i

DIVIDEND PAYOUT POLICY IN THE CHINESE EQUITY MARKET

By

JIANAN GUO

MWM (DEAKIN), BE (FUDAN)

Submitted in fulfilment of the requirements for the degree of

Doctor of Philosophy

Deakin University

November, 2013

sfol

Retracted Stamp

sfol

Retracted Stamp

iv

ACKNOWLEDGEMENTS

To my beloved family and friends, who have provided me with continuing support

throughout my PHD candidature, I owe my sincere appreciation. Specifically, I owe the

greatest debt to my parents who have encouraged me to pursue the higher education

degree. I am also fortunate to receive the love and support of my wife, Huan Sun, who has

shared the sweet and bitter with me, particularly in the final stages of this thesis. This work

is dedicated to them.

It has also been a challenging task to stay motivated and innovative within a balanced

research dissertation. I greatly appreciate my supervisors, Professor Gerald Gannon, Dr

Hong Feng Zhang and Dr Hoa Nyugen, for their enthusiasm and continuous guidance

throughout my PHD candidature. To me, they have set the best examples of a better

academic as well as a self motivated researcher.

Special thanks are owed to Professor Edward Lin and Professor Nava Subramaniam, for their

efforts to sort out the administrative issues during the final stage of this thesis.

The research has also benefited from the comments of attendees at the 18th SFM

Conference, National Sun Yat Sen University (Kaohsiung, Republic of China), 2011 PhD

Conference in Economics and Business (University of Queensland), 6th International

Accounting and Finance Doctoral Symposium (Bologna, Italy) and the 21st Annual

Conference on Pacific Basin Finance, Economics, Accounting and Management.

And, finally, the Higher Degree Research Scholarship and the financial support from the

School of Accounting, Economics and Finance, Deakin University, are gratefully

acknowledged.

v

ABSTRACT

Dividend payout policy is a critical component of corporate financial management, as

well as being important to investors. Empirical studies, however, that focus on the cash

dividend policy in emerging markets, especially the Chinese equity market, are limited. This

thesis attempts to substantiate empirical evidence on the topic of dividend payout policy

with a particular focus on its interaction with ownership structure, stock market abnormal

return and subsequent earnings growth in the Chinese market.

This thesis supplies an empirical analysis of the influence of the ultimate controlling

shareholders, including their types and control rights, on the cash dividend payout

announcement, using a sample of 1,200 Chinese listed firms. The results of this analysis

provide sound support for the viewpoint that the cash dividend policy is related to

ownership structure in the Chinese market. Specifically, the level of the ultimate controlling

shareholders’ control rights, defined as the aggregate direct and indirect shareholdings, is

positively correlated with the likelihood and magnitude of the cash dividend payout.

Further, the outcomes also demonstrate that firms with various types of ultimate

controlling shareholders exhibit divergence in the likelihood and magnitude of cash

dividend payouts. The cash payout policy is also sensitive to firm characteristics such as size,

profitability and financial leverage.

This thesis also investigates the investors’ reactions towards the dividend

announcements with a sample from 1,203 Chinese listed firms. The results indicate that

Chinese minority shareholders respond positively to the stock dividend announcement,

especially in the case of a stock dividend from capital reserves. Unexpected cash dividends

are also positively associated with stock abnormal returns, while unexpected earnings have

vi

little, or even negative, influence on announcement effects. Further, investors respond

discriminatively towards the unexpected cash dividends from the listed firms with various

types of ultimate controlling shareholders. In addition, dividend announcement effect is

also sensitive to idiosyncratic risk and firm size.

This thesis also extends the investigation into how cash dividend announcements

influence the subsequent earnings growth in the Chinese market. Consistent with some

contemporary studies, the results suggest that a higher cash payout is associated with

better future earnings growth, but the positive association diminishes as the payout ratio

increases. The stock dividend decision is negatively correlated with future earnings growth,

which is contrary to the investors’ positive response towards these announcements.

Further, future earnings’ growth is also closely correlated with profitability, dividend yield

and growth opportunity.

Together, the results of this thesis suggest that the cash dividend policy of Chinese

listed firms is worthy of continuous investigation as the Chinese market grows and evolves.

A major contribution of this thesis is that it illustrates that both Chinese investors and listed

firms have acknowledged the crucial role of the cash dividend policy in corporate

governance infrastructure. Correspondingly, this indicates several policy implications. First,

the privatization of Chinese state owned enterprises enhances the diversity of shareholders

and improvesminority shareholders’ positions in the capital market. Second, by establishing

a robust cash dividend payout, Chinese listed firms are able to cope with the agency cost

due to the separation of ownership and control, and consequently improving the efficiency

and the firm valuation.

vii

TABLE OF CONTENTSChapter 1 Introduction ........................................................................................................................1

1.1. Introduction ..............................................................................................................................1

1.2. Research Scope .........................................................................................................................7

Chapter 2 Literature on dividend payout policy ............................................................................... 13

2.1. Introduction ........................................................................................................................... 13

2.2. Lintner model and dividend irrelevance theorem................................................................. 13

2.3. Dividend payout policy, agency cost and corporate governance .......................................... 16

2.3.1. Agency cost and dividend policy..................................................................................... 16

2.3.2. Corporate governance and dividend policy .................................................................... 22

2.4. Dividend policy and stock abnormal return........................................................................... 31

2.4.1. Dividend announcement effect: US and other developed markets ............................... 31

2.4.2. Dividend announcement effect: emerging markets....................................................... 39

2.5. Dividend Policy and Future Earnings Growth ........................................................................ 41

2.5.1. Signalling hypothesis of dividend policy ......................................................................... 41

2.5.2. Relationship between dividend policy and future earnings growth............................... 48

2.6. Summary ............................................................................................................................... . 53

Chapter 3 Review of Chinese Stock Market and Payout Policy ........................................................ 56

3.1. The Evolution of the Chinese Stock Market........................................................................... 56

3.2. Legal Environment of the Chinese Stock Market................................................................... 62

3.2.1. Corporation Law.............................................................................................................. 62

3.2.2. Securities Law ................................................................................................................. 63

3.2.3. Income Tax Code............................................................................................................. 65

3.3. Payout Practice in the Chinese Stock Market ........................................................................ 67

3.3.1. Evolution of regulatory requirement on cash payout .................................................... 68

3.3.2. Snapshot of payout practice in the Chinese stock market ............................................. 70

3.4. Summary ............................................................................................................................... . 73

Chapter 4 Ultimate Controlling Shareholders and Dividend Payout Policy in the Chinese StockMarket............................................................................................................................... ................ 74

4.1. Introduction ........................................................................................................................... 74

4.2. Hypotheses development ...................................................................................................... 79

4.3. Data and descriptive statistics ............................................................................................... 83

4.3.1. Proxies for cash dividend policy, control rights and type of UCSs.................................. 84

viii

4.3.2. Control variables ............................................................................................................. 87

4.3.3. Descriptive statistics ....................................................................................................... 89

4.4. Empirical Analysis................................................................................................................... 92

4.4.1. Cash dividend policy, control rights and cash rights....................................................... 92

4.4.2. Cash dividend policy and types of UCSs.......................................................................... 97

4.4.3. Robustness check.......................................................................................................... 102

4.5. Summary .............................................................................................................................. 106

Chapter 5 Dividend Announcement Effect and Ultimate Controlling Shareholders in the ChineseStock Market............................................................................................................................... .... 108

5.1. Introduction ......................................................................................................................... 108

5.2. Hypotheses development .................................................................................................... 113

5.3. Data and descriptive statistics ............................................................................................. 118

5.3.1. Sample construction ..................................................................................................... 118

5.3.2. Dependent variable....................................................................................................... 119

5.3.3. Independent variables .................................................................................................. 120

5.3.4. Descriptive statistics ..................................................................................................... 123

5.4. Empirical results................................................................................................................... 127

5.4.1. Univariate analysis ........................................................................................................ 127

5.4.2. Multivariate analysis ..................................................................................................... 130

5.4.3. Robustness Check ......................................................................................................... 137

5.5. Summary .............................................................................................................................. 138

Chapter 6 Dividend Payout Policy and Future Earnings Growth in the Chinese Stock Market ...... 139

6.1. Introduction ......................................................................................................................... 139

6.2. Hypotheses development .................................................................................................... 143

6.3. Data and descriptive statistics ............................................................................................. 146

6.3.1. Sample construction ..................................................................................................... 146

6.3.2. Dependent variables – proxies for earnings growth..................................................... 147

6.3.3. Independent variables .................................................................................................. 147

6.3.4. Descriptive statistics ..................................................................................................... 150

6.4. Empirical Analysis................................................................................................................. 154

6.4.1. Univariate analysis ........................................................................................................ 154

6.4.2. Multivariate analysis ..................................................................................................... 157

6.4.3. Robustness test............................................................................................................. 165

ix

6.5. Summary .............................................................................................................................. 167

Chapter 7 Conclusion ...................................................................................................................... 170

7.1. Introduction ......................................................................................................................... 170

7.2. Key findings .......................................................................................................................... 172

7.2.1. Dividend policy, corporate control rights and subsequent equity financing................ 172

7.2.2. Dividend policy, corporate control rights and announcement effect .......................... 174

7.2.3. Dividend policy and future earnings growth ................................................................ 176

7.3. Summary .............................................................................................................................. 177

7.4. Concluding remarks ............................................................................................................. 181

Bibliography ............................................................................................................................... ..... 183

Appendices............................................................................................................................... ....... 196

x

LIST OF APPENDICES

Appendix 4.1 Robustness check: Comparison of various econometric models ………………….. 196Appendix 4.2 Robustness check: Multicollinearilty…………………………………………………………….. 197Appendix 4.3 Robustness check: Selection bias………………………………………………………………….. 198Appendix 5.1 Robustness check: Samples before winsorization…………………………………………. 199Appendix 5.2 Robustness check: Multicollinearility……………………………………………………………. 200Appendix 5.3 Robustness check: Fama Macbeth two step procedure……………………………….. 200Appendix 6.1 Robustness check: Omitting EXPECTED_ROA and/or DIVYIELD……………………… 201Appendix 6.2 Robustness check: Comparison of various econometric models…………………… 202Appendix 6.3 Robustness check: High/Medium/Low CASH_PAYOUT ( CASH_PAYOUT)…….. 204Appendix 6.4 Robustness check: Cash payers with simultaneous stock dividend……………….. 206Appendix 6.5 Robustness check: EPS growth over 2 year window……………………………………… 207Appendix 6.6 Robustness check: Selection bias – Heckman selection model……………………… 208Appendix 6.7 Robustness check: Multicollinearity……………………………………………………………… 208

xi

LIST OF FIGURES

Figure 3 1 Number of listed firms (1990 2011) ................................................................................ 58Figure 3 2 Annual trading turnover (1990 2011), in billion CNY ...................................................... 58Figure 3 3 Total market capitalization (1990 2011), in billion CNY .................................................. 58Figure 3 4 Negotiable/Total market capitalization ratio (1990 2011), in %..................................... 60Figure 3 5 Number of listed firms and percentage of cash dividend payers (1992 2010) ............... 71Figure 3 6 Average cash dividend payout ratio of cash payers (1992 2010).................................... 72Figure 5 1 CAAR trends of cash only payers grouped by the sign of UCEXPECTED_DPS andUNEXPECTED_EPS ........................................................................................................................... 129Figure 5 2 CAAR trends of cash only payers grouped by the types of UCSs .................................. 130

xii

LIST OF TABLES

Table 3 1 Distribution of listed firms with continuous cash dividend payout (1992 2009) ............. 70Table 4 1 Sample distribution by types of UCSs ............................................................................... 86Table 4 2 Variable definition............................................................................................................. 88Table 4 3 Descriptive statistics ......................................................................................................... 90Table 4 4 Pearson's correlation matrix ............................................................................................. 92Table 4 5 Control rights and the likelihood of cash dividends.......................................................... 94Table 4 6 Control rights and the magnitude of cash dividends........................................................ 95Table 4 7 Types of UCSs and the likelihood of cash dividends ......................................................... 99Table 4 8 Types of UCSs and the magnitude of cash dividends...................................................... 101Table 4 9 Difference between control rights and cash rights......................................................... 103Table 4 10 Robustness check: dividend policy and the gap between control rights and cash rights............................................................................................................................... ......................... 104Table 4 11 Robustness check: dividend policy, types of UCSs and the gap between control rightsand cash rights ............................................................................................................................... . 105Table 5 1 Distribution of observations by the combination of stock and cash dividends.............. 119Table 5 2 Variable definition........................................................................................................... 123Table 5 3 Descriptive statistics ....................................................................................................... 124Table 5 4 Pearson's correlation matrix ........................................................................................... 126Table 5 5 Average CARs of different combinations based on earnings shock, dividend shock andsimultaneous stock dividend .......................................................................................................... 128Table 5 6 Stock dividend announcement and stock abnormal returns.......................................... 131Table 5 7 Unexpected earnings, unexpected dividend and stock abnormal returns..................... 133Table 5 8 Control rights and stock abnormal returns..................................................................... 135Table 6 1 Sample distribution number of firms (% of total)......................................................... 146Table 6 2 Variable definitions ......................................................................................................... 149Table 6 3 Summary statistics of FUTURE_EPSG and FUTURE_EBITG by payout decision.............. 150Table 6 4 Descriptive statistics ....................................................................................................... 152Table 6 5 Pearson's correlation matrix ........................................................................................... 153Table 6 6 Mean FUTURE_EPSG (FUTURE_EBITG) comparison ....................................................... 156Table 6 7 Cash dividend and future earnings growth cash only payer sample ........................... 158Table 6 8 Future earnings growth and over investment problem ................................................. 162Table 6 9 Future earnings growth and stock dividend decision ..................................................... 164

1

Chapter 1 Introduction

1.1. Introduction

Dividend policy is important to the corporation and the design of dividend policy is an

essential component of corporate finance infrastructure. The literature on the dividend

policy and practice in developed markets is relatively comprehensive, but the coverage of

the dividend policy in emerging markets is yet to be complete. This thesis supplies recent

empirical research in the area of dividend policy with particular focuses on its interaction

with agency cost of free cash flow and ownership structure in the Chinese equity market.

