and nonessential expenses do not affect announcement day returns in a

Appendix A. Methodology used in computing abnormal, cumulative abnormal, and

The abnormal and cumulative abnormal returns are calculated as follows:

where t BHARs is the usual t-statistic for BHARs ,

You might also like