In the United States and the United Kingdom, the conflict of interest between

shareholders and entrepreneurs, known as the agency cost,1 has raised concerns because

dispersed shareholders cannotmonitor the daily operations of the firm, while themanagers

have the discretion to use the firm’s resources to pursue personal interests (Jensen and

Meckling, 1976). Specifically, the managers are likely to abuse the firm’s free cash flow and

conduct empire building activities, such as take over sprees and investing in projects with

non positive net present value (Jensen, 1986). Consequently, a corporate governance

mechanism is necessary to monitor and restrict managers’ self dealing activities (Barnea et

al., 1981, Fama and Jensen, 1983, Agrawal and Knoeber, 1996, Gillan, 2006, Shleifer and

Vishny, 1997), among which cash dividends become an important measure of internal

corporate governance mechanisms to cope with the agency cost of free cash flow.2 By

1 Agency cost has a broad and multiple aspect definition which includes the stewardship betweenshareholders andmanagement, the asymmetric information between large (inside) shareholders andminority(external) shareholders, and the conflicts of interest between debt holders and shareholders.2 Other internal corporate governance measures include board composition, managerial remuneration andcorporate financial policies (Jensen, 1986, Jensen et al., 1992, Hermalin and Weisbach, 1991).

2

forcing the managers to distribute the excessive cash balance, the shareholders are able to

limit the opportunity of any imprudent investment decisions by the management (Jensen,

1986, Lang and Litzenberger, 1989, Yoon and Starks, 1995, Fuller and Blau, 2010).

Different from the traditional ownership structure in Anglo American markets where

the shareholdings are widely spread and the management is highly independent from the

owners, concentrated ownership is more popular among the civil law countries, including

Europe and most emerging markets, as well as some common law countries (La Porta et al.,

1999, Claessens et al., 2000, Claessens et al., 2002, Faccio and Lang, 2002). Strictly speaking,

the existence of large shareholders is a double edged sword. On the one hand, the largest

shareholders become part of the corporate governancemechanism and positively influence

the cash dividend payout as they make the managers distribute excessive cash and alleviate

the agency cost of free cash flow (Shleifer and Vishny, 1986, Dyck and Zingales, 2004, Truong

and Heaney, 2007, Holderness, 2003). On the other hand, the objective of large

shareholders is to maximize their private benefits, which may well drive them to pursue

unorthodox strategies to expropriate the wealth of the minority shareholders (Grossman

and Hart, 1980, Harris and Raviv, 1988). Empirical evidence suggests that large shareholders

have both motivation and capability to manipulate the dividend policy and benefit

themselves more than minority shareholders (Shleifer and Vishny, 1997).

The dividend policy, including both cash dividends and stock dividends, in the Chinese

equity market has become an interesting topic. Although past empirical studies suggest

Chinese listed firms pay less cash dividends to shareholders than firms in other countries

(La Porta et al., 2000a, Allen et al., 2005), this phenomenon has changed in recent years.

From the investors’ standpoint, cash dividends have become an important form of

3

investment return. On the other hand, listed firms are under continuous pressure from the

authorities to make more cash dividend payouts to external minority investors.3 Further,

listed firms are gradually adopting dividend payout policy as an instrument to deal with the

relationship between investors and managers, as well as between large and small

shareholders. Therefore, this thesis concentrates on dividend payout policy and its

interaction with ownership structure, abnormal stock return and future earnings growth, as

China’s equity market was established to support its transformation towards a market

oriented economy after significant economic reforms were enacted.

Most studies on dividend policy in the Chinese market4 are associated with its unique

features, especially the highly concentrated ownership and the non negotiable shares

under the split share structure.5 For the non negotiable shareholders, cash dividends are

the only justified means to realize their investment returns.6 Some empirical studies, such

as Gul (1999), Cheng et al. (2009) and Huang et al. (2011), take the state owned non

negotiable shares as the proxy of the influence of large shareholders, and report non

negotiable shareholders’ preference for cash dividends. Unlike the aforementioned studies,

the first research question of this thesis will focus on the relationship between cash

3 Because the Chinese equity market was designed to support the entrenched state owned enterprises, mostChinese listed firms adopted a residual dividend policy, that is, investment opportunity overrides the cashdividend payout. In other words, cash dividend policy took second place.4 Dissimilar to United States and other developed markets, the Chinese stock market does not treat sharebuyback (repurchase) as a type of profit distribution. Therefore, in this thesis, dividend policy and payoutpolicy in the Chinese market are interchangeable.5 Split share structure refers to the separation of negotiable and non negotiable shares, which is a legacy ofpartial share issuance privatization. State owned shares and legal person shares stemmed from state ownedshares, which used to be non negotiable until the split share structure reform (SSSR) in 2005.6 Non negotiable shares were forbidden to be traded in the secondary market because the liquidation of thelarge block shareholdings would crush the newly established Chinese stock market. Founding state ownedshareholdersmade this promise to justify the high initial public offering price. Otherwise, no external investorswould have taken part in the partial SIP because they would certainly have been disadvantaged in everyaspect, including corporate control and wealth effect, without this promise.

4

dividend payout and control rights, defined as the direct and indirect shareholdings, of the

ultimate controlling shareholders (UCSs). Compared with cash rights which only include the

direct shareholdings, control rights are better measurements of ultimate controlling

shareholders’ influence on the listed firm. Moreover, after the split share structure reform

(SSSR) was completed and all common shares became negotiable, the negotiability of

common shares was no longer a special feature of the Chinese equity market.

Besides the level of control rights, this thesis also lays emphasis on the types of

ultimate controlling shareholders, as the diversity of Chinese investors has improved

significantly in recent years and the traditional partition of state owned/private investors

can be extended as different types of state owned investors exhibit significant divergence

in their background and operating style (Chen et al., 2009c). For example, a large proportion

of state owned investors still maintain the status of government agencies with

responsibilities beyond economic profits. Intuitively, listed firms with this type of ultimate

controlling shareholder are exposed to lower agency costs as the managers are bonded

politically. However, the minority shareholders in these firms are subject to more

expropriation risk, because the managers are more likely to collaborate with controlling

shareholders. In the meantime, divergence also exists among incorporated state owned

investors as they are affiliated with different levels of government. As per Sun and Tong

(2003), the share issuance privatization in China adopted the policy of ‘‘taking a firm grip on

the large, letting go of the small’’, which meant medium and small state owned enterprises

were floated on the stock market while the government maintained tight control of the

large state owned enterprises. Some of these large state owned enterprises have gradually

become shareholding companies and have received the state owned shares of those

5

floated enterprises from multiple levels of government. In general, the qualities of state

owned enterprises are positively correlated with the level of government, and the state

owned shareholding companies affiliated with central government are deemed to be most

powerful. Not only are they entitled to great support from the central government, but they

are also more likely to own shares in the listed firms with better quality and stronger

operating performance. In addition, they are preferred by financial institutions as their

affiliation with central government enhances their credibility. In comparison, state owned

shareholding companies affiliated with local government usually have less autonomy and

the firms under their control are of inferior quality. The above mentioned differences are

not fully reflected in the previous studies about dividend payout policy and ownership

structures in the Chinese equity market. Therefore, this thesis attempts to fill in this gap

and explore whether the type of ultimate controlling shareholders influences the dividend

payout policy of underlying firms.

The second research question of this thesis is to investigatewhether Chinese investors

respond resiliently to dividend announcements, including stock dividends and changes in

cash dividends. In developed markets, the unexpected cash dividend announcement is

deemed to contain private information from the managers and/or insider shareholders

(Bhattacharya, 1979, Miller and Rock, 1985, John and Williams, 1985, Denis et al., 1994,

Ghosh and Woolridge, 1988, Divecha and Morse, 1983, Lonie et al., 1996, Yoon and Starks,

1995, Amihud and Li, 2006, Graham et al., 2006). Similar studies have been conducted on

the Chinese market, which document that stock dividends7 are more welcomed by

7 In the United States, stock dividends are regarded as a tool to convey managers’ private information, as perthe retained earnings hypothesis (Crawford et al., 2005, Bechmann and Raaballe, 2007), but it relies on theunique accounting principles of the US market.

6

negotiable shareholders8 than cash dividends (Zou et al., 2008, Chen et al., 2009a, Chen et

al., 2002, Yi et al., 2007, Anderson et al., 2011, Cheng et al., 2009). This second research

question follows the methodology adopted in Cheng et al. (2009) and connects it with the

ownership classification in Chen et al. (2009c). The market reaction towards unexpected

dividend announcements from firms with various types of ultimate controlling shareholders

may show divergence if investors are aware of the different backgrounds and features of

these UCSs and incorporate these factors into firm valuation. Further, an unexpected

change in cash dividend announcement from listed firms with different types of ultimate

controlling shareholders may also convey different information due to the fundamental

divergence among these ultimate controlling shareholders. The scrutiny of this research

question will shed more light on whether the Chinese investors regard the cash dividend

not only as a distribution of net profit but also as a method to force the managers to pay

out excessive cash balances and reduce the agency cost.

Apart from the determination of and investors’ reaction towards cash dividends, the

third research question of this thesis is to investigate how the cash dividend payout

influences future earnings growth. As per the traditional ‘pecking order’ theory, a higher

payout ratio leads to lower retention and increases the future cost of funds, which

undermines the future operating performance (Gordon, 1962, Myers, 1984, Rozeff, 1982,

Fama and French, 2002, Ibbotson and Chen, 2003). Some contemporary opinions, however,

indicate there might be a positive correlation between the payout ratio and future earnings

8 The sample window of most studies on the dividend announcement effect in the Chinese market, includingthis thesis, covers the period before SSSR, and the share trading in the secondary market is dominated byindividual investors and private institutional investors, while non negotiable shareholders are not allowed tobe involved.

7

growth as the dividend payout reduces free cash flow and restricts managers’

overinvestment activities. In other words, the underlying firm’s operating performance will

benefit from reduced agency costs of free cash flow (Arnott and Asness, 2003, Zhou and

Ruland, 2006, Gwilym et al., 2006, Huang et al., 2009, Vermeulen and Smit, 2011). With

regard to the Chinese market, there is not yet any literature covering this issue. Therefore,

this third research question will supply some empirical evidence on the relationship

between cash dividend payout and future earnings growth in a transitional economy. The

results may well suggest the significance of agency costs in the Chinese market and provide

a methodology for forecasting the earnings growth based on the dividend level.

1.2. Research Scope

As discussed in the preceding subsection, the main objective of this thesis is to

perform empirical research on how dividend policy interacts with corporate control rights,

announcement effect and future earnings growth in the Chinese stock market. This thesis

is scheduled as follows.

Chapter 2 supplies a review of the existing literature in order to establish three

themes of research that have dominated the dividend policy literature. The first part of this

chapter outlines the first focus of research pertaining to the relationship between corporate

control rights and dividend policy. The second part includes the literature on the

relationship between dividend payout, corporate control rights and stock abnormal return,

while the last part presents the relationship between dividend payout policy and

subsequent equity financing. Based on the traditional dividend theory and agency theory,

the relationship between dividend policy and controlling shareholders is relatively

8

persistent in a developed market (Truong and Heaney, 2007). The literature on dividend

announcement effect is also quite consistent, that is, the dividend announcement carries

private information from corporate insiders (Grinblatt et al., 1984, Bechmann and Raaballe,

2007, Cheng and Leung, 2008). It is arguable, though, whether the cash payout has positive

or negative impacts on the future earnings growth (Fama and French, 2002, Zhou and

Ruland, 2006, Huang et al., 2009). With all three focuses of research, the literature review

in the following chapter highlights the importance of dividend payout policy as a solution

to the agency cost in the corporate governance area. Specifically, the potential connection

between dividend payout policy and ownership structure, as well as the impact of dividend

payout on stock abnormal return and future profitability, form the main themes of this

chapter, which instructs the design of empirical research in the following chapters.

Chapter 3 provides a brief review of the history and legal framework of the Chinese

stock market, as well as the dividend payout practices of Chinese listed firms. The capital

market in the Chinese economy was designed to support the state owned enterprises to

convert from the subsidiaries of multiple level governments to independent entities via

partial share issuance privatization (SIP). Meanwhile, the Chinese civil law legal system is

influenced by its unique political infrastructure and generally regarded as weak in terms of

investor protection (La Porta et al., 1998). Consequently, the pyramid structure has become

a popular format to facilitate controlling shareholders to redirect the listed firms’ financial

resources to other controlled entities. Recently, the securities regulatory authority, China

Securities Regulatory Commission (CSRC), promulgated some regulations which are tough

on the listed firms and relevant controlling shareholders, such as restricting controlling

shareholders from occupying listed firms’ financial resources.

9

Chapter 4 conducts an empirical study of how the level of control rights and the types

of UCSs affect cash dividend policy among Chinese listed firms, based on a sample between

2003 and 2010. Specifically, this chapter connects and extends the works of Chen et al.

(2009c) and Cheng et al. (2009), which investigate the relationship between ownership and

operating performance, and the connection between negotiable/non negotiable shares

and dividend preference, respectively. The likelihood and the magnitude of cash dividend

policy are examined in both univariate and multivariate analysis. The propensity of cash

dividend policy is proxied by a dummy variable and the magnitude of cash dividend policy

is measured by the cash payout ratio. UCSs’ control rights are introduced as the proxy of

ownership. In addition, four dummy variables, which indicate the types of UCSs, are

included along with the level of control rights. The analysis suggests that both the

propensity and the magnitude of cash dividend payout are positively associated with the

control rights, consistent with the findings in Cheng et al. (2009) that non negotiable

shareholders prefer cash dividends. Similar to Chen et al. (2009c), different types of UCSs

are found to have divergent impacts on the propensity and magnitude of cash dividend

payouts. Among the listed firms with state owned ultimate controlling shareholders, listed

firms controlled by state owned enterprises affiliated with central governments (SOECG)

have a higher probability of announcing a cash dividend, while listed firms controlled by

state owned enterprises affiliated with local governments (SOELG) have a lower probability

of announcing a cash dividend. In terms of cash dividend magnitude, listed firms with state

owned UCSs do not exhibit outstanding cash dividend payout ratio, while listed firms

controlled by the State Asset Management Bureau (SAMB) report a lower cash dividend

payout ratio than other firms in the sample. Contrary to the traditional thinking, listed firms

10

under the control of private investors show a significantly higher cash dividend payout ratio

and a marginally larger probability of making a cash dividend payout than firms with state

owned ultimate controlling shareholders. This empirical chapter contributes to propose

control rights, rather than non negotiable shareholdings, to be used as the measurement

of influence from large shareholder, as well as improve the traditional state private

segregation of shareholders in the Chinese market with a scrutiny into the various types of

state owned large shareholders.

Chapter 5 undertakes an event study on the stock and cash dividend announcement

effect, as well as the divergence in announcement effect among the firmswith various types

of UCSs. As the past literature finds that Chinese investors react more positively towards

stock dividend announcements than cash dividend announcements, this chapter explores

whether this pattern has continued since the Securities Law was enforced in 1999. The

announcement effect is measured by the cumulative abnormal return across different

event windows. The stock dividend is proxied by a dummy variable, while the cash dividend

shock is measured as the unexpected cash dividend adjusted by industry average growth. It

is found that the cumulative abnormal return responds positively to the stock dividend

announcement, especially in the case of stock dividends from capital reserves. This finding

contributes to the literature in Chinesemarket, which, for the first time, separates the stock

dividend from retained earnings from the stock dividend from capital reserve.

Meanwhile, unexpected cash dividends have a marginally positive contribution to

cumulative stock abnormal returns. The results are consistent with past literature that

minority shareholders prefer stock dividends while non negotiable shareholders prefer

cash dividends (Cheng et al., 2009). Meanwhile, UCSs’ control rights have little influence on

11

the cumulative abnormal return, but the unexpected cash dividends from firms with

different types of UCSs lead to divergent cumulative stock abnormal return. Specifically,

market reaction to unexpected cash dividends from firms with private or SAMB UCSs is

generally positive in the short and medium event windows. Dissimilar to previous studies,

the earnings shock has no significant impact on the cumulative abnormal return. The above

finding makes contribution to the dividend announcement effect literature in the Chinese

market and examines the effectiveness of CSRC’s regulations on cash dividend payout.

Chapter 6 conducts an empirical analysis of the relationship between current cash

dividend payouts and future earnings growth in the Chinese market, based on a sample

taken between 2001 and 2010. This chapter extends the works of Arnott and Asness (2003),

Zhou and Ruland (2006) and Vermeulen and Smit (2011) to the Chinese market with the

purpose of investigating whether cash dividend policies of Chinese public firms follow the

pecking order theory or the free cash flow hypothesis. Further, this chapter explores a

potential quadratic relationship between cash dividend payouts and future earnings

growth, which can be regarded as a combination of pecking order theory and free cash

hypothesis. Additionally, the influence of stock dividend announcements on the future

earnings growth is also examined.

Following the aforementioned literature, the future earnings growth is measured by

growth in earnings per share and earnings before interest and tax per share over a one year

window. The results indicate that the cash payout ratio and a change in cash payout ratio

has a positive influence on the future earnings growth, which is generally consistent with

Arnott and Asness (2003), Zhou and Ruland (2006) and Vermeulen and Smit (2011). The

future earnings growth of firms with higher over investment potential, proxied by Tobin’s

12

Q, benefits more from higher cash dividend payout. The proposed non linear relationship

between cash dividend payout and future earnings growth is not significant, although the

contribution from cash payout to future earnings growth diminishes marginally as the cash

payout increases. Finally, a stock dividend announcement does exert a negative impact on

the future earnings growth, which is in line with the free cash flow hypothesis. This chapter

makes contribution to interpret the cash dividend as a tool to deal with the agency principal

problem, as well as supply an explanation to the widely discussed ‘stock dividend puzzle’ in

the Chinese market.

Chapter 7 provides a summary and conclusion. In the traditional corporate finance

research, dividend policy is associated with firm valuation and corporate governance

mechanisms. The results suggest that dividend policy manifests itself as a mechanism to

mitigate the free cash flow problem and improve the efficiency of operations in the Chinese

market. Importantly, dividend policy is affected by the ownership structure and

consequently exerts an influence on the operating performance and firm valuation. These

findings suggest some further research into the agency cost and relevant tunnelling

behaviour of ultimate controllers is necessary.

13

Chapter 2 Literature on dividend payout policy

2.1. Introduction

This chapter provides a literature review of the previous studies on the three main

research themes covered in this thesis. Section 2.2 briefly outlines Lintner’s model and

Miller and Modigliani’s dividend irrelevance theorem. Section 2.3 outlines the first theme

pertaining to the relationship between cash dividend, agency cost and corporate control

rights. The first part of Section 2.3 addresses the studies on the theoretical views about the

relationship between cash dividend, agency cost and corporate control rights, while the

second part of Section 2.3 elaborates on the signalling hypothesis of cash dividends,

whereas the last part of Section 2.3 discusses the relationship between dividend policy and

ultimate controlling shareholders. Section 2.4 reviews the literature on the dividend

announcement effect, which is comprised of two parts the theoretical viewpoints and the

empirical evidence. Section 2.5 deals with the second theme, the research on the

relationship between dividend policy and future earnings growth, which consists of two

parts. The first part of Section 2.5 focuses on the free cash flow hypothesis of dividend

policy, whereas the second part covers the empirical research on how dividend policy is

correlated with future earnings growth. Finally, Section 2.6 summarizes the literature on

which the following chapters are based.

2.2. Lintner model and dividend irrelevance theorem

Lintner (1956) proposed that managers designed the dividend policy deliberately

rather than simply distribute the residual of net profit after reserving sufficient funds for

14

future investment plans. Managers would maintain a steady dividend payout unless they

could confirm that the change in the updated earnings level was sustainable. Furthermore,

Lintner (1956) posited that managers are concerned more about the target payout ratio

than the absolute changes in earnings amount. Fama and Babiak (1968) confirm the

robustness of net profit as an earnings parameter and supply empirical evidence to support

the Lintner model with little serial dependence on the disturbances, based on a sample of

United States industrial firms. Ang (1975) suggests the coherence between dividends and

earnings is faster than normally expected in the long run and short run components. Kalay

(1981) document a negative relation between earnings uncertainty and payout ratio in

cross sectional tests, but a similar correlation is not observed in the time series analysis.

Moreover, based on a worldwide firm level sample, Chay and Suh (2009) conduct cross

sectional analysis and report a significant negative association between cash flow

uncertainty, proxied by the volatility in stock returns, and cash payout policy, including both

the magnitude and propensity of the cash dividend.

Besides the traditional cash dividend payout, share repurchase has become more

popular in recent years. Brav et al. (2005) conducted a survey among 384 financial

executives and found share repurchase was preferred, not only because of its flexibility in

distributing residual cash flows, but also its function of adjusting the capital structure

instantaneously. Skinner (2008) reports only a small portion of public firms contribute to

the aggregate cash dividend payout and argues that the traditional version of the Lintner

model is unable to capture the evolution in payout practice because the sticky dividend

policy results in a weakening correlation between dividends and earnings. Instead, the

speed of adjustment improves over time and becomes more significant when the payout is

15

measured by total cash distribution, including share repurchase. Some empirical evidence

suggests that newly listed firms arewary of initiating cash dividends and the amount of cash

dividend payouts is concentrated within a relatively small number of large firms (Fama and

French, 2001, DeAngelo et al., 2004, Skinner, 2008).

Another cornerstone in dividend literature is the Miller Modigliani (MM) dividend

irrelevance theorem. Miller and Modigliani (1961) put forward the theory that

shareholders’ wealth level is only determined by the firm’s investment policy but is not

affected by the dividend policy, given a frictionless market. Similar to Lintner’s suggestion

of a target dividend payout ratio, Miller and Modigliani (1961) theorize that the change in

current dividend will convey some information about future target payout ratios. It is the

shareholders’ speculation on this information, rather than the dividend policy, that drives

the share price. Miller and Modigliani (1961) also discuss various types of market

imperfections, such as income tax and transaction costs, and shed light on the directions of

subsequent payout policy research, such as tax clientele effect and agency cost theory.9

TheMM dividend irrelevance theorem has been continuously debated, especially its

assertion that only investment policy contributes to firm value in a frictionless market.

DeAngelo and DeAngelo (2006) posit that dividend policy is as important as investment

decisions in firm valuation, and the optimal payout ratio is within a range rather than at a

certain level when the retention of free cash flow is allowed. In the framework of MM’s

dividend irrelevance theorem, managers have to pay out the present value of future free

cash flow if the decision is in the best interests of shareholders. DeAngelo and DeAngelo

9 The agency cost theory can be subdivided into an incomplete contract (free cash flow problem) and anasymmetric signalling hypothesis (asymmetric information).

16

(2007) argue MM’s dividend irrelevance theorem may well produce some side effects

because the above assumption can be easily challenged by agency principal problems and

stock bubbles.10

2.3. Dividend payout policy, agency cost and corporate governance

2.3.1. Agency cost and dividend policy

The separation between ownership and control is the source of the agency cost

theory and modern corporation theory in financial economics (Ross, 1973). Entrepreneurs

with intelligence capital and investors with financial capital form a collaborative relationship

in which the investors are the principal and the entrepreneurs work as the stewards.

Shareholders delegate the firm resources to entrepreneurs (directors) with expertise and

professional judgement, whichmakes directors the actual controller of the firm’s assets and

they are assumed to work in the best interests of the principal. However, as the assumption

of homo economicus11 holds, the entrepreneurs’ interest is not always aligned with that of

the shareholders’, which results in the potential conflict of interest in which the directors

will maximize their own benefits at the cost of shareholders (Jensen and Meckling, 1976).

Because it is unrealistic to develop a perfect contract in detail to define the rights and

responsibilities of both principal and agent, shareholders have to utilise other methods to

10 Agency principal problem caused by separation of ownership and control results in managers’ preferencefor retaining more free cash flow and conducting self dealing activities at the expense of shareholders.Irrational exuberance in the stock market tempts managers to issue new shares at overvalued prices, evenwhen there is no ideal investment opportunity.11Homo economicus refers to rational decision making purely based on the principle tomaximise self interest.

17

minimise this problem and incur agency cost, such as a competitive remuneration package,

monitoring of costs and residual losses.12

The existence of the agency principal relationship installs managers as the actual

controller of the firm’s assets who are likely to abuse the free cash flow, which is ‘the cash

flow in excess of that required to fund all projects that have positive net present value’

(Jensen, 1986). External investors may tackle this free cash flow problem by demanding

more cash dividends.13 Rozeff (1982) posits an optimal dividend payout ratio to minimize

the aggregation of agency costs and floatation costs of external equity funding. To reduce

the cash holding by paying cash dividends will mitigate the agency cost, as the managers’

inclination to have a spending spree (e.g., over invest) is suppressed, but the side effect is

that the firm has to raise external capital when projects with positive net present values

emerge. There is a trade off between them though as the benefit from decreased agency

costs may well be overridden by the higher floatation costs of subsequent equity financing.

Crutchley and Hansen (1989) outline three methods to deal with agency cost,

including managerial shareholdings, cash dividend and debt financing. Managers would

adopt the least costly financial policy to reduce the agency cost and increase firm value for

both shareholders and themselves. It is found that managers are not inclined to use cash

dividends when the floating cost is high, but are likely to distribute more cash when firm

size is larger, which is in line with the agency cost hypothesis. Denis et al. (1997) conclude

12 Monitoring cost refers to the dissipated resources spent on supervising the directors, such as auditing fees.Residual loss is the economic consequence of managers’ decisions which fail to maximize shareholders’wealth. Large shareholders may face agency cost stemming from ‘free rider problems’, that is, minorityshareholders will take advantage of the conflict between large shareholders and directors. Entrepreneurs alsoincur ‘bonding cost’ to affirm their commitment to the company. Jensen andMeckling (1976) also explore theagency cost of debt.13 Debt cost is another solution with similar purpose to decrease the cash balance. But it may lead to theagency cost between shareholders and debt holders.

18

that the firm value is undermined by the size of managerial equity ownership and the

presence of external block shareholders, because these factors are inversely related to the

level of diversification.

Simultaneously, Easterbrook (1984) suggests the divergence in the level of risk

aversion between shareholders and managers leads to another type of agency cost. As

entrepreneurs’ human capital and interest in the firm is less diversified, they are more risk

averse than investors who prefer taking more financial risks to exploit the tax shield from

debts. Similar to Rozeff (1982), the primary market will discipline managers’ behaviour

when they pursue additional external equity funds. Even if no additional external equity

capital is needed, a cash distribution will alleviate the divergence in the risk appetite

between shareholders and entrepreneurs. By contrast, Blau and Fuller (2008) posit that

corporate management incorporates the operating flexibility into the design of dividend

policy. Fundamentally, the agency principal relationship assumes managers have more

expertise than shareholders. By decreasing dividend payments and storing liquidity,

managers obtain flexibility which enables them to be engaged in profitable investment

projects, but shareholders may not agree with the managers’ decision because fewer

dividends are unappealing. Further, shareholders will supply further equity funds when the

stored liquidity is insufficient, therefore, managers face a trade off between investment

flexibility and the current dividend.

Some recent studies argue that it would be hard for managers to redirect financial

resources as the variability of future cash flow makes it unverifiable. In other words, the

managers’ control on the firm assets may be very expensive. Therefore, the focus of

research on agency cost and dividend policy turns to the relationship between insiders and

19

outside equity. Fluck (1998) investigates the function of outside equity and argues that

agency cost can be alleviated by debt and outside equity, because investors will evaluate

the probability of whether their future cash rights can be realized, and their decision to

purchase the outside equity indicates they are satisfied with the expected outcome. Not

only does debt impose fixed financial charges on management as a discipline, but also puts

the managers’ fate in the debt holders’ hands when the put option on the firm assets held

by outside equity investors is exercised in the case of bankruptcy. Entrepreneurs pay cash

dividends to outside equity holders in order to convince them, to some extent, that their

future cash rights are honoured.

Unlike Jensen and Meckling (1976), Myers (2000) argues that managers pay cash

dividends to external shareholders in order to extend the agency principal contract,

because the future return on investment is subject to market uncertainty, and managers

are attempting to mitigate a negative response from shareholders. In other words, cash

dividends are introduced as a mechanism to sort out the agency cost as it ensures the

investors will hold the investment longer; and outside equity becomes a supplementary

solution when shareholders receive sufficient cash dividends.

The empirical evidence on the agency cost of free cash flow has been widely

documented in both developed and emerging markets. Born and Rimbey (1993) review the

hypothesis in Easterbrook (1984), that the signal embedded in the cash dividend change is

ambiguous because it relies on investors’ capability to distinguish growing firms from

disinvesting firms (i.e., mature firms). Their outcomes indicate that firms with subsequent

financing are more likely to pay fewer cash dividends, but the market reaction to these

20

firms’ announcements is much larger compared with the latter, which provides supportive

evidence to underpin the agency cost model developed by Easterbrook (1984).

In a sample of Canadian firms with a large percentage of internal shareholders, Eckbo

and Verma (1994) report significant agency costs exist when owner managers hold the

dominant voting power in the firm, because such firms always pay low cash dividends or

even omit the cash distribution altogether. They conclude the cash dividend is the result of

consensus among various shareholder groups with heterogeneous backgrounds.14

Mercado Mendez and Willey (1995) observe that the cash dividend policy of the

banking industry is positively influenced by bank size, but negatively associated with poor

diversification, which means the management of banks partially takes the agency cost into

account when they design the financial policy. Similar to the banking industry, utility

companies are also strictly regulated and different from other firms in many features, such

as their capital structure and the appointment of senior management (Hansen et al., 1994).

The utility industry always faces political pressure as any change in ratemay cause rebounds

from the authority and the public. Besides the traditional agency shareholder problem, the

shareholder regulator conflicts also influence the firm’s financial policies. Hansen et al.

(1994) argue that equity finance following a large dividend distribution puts the firm under

the microscope of the underwriter, whose due diligence report will work as an effective

check on the details of the firm. Consistent with the agency cost hypothesis, their empirical

results suggest the floating cost of new equity and ownership concentration has negative

impacts on the dividend payout ratio of utility firms.

14 Eckbo and Verma (1994) also indicate some evidence on the tax clientele effect because the magnitude ofthe cash dividend will increase when the firm is controlled by institutional shareholders with a lower marginaltax rate.

21

Lie (2000) investigates firms’ decisions to deal with excess cash balances by three

types of announcements, special dividend, regular dividend increase and self tender offer.15

Special dividend and self tender offers are generally used to deal with the excess cash from

non recurring events, while a regular increase in cash dividends follows the increase in cash

due to recurring events. Empirical results suggest investors react actively to self tender

offers and special dividends, but sluggishly to regular dividend increases and small special

dividends, which supplies evidence to the free cash flow hypothesis that a large cash

distribution is to alleviate the agency problem.

Brockman and Unlu (2009) look into the substitution between debt and cash dividend

policy by investigating how the credit rights across various markets influence the cash

dividend decision. They argue that managers would make compromises with creditors in

the issue of dividend policy because creditors are concerned about the potential wealth

transfer from them to shareholders. Therefore, a more conservative dividend policy would

satisfy both creditors and shareholders. Their empirical results suggest that the quality of

protection on credit rights is positively correlated with the probability and magnitude of

cash payouts. In other words, the agency cost of debt carries substantial weight when

managers decide on the cash distribution to shareholders, which underpins the reality that

creditors are more involved in corporate financial policy.

Further, Brockman and Unlu (2011) explore agency cost in the frame of the lifecycle

theory of dividend payouts. The lifecycle theory of cash dividend advocates that firms in the

growth stage are less likely to distribute cash to shareholders, while firms in the mature

15 Self tender offers are a type of share buyback, but only target a certain group of shareholders. It is a methodused to avoid hostile takeover.

22

stage are more likely to do so, because the investment return of the former’s financial

resources is much higher than that of the latter.

Apart from the empirical evidence in favour of the lifecycle theory, they confirm the

existence of an agency cost inclusive lifecycle theory of cash dividends. Because managers

need to distance themselves from potential suspicion of hoarding financial resources, they

will pay more cash dividends if the environment of transparent disclosure is not available,

which underpins the validity of cash dividends as a solution to agency costs.

2.3.2. Corporate governance and dividend policy

The research on agency cost depends on the reality in the Anglo American markets

that shareholdings are widely dispersed among investors, and managers take advantage of

this situation because it is relatively difficult for scattered shareholders to take collective

action against the management. In other markets, however, the concentration of

ownership and the presence of large shareholders have more impact on corporate

governance issues in those markets.

Shleifer and Vishny (1986) attempt to develop a model and explain how a large

minority shareholder takes the responsibility of monitoring the company’s management,

and deals with the agency principal conflicts when diffused minority shareholders are

actually free riders. Cash dividend becomes a subsidy to large shareholders because they

need to be compensated for the additional monitoring work. Subject to higher marginal tax

rates, minority shareholders are disadvantaged when they receive cash dividends, but

regard it as reasonable because they share the benefit of the improved share price

23

stemming from large shareholders’ extra surveillance of management.16 Large shareholders

can also be the source of an agency problem between large and small shareholders, as large

shareholders may collaborate with managers and expropriate small shareholders’ wealth.

Although Barclay et al. (2009) argue that there is weak evidence to support the connection

between large shareholders and cash dividend payouts, their argument targets the tax

clientele effect and does not rule out the possibility of collaboration between large

shareholders and managers.

Jensen et al. (1992) argue that the contribution of a cash dividend to alleviate agency

cost is less significant when the underlying firm reports a higher percentage of insider

ownership, because firm specific features cause the endogeneity of dividends and

ownership. Their empirical results propose that insider ownership has a significant negative

correlation with dividend payout and debt level. Similar to Jensen et al. (1992), Crutchley

and Jensen (1999) contend that capital structure, including the level of insider ownership

and institutional shareholdings, and dividend policy are designed simultaneously in order

to minimize the agency cost. They observe that institutional shareholdings work as an

alternative monitoring mechanism which is able to substitute the cash dividend. They also

report a statistically significant curvilinear relationship between dividend payout and inside

ownership.

When institutional investors gain significant power in the financial industry, the

function of dividend signalling is not only limited to conveying information on firms’

operation but is also utilized to attract institutional investors. Allen et al. (2000) point out

16 Recent studies, such as Barclay et al. (2009), document weak connections between institutionalshareholders and dividend paying firms.

24

that the presence of institutional shareholders enhances corporate governance and the

information content embedded in the dividend change. As institutional investors have the

expertise to evaluate the performance and quality of listed firms, bad firms will incur

significant costs if they attempt to improve their reputation by mimicking good firms’

payout policy. Institutional shareholders will take action against incompetent managers by

dumping their shareholdings and facilitating potential takeover activities, when they reduce

cash dividend. They conclude that both dividend policy and firm value is positively

influenced by the existence of institutional investors. Meanwhile, Chae et al. (2009) report

that the effectiveness and direction of corporate governance on the cash dividend depends

on firms’ financial constraints. Firms will reduce the payout ratio when they face tougher

external financing constraints even if corporate governance is enhanced.

Based on Lintner (1956) and Fama and Babiak (1968), Short et al. (2002) document a

positive interaction between institutional shareholders and dividend policy based on UK

panel data. The positive relationship is enhanced when the underlying firm’s profitability is

strong. In the meantime, an inverse relationship between managerial shareholdings and

dividend payout is observed, which supplies evidence to the free cash flow hypothesis. Khan

(2006) also uses a panel of 300 UK firms between 1985 and 1997 to investigate the

relationship between dividend policy and ownership structures. The result suggests a

negative and curvilinear relationship between the shareholding of the top five largest

shareholders and dividend payouts. In addition, the individual shareholdings are negatively

related to the dividend payout, while the level of shares held by insurance companies has a

positive relationship with the dividend payout. The author attributes the outcomes to either

25

the fact that cash dividends substitute for the poor corporate governance, or that powerful

shareholders exert influence on the management on the issue of corporate financial policy.

Baba (2009) reports that a larger increase in foreign ownership has a positive

influence on the likelihood of cash dividend announcements and the probability of an

increase in cash payout magnitude among Japanese listed firms. Although foreign investors

only have a relatively small stake in the Japanese market, the results indicate that active

external shareholders will exert considerable influence on the dividend policy. The empirical

results of Grinstein and Michaely (2005) find there is a significant positive relationship

between share repurchase and institutional shareholdings, while the correlation between

institutional shareholdings rendering a higher dividend payout and cash dividend payout is

ambiguous, although dividend paying companies have higher percentages of institutional

shareholdings.

The legal environment is another important issue which affects cash dividend policy.

In general, the quality of legal protection will determine to what extent the shareholders’

property rights can be endorsed, and mitigates the damage to their wealth caused by

agency costs. In other words, besides the conventional corporate governance

infrastructure, an effective legal system becomes a back up solution to the agency problem.

La Porta et al. (2000a) conduct a cross border analysis on how legal systems influence the

dividend policy. Two testable models, ‘outcome model’ and ‘substitute model’, are

developed to explore how firms design their dividend policy. The ‘outcome model’ posits

that the dividend policy is an outcome of an effective legal system on the issue of

shareholders’ property rights. The payment of a cash dividend is the result of minority

shareholders exercising their entitlement to extract cash from the firm, which shows the

26

power of the legal system to improve the self motivated corporate governance system.

Rather than the causal relationship, as proposed in the ‘outcome model’, the ‘substitute

model’ hypothesizes that cash dividends and legal protection are substitutes. The

foundation of this hypothesis is that the subsequent equity finance needs the firm to

establish a remarkable reputation in the capital market in order to receive decent treatment

from the investors, including both active participation in seasoned equity offerings and

lower funding costs. The ‘outcome model’ predicts that dividend payout ratios should be

higher in those countries with stronger corporate governance.

Based on the regression results from a sample of 4,103 dividend paying public firms

from 33 legal regimes, La Porta et al. (2000a) document there are significant differences

between the dividend payout patterns of civil law regimes and that of common law regimes.

The empirical evidence is inclined to support the ‘outcome model’ as providing stronger

legal protection (in common law regimes) and it is associated with higher cash dividend

payouts. Meanwhile, they also report a side effect of this relationship which is the lower

growth rate in common law regimes, as the minority shareholders are more likely to

withdraw cash from the listed firm which may well undermine the long run growth.

Farinha and Lopez de Foronda (2009) compare the relationship between dividend

policy and insider ownership in different legal regimes with civil law and common law legal

systems. In common law regimes, the cash dividend payout is, in general, negatively

correlated with insider ownership in a common law legal system. During this descending

process, payout ratio will rebound when the insider shareholdings are significant but non

dominating, which reflects the insider shareholders’ attempt to compromise outside

investors’ demand for decreasing the cash controlled by insider owners. The pattern is

27

totally opposite in civil law regimes, where cash dividend increases as the internal

shareholding increases, but with a retrace during the ascending process. From the author’s

viewpoint, the divergent pattern suggests dividend policy works as a corporate governance

mechanism in regimes with different legal systems and distinct agency costs.

Adjaoud and Ben Amar (2010) test whether outcome hypothesis or substitution

hypothesis is more suitable to explain the relationship between dividend policy and

corporate governance quality, with a sample of 714 observations listed on the Toronto

Stock Exchange, and find supportive evidence that dividend policy is a kind of corporate

discipline mechanism. They use the Globe & Mail annual corporate governance index, as

well as four sub category scores board composition, shareholding and compensation

issues, shareholder rights issues and corporate governance disclosure policy to assess the

robustness of corporate governance mechanisms. Their regression results indicate that

strong corporate governance is associated with a higher dividend payout ratio, which is

similar to Farinha and Lopez de Foronda’s (2009) findings on the common law regime.

Jiraporn and Ning (2006) posit agency costs play a determinant role when firms design

a dividend policy, by examining the association between dividends and the strength of

shareholder rights. Contrary to La Porta et al. (2000a), their empirical evidence is inclined

to underpin the substitution model, that is, there is an inverse relationship between

dividend payout and shareholder rights. Firms would announce higher dividends when

shareholder rights aremuchweaker, as themanagement needs to convince themarket that

they would not expropriate shareholders’ wealth. The result is that dividends replace

shareholder rights as a corporate governance mechanism.

28

Some studies argue the ‘outcomemodel’ proposed by La Porta et al. (2000a) may only

be suitable for markets where the shareholding is widely dispersed. In fact, the

phenomenon of shareholding concentration is widely seen in markets other than UK and

US. Gugler and Yurtoglu (2003) contend that the change in dividend payout reflects the

severity of the conflict between large and small shareholders.With a sample of 266 German

listed companies, they find that the payout will be reduced when the largest shareholder

increases its stake, but is increased when the second largest shareholder acquires more

outstanding shares. The first result indicates the free cash flow problem deteriorates as the

largest shareholder has more voting rights, while the second phenomenon suggests the

second largest shareholder attempts to challenge the largest shareholder’s monopoly on

the firm’s financial resources.

Besides the level of shareholdings, Gugler (2003) contends that the identity of the

largest shareholder influences the underlying firm’s payout policy. Gugler (2003) defines

the four types of largest shareholders as families, banks, foreign investors and state

governments, because they have unique backgrounds and interests. A system of

simultaneous equations is applied to the panel data of 214 non financial Austrian firms

between 1991 and 1999. Significant divergence among the different types of largest

shareholders is documented. The listed firm with a state government as the largest

shareholder smooths the dividend policy, while family controlled firms exhibit more

volatility in the dividend payout and are more likely to cut cash dividends. The dividend

smoothing activity, however, is only marginally important in firms with banks or foreign

investors as the largest shareholder. The investment cash flow, proxied by the research and

29

development expenditure, is inversely related to the dividend policy, which supports the

maturity hypothesis that mature firms will distribute the cash to shareholders.

Faccio et al. (2001) argue that the existence of large shareholders has changed the

payout pattern significantly. Investors in both European and Eastern Asian firms tightly

affiliated with single large shareholders receive more cash dividends, because they are

aware of the potential for expropriation. When the affiliation is loose, though, the cash

dividends of European firms are not affected, whereas similar Asian firms reduce their cash

payout because multiple large shareholders, also known as ‘parties acting in concert’,

collaborate and exacerbate the expropriation.

Renneboog and Trojanowski (2007), using UK panel data, investigate whether

corporate control structure influences the dividend policy. Their results show a significant

negative relationship between the largest shareholders’ voting rights and payout ratios, and

the pattern is irrelevant to the type of largest shareholder. Meanwhile, the traditional

connection between earnings and cash payout, as suggested by Lintner (1956), still holds,

but is undermined by the existence of a large shareholder and/or a coalition among large

shareholders. Large shareholders attempt to optimize the aggregate costs from both free

cash flows and subsequent equity financing. Not only does the result support the gradual

adjustment in dividend payout (i.e., smoothing the dividend), but it also underpins the

pecking order theory where the funding cost is more relevant when management design

the payout policy.

Wei and Zhang (2008) analysed eight East Asian emerging markets before the Asian

financial crisis, and found that the sensitivity of a firm’s capital expenditure to its free cash

flow is negatively correlated with the cash rights held by the largest shareholder, but

30

positively correlated with the magnitude of the divergence between the largest

shareholders’ cash rights and control rights. These results can be explained by the free cash

flow hypothesis, because the largest shareholder would occupy the free cash flow of listed

firms and expropriateminority shareholders’ wealth. As the divergence between cash rights

and control rights enlarges, the controlling shareholder has more incentive to exploit

additional economic benefits from the underlying firm via overinvesting its free cash flow.

With regard to the relationship between dividend policy and ownership structure, Gul

(1999) reports that government ownership is positively associated with Chinese listed firms’

cash dividend payouts, based on a sample of Chinese listed firms between 1991 and 1995.

Similarly, Cheng et al. (2009) document a positive relationship between state owned shares

and cash dividend payouts, as well as a positive relationship between private shareholdings

and stock dividend.17 Further, Huang et al. (2011) also report a positive contribution from

non negotiable shares on both the magnitude and likelihood of cash dividend

announcements with a sample between 1994 and 2006. All these studies attribute the

positive relationship between non negotiable shares and cash dividend to the unique

ownership structure in the Chinese equity market, especially the non negotiability of state

owned shares, because a cash dividend is the only justified form of investment return from

the viewpoint of non negotiable shareholders.

17 Stock dividends in the Chinese market contain two types, stock dividend from retained earnings and stockdividend from capital reserves, which are similar to bonus shares and stock splits in a developed market,respectively.

31

2.4. Dividend policy and stock abnormal return

2.4.1. Dividend announcement effect: US and other developed markets

The relationship between dividend policy and stock abnormal return is a test of the

efficient market hypothesis (EMH), and themainstream research on the stock price reaction

to dividend announcements surrounds the information content embedded in dividend

change, omission and initiation. Large firms are reluctant to change dividends unless the

managers are assured about the sustainability of operating profits (Lintner, 1956). Investors

are likely to take the dividend announcement as an objective indicator of firm performance,

because external stakeholders lack detailed information about the firm apart from that

obtained from financial reports. Assuming the management conveys information via

dividend announcements, investors react vigorously to the change in dividend policy. On

the other hand, this mechanism is also known by managers who then design the dividend

policy carefully in order to avoid any undesired impacts on the firm value.

Pettit (1972) attributes the ‘sticky dividend policy’ to themanagement’s concerns that

the announcement of dividend changes will be used to assess share price. Based on a

sample of 625 dividend change announcements between 1964 and 1968, the author finds

investors react negatively to dividend decreases and positively to moderate dividend

increases (10 to 25%).18 The results of Pettit (1972) are affirmed by Kwan (1981) who applies

the models in Lintner (1956) and Fama and Babiak (1968) on a small sample with better

specification of information content. Higher unexpected changes in dividend (larger than

10%) are associated with higher cumulative abnormal returns around the event window.

18 But investors’ reaction becomes sluggish if the dividend increase is smaller than 10% or larger than 25%.

32

Bhattacharya (1979) proposed that managers use the cash dividend as an opportunity

to signal the expected cash flow because of the different tax rates on dividend income and

capital gain. Intuitively, investors respond differently based on the direction of news. A

positive cash dividend shock or dividend initiation is viewed as a signal of the manager’s

optimism about future cash flow, while a lower than expected cash dividend

announcement shows the managers’ lack of confidence. The argument in Bhattacharya

(1979) is that it treats the dividend, to some extent, as a commitment to shareholders,

although investors should be aware that they are only entitled to the residual value.

Aharony and Swary (1980) use quarterly data from the US capital market, and show

the information content of changes in dividend is more useful in conveying information to

the public than that from the corresponding earnings announcement. The stock abnormal

return around the dividend announcement independent of the earnings announcement is

significantly positive (negative) when there is a dividend increase (decrease). Moreover,

with a sample across the window between 1947 and 1968, Charest (1978a) finds that the

monthly return after the dividend change announcement is consistently large, especially in

the case of a dividend decrease, which suggests a significant market anomaly. The author

attributes the result to the under reaction of investors, but cannot rule out the possibility

of measurement error in the abnormal return. On the other hand, Divecha and Morse

(1983) also document a significant stock abnormal return when there is an increase in cash

dividends, but their sample is restricted to a shorter window from 1977 to 1979.

Woolridge (1982) attempts to investigate the information content of dividend

changes independent from the investors’ expectation on dividends and the noise of

simultaneous earnings announcements. Their research design focuses on unexpected

33

dividends after explicitly controlling for the earnings shock. The sample contains 200 shares

and 376 dividend change announcements, and the regression results indicate dividend

shocks are positively related to cumulative abnormal return, which supports the

information content hypothesis of dividends.

Benesh et al. (1984) supply further empirical evidence on the hypothesis that

managers transmit private information by their design of dividend policy. Based on the

sample of substantial changes in cash dividends (initiation, omission and no less than 25%

change), the summary statistics of cumulative abnormal returns indicate that negative

dividend shocks have a prominent downward impact on stock prices even though the

market has anticipated the forthcoming news to a large degree. Similarly, the market

reaction to initial dividend declarations is found to be substantial and much greater than

previously found for favourable dividend classifications in general.

Kalay and Loewenstein (1985) argue that the observed abnormal return during the

announcement period is due to the dynamic pattern of beta, because the systematic risk

around the event window is more likely to increase. Their empirical results support their

hypothesis of increased systematic risk, but this spike in beta does not fully explain the

excess return across the event window. Following this thread of logic, Kalay and

Loewenstein (1986) posit that the timing of a dividend announcement is key to explaining

the announcement effect, because management, as investors expect, would postpone the

announcement of a dividend decrease (i.e., bad news). They report that the level of

dividend reduction in the deferred announcement is larger than other similar

announcements, which results in significant negative abnormal returns.

34

Ghosh and Woolridge (1988) find shareholders are more sensitive to a downward

change in dividends. Their multi variable regression suggests that the level of decrease in

market capitalization depends on the magnitude of the dividend cut, the size and risk

feature of the firm, and the stock return of the previous period. The magnitude of capital

loss can be exacerbated by simultaneous underperforming earnings results, but mitigated

by continuous poor operating performance or concurrent stock dividend. If the reason for

dividend reduction is related to capital expenditure, the negative impact on share price is

marginally attenuated. They find sufficient evidence that investors consider the dividend

change announcement, along with other simultaneous information, when they assess the

information content conveyed by management.

Venkatesh (1989) compared the influence of dividend initiation with that of earnings

announcements and found that the market reaction to a quarterly earnings announcement

becomes weaker when it follows the dividend initiation, which implies a partial substitutive

relationship between dividends and earnings announcements. Further, they report a

significant drop in market volatility, especially the firm level volatility, after the initiation of

a cash dividend. The author attributes this phenomenon to the possibility that shareholders

are temporarily dominated by the dividend imitation and overlook other signals following a

dividend announcement.

Michaely et al. (1995) study the market reactions to initiations and omissions of cash

dividend announcements. In line with previous literature, they report that the abnormal

return is more significant in the case of dividend omission announcements than in those of

dividend initiation announcements. With a sample of 561 dividend initiations and 887

dividend omissions, they document sustained drifts in share price which continue in the

35

same direction (but weaker in strength) in the following year. For instance, the average

cumulative abnormal return in the 3 day window around dividend initiation events is 3.4%,

which is followed by an average 15.6% excess return in the subsequent three year window.

Further, the price drift linked with dividend initiation and omission is far more outstanding

than that following earnings shocks. They apply a trading strategy based on the above

mentioned momentum of dividend shocks with a long short strategy and the simulation

shows plausible positive returns in 22 out of 25 years.

Lee (1995) also investigates the response of stock prices to dividend shocks. The

methodology used is a bivariate model of stock prices and price dividend spreads. Different

from previous research designs, Lee (1995) models the dividend as a combination of a

permanent component and a temporary component, which are linked to stock prices via a

stock price valuation model. The initial responses of stock price to both permanent and

temporary shocks are significantly strong. Further, the price dividend spreads can be

explained by the temporary shocks in dividends, which indicate that large swings in share

prices are caused by temporary shocks. The author argues that a dividend is an imperfect

information transmission mechanism, as investors are unable to distinguish permanent

shocks from temporary shocks.

Docking and Koch (2005) conjecture that the direction andmagnitude of impacts from

dividend change announcements on the share price are affected by the momentum effect.

Their empirical results suggest that, if the dividend shock is against the latest tendency, for

example, dividend reduction (increase) during upward (downward) movement, the market

reaction will be much greater. The authors attribute these outcomes to a mixture of the

36

dynamic rational expectation equilibriummodel with the behaviour factors of the investors’

sensitivity to market momentum and volatility.

As the popularity of share repurchase evolves, it is argued that managers rely more

on share repurchase than on traditional cash dividends to transmit private information.

Choi and Chen (1997) report that the abnormal return caused by stock repurchase

announcements is much larger than that of dividend increase announcements, after

controlling for idiosyncratic factors. It is also found that share repurchase announcements

are more likely to be followed by a significant drop in systematic risk and better ratings

issued by financial analysts. Some survey results also suggest managers prefer share

repurchase because of its flexibility (Brav et al., 2005). Changes in dividends have become

less informative than they used to be three decades ago, partly due to the fact that investors

are used to an inelastic dividend policy. Skinner (2008) finds that share repurchase has

become more important than cash dividends as a method to distribute cash, because the

dividend policy of large firms becomes more conservative and share repurchase exhibits

greater flexibility.

Different from Kalay and Loewenstein’s (1985) study which posits a short run

decrease in systematic risk around dividend announcement, Grullon et al. (2002)

hypothesize that the change in cash dividends signals changes in the fundamental features

of underlying firms, which influences their share prices. For firms which increase dividends,

it is likely that they are approaching a maturity stage and the level of systematic risk has

decreased significantly, which causes positive reactions in the stock price. It is also found

that those firms with increased dividends are inclined to freeze their investment

expenditure and experience inferior profitability in the following years. Empirical results

37

suggest these stocks with increased dividends but decreased systematic risks report best

share price performance in the next three years, which is explained as a change in the

systematic risk (i.e., over estimation on expected return).

Similar studies have been conducted on developed markets other than in the US.

Balachandran (2003) looked into the share price reaction to firm’s dividend reduction

announcements in the UK. Empirical results indicated negative stock returns are significant

following both interim and final dividend reduction announcements, but are much stronger

in the case of interim dividend reductions. The negative return around the reduction in the

final dividend is more likely to be reversed in the short term. The author also documents a

significant correlation between magnitudes of dividend reduction and the excess return in

pre event and post event observation windows. Del Brio and de Miguel (2010) apply the

signalling hypothesis proposed by John and Lang (1991) to the Spanish market. The authors

found that stock abnormal returns of Spanish listed firms are sensitive to dividend

announcement, but not fully explained by dividend signalling hypotheses. Instead, they are

more concerned about the insiders’ trading activity. The excess return is significantly

positive (negative) when the cash dividend increases (decreases) along with the increase

(decrease) of shareholdings by informed insiders.

Besides the practice of cash dividend payouts, stock dividends are also popular in both

developed and developing markets, although they are not a cash distribution. Charest

(1978b) reports that the excess return in the three months after stock splits (i.e., stock

dividend) is significantly different from zero. Grinblatt et al. (1984) report that there are

positive share price reactions to stock dividends and stock split announcements

independent of other firm specific announcements. Although the announcement date of

38

return of stock dividends and stock splits is not correlated with forecasts of future increases

in cash dividends, the announcement effect can be interpreted by several signalling

hypotheses. One of them is the ‘retained earnings hypothesis’, that is, the reduction in

retained earnings (equity) due to stock dividends19 announcement suggests management’s

confidence in future cash flows. In other words, the strong expectation of the firm’s

operation is the foundation for the stock dividend decision. The other signalling hypothesis

is the ‘attention hypothesis’ which posits that underpriced firms use stock dividends to

catch the attention of various market participants.

Bechmann and Raaballe (2007) investigate stock dividends and stock splits based on

a sample from the Copenhagen Stock Exchange. It is reported that the announcement effect

of stock dividend is closely linked to a firm's payout policy. They find that firms with a split

factor of less than two report an average announcement effect of 4.23%, and they are able

to increase the cash dividend with a magnitude similar to the increase in share capital

caused by stock dividends. Stock dividends with a split factor of two or more are associated

with a permanent increase in cash dividends, but the magnitude is less than the percentage

increase in share capital. These shares report amarginally positive abnormal return (0.08%),

on average, around announcement, which is in line with the retained earnings/signalling

hypothesis.

19 According to generally accepted US accounting principles, stock dividends of 25% or less will reduce themarket value of stock dividends from the firm’s retained earnings. The accounting treatment of stockdividends beyond 25% is the same as that of stock splits.

39

2.4.2. Dividend announcement effect: emerging markets

Research on the dividend announcement effect in emerging markets has become

popular in recent years. Chen et al. (2002) investigate the information content of the

concurrent earnings, stock dividend and cash dividend announcement in the Chinese

market. Based on a sample of 1,232 observations, their results suggest stock abnormal

return is more associated with earnings announcements than cash dividend

announcements. Stock dividend announcements appear to work as catalysts to earnings

announcements, in that, they enhance (dampen) stock return when earnings news is

positive (negative).

The potential limitation of Chen et al. (2002) is that its sample observation window

covers the very early period of the Chinese stockmarket (1994–1997). Recent research with

longer observation windows finds different results. Chen et al. (2009a) conducted an event

study on announcements between 2000 and 2004, and document that the change in cash

dividends is positively correlated with stock abnormal return in the Chinese market.

Different from the traditional signalling hypothesis, they find that a positive abnormal

return is reported after both increases and decreases in cash dividends. The implication is

that investors welcome cash dividends and are indifferent to the change in dividends. Their

cross sectional analysis suggests that both cash dividend yield and the ratio of non

negotiable shares influence the announcement effect of cash dividend changes. They also

acknowledge that the pattern of announcement effect changes significantly due to the

implementation of some administrative orders.

Cheng et al. (2009) adopt a different measurement of information content embedded

in dividends. Instead of changes in the cash dividend, they utilize unexpected dividends

40

along with unexpected earnings. They report that unexpected earnings contribute to the

abnormal return within short term event windows, but there is no significant contribution

from unexpected dividends. Instead, the change in stock dividend per share shows a

significant contribution to abnormal return around the announcement period, which

supplies evidence that a stock dividend decision, although not classified as a real

distribution, is a key indicator of stock abnormal return in the Chinese market.

Unlike mainland China’s, the Hong Kong market has been operating for decades and

the payout pattern of public firms listed on the Hong Kong market has been studied. Cheng

et al. (2007) test the information content of simultaneous earnings announcements and

dividends announcements in the Hong Kong market. The listed firms in the sample are

grouped based on firm specific features, such as concentrated control rights held by family,

insufficient corporate transparency and tax free cash dividends. Their results indicate stock

prices react vigorously to both unexpected earnings and dividend changes, but the

unexpected dividends contribute more to the announcement effect, which contradicts the

traditional viewpoint that earnings announcements carry more information.

Cheng and Leung (2008) examine whether managers of Hong Kong listed firms take

advantage of the private information they have by trading stocks before the simultaneous

earnings and dividend announcements. Their findings support the signalling hypothesis

because insiders would buy (sell) before good (bad) news. They also report a significant

correlation between pre event insider trading behaviour and the announcement effect.

Besides the Chinese and Hong Kong markets, some empirical research has been

conducted on other emerging markets. Hussin et al. (2010) investigate the announcement

effect of both cash dividends and corporate earnings on stock prices on theMalaysian Stock

41

Exchange, in order to test the semi strong efficient market hypothesis. Their sample

consists of 120 firms listed on the Main Board of Bursa Malaysia, which reported earnings

and dividend announcements in 2006. The abnormal returns around the announcement

date are positive (negative) when there is a dividend increase (decrease), which can be

mainly attributed to the information content embedded in the cash dividend.

Ali and Chowdhury (2010) look into the stock price reactions to cash dividend

announcements in the private commercial banks of Bangladesh, and test whether the

information content is significant. They conduct a standard event study on the dividend

announcement of 25 listed banks. The cumulative abnormal returns around the event

window indicate little investor response to the announcements. They conclude that

investors are more concerned with the trading activity of insiders and other influential

factors. In other words, a dividend announcement in the Bangladesh market conveys little

information about the private information held by managers.

2.5. Dividend Policy and Future Earnings Growth

2.5.1. Signalling hypothesis of dividend policy

Besides the function of rewarding the shareholder andmitigating the agency cost, the

cash dividend is utilized as a tool to convey themanagers’ viewpoint on the firm’s operation,

or more precisely, the inside information which is not available to outside equity holders

(i.e., asymmetric information). Different from the previous section, this section

concentrates on whether the assumed signal in the cash dividend is realized.

Beaver (1968) discusses whether investors abstract information content from firms’

annual earnings announcements and how it influences the investors’ expectation on future

42

earnings and the market value of the firm. Watts (1973) tests the hypothesis that cash

dividends contain information about the firm’s future earnings. The regression results

indicate that the change in future earnings is linked to current unexpected dividend

changes, but the change in future earnings is not economically significant.

John and Williams (1985) develop a theoretical model which incorporates the effect

of asymmetric information and explains the phenomenon that many firms are engaged in

simultaneous cash dividend announcements and seasoned equity offerings. A taxable

dividend acts as an information source about the firm’s future earnings, which is not

unveiled through periodic announcements. With the existence of asymmetric information,

the transaction cost attached to the equity issuing process and the tax related to cash

dividend payments are the elements which distinguish good firms from bad. The

management has the incentive to signal the good news because managerial shareholdings

would benefit from investors’ reaction to positive dividend shocks. This signalling

equilibrium still holds when multiple signals, such as share repurchase and investment

plans, are presented to the market (Williams, 1988).

By contrast, Miller and Rock (1985) propose a model focusing on consistent signalling

equilibrium in conditions of asymmetric information and share trading, in which cash

dividend announcements work as effective signals. In Miller and Rock’s (1985) framework,

the positive cash dividend shock is not always treated as an encouraging signal if the source

of the dividend was once allocated for capital expenditure (i.e., an underinvestment

problem). The market will punish the underlying firm if shareholders realize that managers

attempted to mislead them by paying unnecessarily excessive dividends. Miller and Rock

43

illustrate that lower managerial shareholdings are accompanied by higher cash distribution

and lower investment level.

Ofer and Thakor (1987) argue that the cost attached to the subsequent equity offering

following a cash disbursement in dividends or share repurchase will become a substantial

loss and impair shareholders’ wealth. Ofer and Thakor (1987) contribute a model which is

able to accommodate both cash dividends and share repurchase. As per their findings, cash

dividends and share repurchase do not prevail over each other when taken as vehicles to

deliver information to outside equity investors. Share repurchasemay lead to a higher stock

price response, but this benefit will be offset by the cost of equity offerings, while a cash

dividend is disbursed from internal funds with no need to seek additional equity capital.

Ofer and Siegel (1987) use analysts’ earnings forecasts as the proxy of market expectation

and report that analysts are likely to revise up their forecast when unexpected dividend

changes hit the market.

Yoon and Starks (1995) claim that dividend announcements may not reveal

information about managers' investment policies, as the change in future investment cash

flow is positively correlated with the change in current cash dividends, while the investment

opportunity, represented by Tobin’s Q, does not affect the stock abnormal return. Equity

analysts are more likely to update their forecasts of current earnings after a dividend

announcement. These results are more inclined towards the signalling hypothesis. But

DeAngelo et al. (1996) argue that the quality of a cash dividend as a signal of the firm’s

future cash flow is dubious because themanagementmight be over optimistic about future

performance, and the popular dividends smoothing practice undermines the reliability of

the information content contained in cash dividends. Based on a sample of 145 NYSE listed

44

firms with their first earnings decrease after a long history of earnings growth, DeAngelo et

al. (1996) find changes in cash dividends do not indicate changes in future earnings and

capital expenditure.

Similarly, Noe and Rebello (1996) focus on managerial optimism and how it affects

the firm’s dividend policy, subsequent equity financing and investment policy. They

document that dividend policy and subsequent equity financing become an ideal signalling

mechanism when adverse selection and managerial optimism co exist. In their opinion, the

signalling function of dividend change needs some supplementary conditions when the

management expects investors to capture the information content. At the same time,

Dhillon and Johnson (1994) argue that dividend change more likely reflects the hypothesis

of wealth redistribution between shareholders and bondholders; there is little evidence to

rule out the information content hypothesis.

Bernhardt et al. (2005) investigate whether the information content of dividends is a

Spencian type of signal by testing the monotonic hypothesis that the gap between tax rates

on cash dividend and capital gains is positively related to the value of information contained

in the dividend announcement. With robust nonparametric methodologies, they report

that the abovementioned positive correlation does exist, but it is not consistent as the level

of the dividend signal changes, which suggests that existing signalling models are not

universally applicable.

Bechmann and Raaballe (2010) develop a class of signalling models and look into the

efficiency of the information signal conveyed by the cash distribution announcement.

According to their models, a taxable cash dividend is a necessary component to establish a

robust signal. When cash dividends of good firms are unable to distinguish them from bad

45

firms, good firms need to announce a share repurchase to enhance the signal, but at the

cost of cutting investment or raising new equity capital. The key assumption in their models

is that good firms derive more marginal benefit from investment than bad firms.

Consequently, the opportunity cost of cutting back investment would be much higher for

good firms. Therefore, the cost benefit of cash distribution as a signal should be

reconsidered by the management.

A great deal of empirical research about the signalling function of cash dividends

concentrates on the special cases of dividend initiation and omission, because they are

deemed to contain more information about the expected future cash flow. Asquith and

Mullins (1983) utilize a sample of 168 firms with independent announcements of dividend

initiation, or dividend resumption after halting for 10 years, and document significant

positive abnormal returns around the announcement period. Further, the magnitude of

abnormal returns is positively correlated with the size of the cash dividend announced. In

addition, any subsequent increase in cash dividend following the initiation renders higher

abnormal returns. This finding is in line with the signalling hypothesis because the increase

in cash dividend affirms the information from the dividend initiation announcement.

Lang and Litzenberger (1989) discuss whether the cash flow signalling hypothesis or

the agency cost of free cash flow exerts more pressure on the share price. They find that

firms with less than unity Tobin’s Q report higher average stock returns around

announcement period, when the underlying firm declares substantial dividend change. Less

than unity Tobin’s Q indicates that the market believes the underlying firm has a significant

over investment problem and its market value is impaired by the projects with negative net

present values (NPVs). When there are unexpected changes in the dividend payout,

46

investors treat it as a change in the free cash flow and the over investment problem will be

mitigated or exacerbated after the cash dividend announcement.

John and Lang (1991) study the linkage between insider trading and dividend

announcements in order to test the potential signalling behaviour. Their signalling model

with endogenous insider trading suggests that the market reacts significantly to the trading

action of the insiders before the dividend announcement. When the insiders liquidate their

shareholdings before dividend initiation announcements, external investors take it as a

negative signal that the firm’s growth momentum is fading. Consequently, the abnormal

returns of these stocks are negative and far below other firms with no insider trading or

insiders’ increasing shareholdings. Furthermore, a dividend increase may cause divergent

market feedback depending on the firm’s investment opportunities.

Denis et al. (1994) extend the idea of information about future cash flow in Lang and

Litzenberger (1989), and test whether the change in dividend will be interpreted by the

market participants as signals of cash flow, symptoms of over investment or the clientele

effect. With a sample of 5,992 dividend increases and 785 dividend decreases between

1962 and 1988, they find that the announcement effect is positively correlated with the

standardized changes in dividend, while unrelated to the Tobin’s Q. It is also reported that

analysts are more likely to revise the earnings forecast after dividend change

announcements. Different from what is hypothesized in the agency cost of free cash flow,

firms’ dividend change announcement is positively correlated with the following change in

capital expenditure. In other words, firms are likely to invest more (less) after they

announce dividend increases (decreases). Therefore, their empirical results supply sound

47

support to the cash flow signalling hypothesis of dividends, but offer limited evidence in

favour of the over investment hypothesis.

Although a stock dividend is not a real cash distribution, it is also considered as a tool

to mitigate the asymmetric information between management and external shareholders.

It seems stock dividends and stock splits do not occupy many financial resources because

there is no cash outflow. However, according to the surveys conducted by Elgers and

Murray (1985), managers acknowledge that the costs of stock dividends and stock splits are

not negligible. It raises a similar argument to the cash dividend that costly stock dividends

(splits) are another vehicle to enable insiders to provide more information about the firm.

As illustrated by Elgers and Murray (1985), a small quantity of stock dividends is more likely

to be the manager’s demand to signal optimistic expectations on future performance. They

also report that the cash balance is not the sole initiator of the stock dividend decision,

although some managers do take it into account. The survey result indicates that managers

design the stock dividend announcement with the full insight of their own firms. Similar

results are also reported in Lakonishok and Lev’s (1987) research which conducts empirical

analysis on the motivation of stock splits or stock dividend decisions, and the positive

market reaction to that news. Besides the normal pursuit to restore share prices to a

‘normal range’, Lakonishok and Lev (1987) find some supportive evidence of the signalling

motive of stock splits, but attribute the intention of stock dividends to an attempt to

substitute it for a relatively small cash dividend.

McNichols and Dravid (1990) supply further evidence that management signals

private information about future cash flow by stock splits and stock dividends. After

controlling for firm specific parameters, both earnings forecast errors and stock abnormal

48

returns are highly correlated with split factors. Consequently, investors update their

evaluation of the firms’ value based on the private information conveyed via split factors.

Their analysis also indicates that the abnormal return around stock dividend

announcements is still significantly correlatedwith split factors, even after earnings forecast

errors are included in the control variables. It implies, too, that the information content in

the stock dividend announcement goes beyond the future earnings and is not fully

explained by the signalling hypothesis.

2.5.2. Relationship between dividend policy and future earnings growth

As per conventional wisdom, cash dividend payouts will reduce the cash balance and

increase the probability of raising external capital in the future, which results in more

financial cost. Gordon (1962) proposes the share price valuation model which only takes

into account the present value of future cash flow. Future cash flow is determined by

current investment, which is influenced by the retention ratio. In his model, the trade off

between current payout ratio and future growth rate dominates the valuation of a common

share. As the retention increases, more internally raised funds are available to be used for

capital expenditure and enhance the future growth rate. Therefore, Gordon (1962) is the

founder of the conventional school of thought about the negative relation between

dividend payout and future earnings growth.20

Myers (1984) raises the question of how managers determine their capital structure,

and whether there is a target debt to value ratio or do they just follow the pecking order in

20 Similar viewpoint is raised in Rozeff (1982) although his explanation is based on the trade off betweenagency cost of free cash flow and funding cost of subsequent equity financing. Ibbotson and Chen (2003) alsopoint out that a lower dividend payout ratio implies a higher future growth rate.

49

which managers prefer internal finance (i.e., retained earnings) to external finance.

According to the pecking order theory, firms will benefit from the cheaper funding cost if

they reduce the cash distribution to shareholders, which underpins the traditional

viewpoint that more retention (less payout) will enhance the future earnings growth.

Fama and French (2002) compare the trade off theory and pecking order theory by

studying firms’ profitability, cash dividends, investments and capital structure. Based on a

sample of more than 3,000 firms between 1965 and 1999, they find that higher dividend

payouts occur in the firms with stronger profitability and less capital expenditure. Unlike

the MM proposition II, firms which report higher earnings and more investment are usually

less leveraged. The dividend payout ratios of firms with more capital expenditure are low,

on average, in the long run. The short term variation in investment is funded by debt and

will not influence the dividend payout (i.e., the sticky dividend policy holds). In other words,

the pecking order theory works quite well in explaining the firm’s financial decisions

regarding investment, dividends and capital structure. Their results support the

conventional thought that the retention of net profits is closely linked to better operating

performance.

Harada and Nguyen (2005) extend the research on dividend signalling to the Japanese

market and examine the relationship between dividend adjustments and subsequent

operating performance. Differing from previous research, they develop a predictive model

of conditional dividend adjustment. They show that the change in dividend is positively

correlated with future earnings. Meanwhile, the risk adjusted returns in the long

observation window are in line with the forecast change in earnings.

50

As corporate finance research evolves, the relation between cash dividend and future

earnings growth is re examined, and a contemporary school of thought is developed which

posits that cash dividend payouts and future earnings growth are positively correlated

because of the dividend signalling. Healy and Palepu (1988) suggest that positive (negative)

earnings growth is associated with the decision of initiating (omitting) dividends.

Meanwhile, changes in the subsequent earnings growth are positively correlated with the

announcement period abnormal return, which indicates investors’ capture of the signal

embedded in the cash dividend announcement. It is also underpinned by the evidence that

the share price reaction to subsequent earnings announcements is less significant, as

investors are aware of the upcoming change implied in the previous dividend

announcement.

Benartzi et al. (1997) argue that the change in cash dividend does not indicate the

change in future earnings. They challenge the traditional implication that the change in cash

dividends signals the future performance. Firms that increase dividends reported strong

performance in the current and previous year, but do not exhibit superior earnings growth

in the subsequent year. Instead, firms which cut dividends are inclined to show robust

earnings growth in the following year. Consistent with Lintner (1956), they report that

dividend increasing firms are more likely to avoid a drop in future earnings, although the

size of the dividend increases is not an effective parameter to forecast future earnings. Their

empirical results also show that dividend increasing firms report modestly positive

abnormal returns in the following three years. In reply to Benartzi et al. (1997), Nissim and

Ziv (2001) investigate the relation between changes in cash dividend and future

profitability, measured in terms of either future earnings or future abnormal earnings. They

51

argue that the results in Benartzi et al. (1997) are influenced by both the specification issue

(the measurement error of the dependent variable) and a missing control variable (return

on equity). Based on a sample of about 700 firms between 1963 and 1998, they report that

dividend changes supply information about earnings in the following financial years after

controlling for market and accounting data. The correlations between dividend changes and

the earnings change in the following two years are significantly positive. They attribute the

outcomes to the signalling hypothesis that management conveys information about the

firm’s operation by changes in the dividend.21

Meanwhile, following Jensen and Meckling (1976) and Jensen (1986), some

researchers argue that although the reduction in the cash balance as a result of paying

dividends may increase the funding cost, it also mitigates the agency cost and prevents

management from self dealing activities at the cost of shareholders’ wealth (i.e., agency

cost).22 Because cash dividend works as a mechanism to monitor the management, the

future earnings growth will benefit from the cash dividend payment. Therefore, a positive

correlation may well exist between current cash dividends and future earnings growth.

Arnott and Asness (2003) investigate whether the cash dividend payout is a good

predictor of future earnings growth. Instead of firm level data, they use the S&P500 Index

because the earnings growth and average dividend payout ratio of a portfolio are less

affected than the temporary anomalies of individual firms. Their descriptive and regression

evidence strongly indicates that the change in expected future earnings growth is in the

same direction as the change in current payout ratio, which is not attributable to factors

21 Other relevant research on the dividend signalling hypothesis, such as Ofer and Siegel (1987), Denis et al.(1994) and Cheng et al. (2007) is reviewed in the previous subsection.22 For a comprehensive literature review on agency cost of free cash flow, please refer to Section 2.2.2.

52

such as the mean reversion in earnings and the change in dividend yield. Their explanation

for these results is the monitoring function of cash dividend on management’s empire

building activities.

Gwilym et al. (2006) extend the research to 11 regimes in the Organization for

Economic Co operation and Development (OECD), including the US. Following Arnott and

Asness (2003), they also use the index portfolio payout level instead of firm payout.

Although the country specific features are divergent, their regression results are generally

consistent with what has been documented in Arnott and Asness (2003), that is, a positive

correlation between current dividend payout and future earnings growth, however, a

higher payout ratio does not imply stronger growth in real dividends.

Zhou and Ruland (2006) extend the research of Arnott and Asness (2003) to US firm

level data because the coverage of the S&P500 Index may not reflect the whole picture of

US listed firms, and aggregate results may not be applicable to individual firms. With a

sample from 1950 to 2003 and using the Fama Macbeth procedure (Fama and MacBeth,

1973), they show that firms with a high dividend payout ratio exhibit much better future

earnings growth in all short, medium and long observation windows. They also conduct a

comprehensive robustness check, such as alternative measurement of dividend payouts,

and mean reversion in earnings and industry effects, which suggests consistent outcomes.

Similar to Arnott and Asness (2003), Zhou and Ruland adopt the agency cost of free cash

flow as an explanation for their results.

Some recent literature also shows a similar pattern in developing markets. For

example, Vermeulen and Smit (2011) look into the payout ratio and subsequent earnings

growth in the South African market. With a large sample of more than 12,000 firm years

53

across the period from 1973 to 2009, they find a significant correlation between higher

dividend payouts and stronger subsequent earnings growth.

Huang et al. (2009) apply the research idea of Zhou and Ruland (2006) to the Taiwan

market and examine the relationship between the dual dividend23 practice and subsequent

earnings growth. They focus on whether the proportion of cash dividend and stock dividend

has any association with the subsequent earnings growth, because cash dividends and stock

dividend are assumed to have different functions. Specifically, cash dividend works as a

mechanism to deal with agency cost (i.e., the free cash flow problem), while a stock

dividend is to signal management’s confidence in the future performance (i.e., signalling

hypothesis). Their empirical evidence indicates that the significant positive correlation

between total payout ratio and subsequent earnings growth only exists in the sub sample

with a balanced dual dividend, which supports a new ‘balanced dividend hypothesis’.

2.6. Summary

This chapter provides a review of both theoretical and empirical research that deals

with how dividend policy interacts with corporate ownership, future earnings growth and

stock abnormal return around the event window. Three major issues are covered in this

chapter regarding the function of dividend policy.

First, due to the agency problem caused by separation of principal and entrepreneur

(Jensen and Meckling, 1976), the cash dividend policy is deemed to work as a mechanism

to mitigate agency cost. Apart from the traditional definition of agency cost, the

concentration of ownership has created another type of agency problem between

23 Dual dividends mean a combination of cash dividends and stock dividends.

54

controlling shareholders and minority shareholders. Controlling shareholders have the

motivation and capability to either expropriate the wealth of minority shareholders by

tunnelling the resources of underlying listed firms, or to provide support when the listed

firm needs financial assistance. The dividend payout policy, including the decision to pay a

dividend and the magnitude of the dividends, is influenced by the level of control rights and

the background of the controllers. The existing literature is inclined to support the

viewpoint that concentrated ownership will provide a remedy to agency costs, however,

whether or not this viewpoint is applicable in emerging markets needs more empirical

research, because the agency cost, especially the free cash flow problem, ismore prominent

in emerging markets.

Second, it is documented that dividend policy conveys private information held by

insiders. The magnitude and significance of the announcement effect reflects whether

investors accept the signals. In general, testing the announcement effect is equivalent to

testing the efficient market hypothesis. Existing models use either changes in the cash

dividend or unexpected cash dividends to measure the information content of dividend

announcements. Meanwhile, dividend announcements are usually accompanied by a

concurrent firm level announcement, such as earnings. Therefore, the key tomeasuring the

information content accurately in the dividend announcement is to control the noises from

other simultaneous announcements. Besides cash dividends, a stock dividend (i.e., bonus

shares) is also regarded as a channel to release information about managers’ opinions on

the firm’s operating performance.

Finally, the relationship between dividend policy and future earnings growth is also

relevant to the free cash flow problem. Managers, as well as controlling shareholders, are

55

more likely to abuse the financial resources of a firm under their control. Paying cash

dividends would decrease the free cash flows and lessen the probability of conducting

negative NPV projects. According to traditional thought, the subsequent earnings growth is

related to cost of capital, which can be decreased by more retention and lower/fewer cash

payouts. Therefore, the traditional school of thought proposes a negative correlation

between payout ratio and subsequent earnings growth. Recent empirical studies have

supplied much evidence to support this contemporary theory (Arnott and Asness, 2003,

Zhou and Ruland, 2006).

56

Chapter 3 Review of Chinese Stock Market and Payout Policy

3.1. The Evolution of the Chinese Stock Market

The structure of the Chinese economy before 1978 is a typical example of a highly

centrally planned economy, which has strong execution but lacks flexibility in the allocation

of various resources as the mechanism of price is disabled (Von Mises, 1951). In 1978, the

leading party, the Chinese Communist Party (CPC), decided to reform the national economy

into a more market oriented mechanism, in order to improve the people’s living standards

and adapt to the irrevocable trend of globalisation. With the approval of the central bank,

the People’s Bank of China (PBOC)24, Beijing Tianqiao Department Store Co Ltd and Shanghai

Feile Audio Visual Co Ltd were allowed to sell common shares in July 1984. The first over

the counter market for trading the outstanding common shares was established in

September 1986, and the first securities company, Shanghai Wanguo Securities Company,

was founded in 1988.25

All the above events were only initial experiments to explore the feasibility of

establishing a capital market in China. The slow progress in the development of the Chinese

capital market was mainly due to the political debate about public ownership in a socialist

country which excluded private ownership. The drawbacks of public ownership were

reflected in the poor efficiency and productivity of the state owned enterprises because the

performance of the organisation was not related to the remuneration of the employees.

The fall of the Communist bloc in Eastern Europe had alerted the ruling Chinese communist

24 PBOC used to be the only financial institution in mainland China until 1984, when its commercial bankingbusiness was separated to establish the Industrial and Commercial Bank of China (ICBC).25 The securities company is similar to an investment bank in developed markets.

57

party to the possibility that the lag in economic developmentmight provoke further political

unrest. Thereafter, the CPC decided to set aside the ownership debate and accelerate the

economic reforms. Their resolution was underpinned by the establishment of the Shanghai

Stock Exchange in November 199026 and China Securities Regulatory Commission (CSRC) in

October 1992.27

One important mission of the Chinese equity market is to convert state owned

companies to modern enterprises through partial share issuance privatization (SIP). Since

the establishment of the two stock exchanges, the number of listed companies has

increased from 6 in 1990 to 2,301 in 2011,28 as shown in Figure 3.1. The slow pace of the

growth in the number of listed companies is due to the specific features of the initial public

offering (IPO) process. A quota system was implemented by which the central government

and provincial government selected the candidates to be floated on the stock market.29

Since 1998, an approval system has replaced the quota system which is run by the Stock

Issuance Examination Commission, a subdivision of CSRC. Compared with the initial sluggish

growth in the number of listed companies, the annual trading turnover and total market

capitalization has reported significant jumps since the introduction of the approval system.

The annual trading turnover peaked in 2010, while the market capitalization as of year end

reached 33 trillion CNY in 2007, as shown in Figures 3.2 and 3.3.

26 Shenzhen Stock Exchange was approved by PBOC in April 1991 and incorporated in July 1991.27 But CSRC did not fully take over the function of monitoring and regulating the security market from theState Council of the People’s Republic China and PBOC until the Securities Law was enforced in 1997.28 The figures for the Shenzhen Stock Exchange include main board, growth enterprise market (GEM) board,and small and medium enterprise (SME) board.29 Due to the accumulated quota from previous periods and the time lag between IPO application and listingdate, the quota system was terminated in 2002.

58

Figure 3 1 Number of listed firms (1990 2011)

Source: China Securities Regulatory Commission

Figure 3 2 Annual trading turnover (1990 2011), in billion CNY

Source: China Securities Regulatory Commission

Figure 3 3 Total market capitalization (1990 2011), in billion CNY

Source: China Securities Regulatory Commission

0 5 24 76 118 128 227 349 400 452 502 500 494 493 526 531 579 677 748 8001004

1336

7 7 29106 170 185

288373 426 472

560 637 706 771 827 821 832850 855 857

883

924

0

500

1000

1500

2000

2500

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Shenzhen Shanghai

0 1 23 226 566 305 903 1,356 1,236 1,689 3,114 1,993 1,647 2,056 2,625 1,9085,731

30,229

17,960

34,53630,280

23,614

0 3 42 126 238 92 1,203 1,674 1,111 1,4222,926

1,338 1,070 1,072 1,534 1,204

3,197

15,280

8,603

18,82323,983

18,282

0

10,000

20,000

30,000

40,000

50,000

60,000

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Shanghai Shenzhen

2 7 63 209 249 245 535 913 1,063 1,452 2,679 2,715 2,502 2,951 2,580 2,290

7,139

26,939

9,761

18,450 17,93214,852

0 8 50 128 104 89 419 819 874 1,1822,105 1,544 1,271 1,220 1,077 897

1,756

5,626

2,373

5,8678,504

6,620

0

5,000

10,000

15,000

20,000

25,000

30,000

35,000

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

Shanghai Shenzhen

59

The above facts suggest the Chinese equity market was designed and established by

the government, which is different from most developed markets that were based on the

spontaneous order of market participants and independent regulators. This fundamental

difference determines that the capital raising function of the Chinese stock market

overrides its investment function in order to support the economic reforms, which is

criticized by market participants as the listed firms are more advantaged than the minority

investors. This situation has changed since the Securities Law was promulgated in 1997,

which sets up the securities regulatory framework and justifies CSRC’s position as the

regulator of the securities industry.30 Furthermore, CSRC encourages the interaction

between the Chinese stock market and global equity markets. On the one hand, more and

more Chinese companies are listing in overseas markets, companies such as People’s Life

Insurance Company of China, Chinese National Petroleum Corporation (CNPC), Sino

Petroleum Corporation (Sinopec) and China National Offshore Oil Corporation (CNOOC)

while, on the other hand, the Chinese equity market has been conditionally open to foreign

investors since the introduction of the Qualified Foreign Institutional Investor (QFII)

regulations.31 As part of the agreement to join the World Trade Organization (WTO), the

opening of China’s financial markets is inevitable. Since the promulgation of the Temporary

Regulation on Domestic Securities Investment by Qualified Foreign Institutional Investors in

200232, UBS AG became the first QFII to trade in the Chinese A share market which was

30 Security Investment Funds Law, which regulates the mutual funds industry, was implemented in 2003.31 Originally, the Chinese stock market was designed to separate local investors from overseas investors byallowing them to hold shares denominated in foreign currencies, US Dollars for B shares issued in ShanghaiStock Exchange, and Hong Kong Dollars for B shares issued in Shenzhen Stock Exchange. Foreign investorswere only allowed to trade B shares which are denominated in foreign currencies. B share market was opento domestic investors in 2001.32 It was ceased in 2006 and replaced by ’Regulation on Domestic Securities Investment by Qualified ForeignInstitutional Investors’.

60

previously only available to domestic investors. At the end of 2011, 142 QFIIs has been

approved by CSRC with a total investment quota of more than US$20 billion.

Figure 3 4 Negotiable/Total market capitalization ratio (1990 2011), in %

Source: China Securities Regulatory Commission

Another crucial milestone was the split share structure reform (SSSR) in 2005. As

mentioned, the partial SIP in the early stages of the Chinese equity market created a

significant portion of non negotiable state owned shares and legal person shares33 that

became a barrier to further development in the stock market, as the controlling

shareholders exploited their advantage with the majority voting rights. Effectively, SSSR

aims to float all non negotiable shares34 and the solution is that the non negotiable

shareholders provide considerations to the negotiable shareholders in exchange for their

approval of floating non negotiable shares.35 To date, more than 99% of listed firms have

33 Non negotiable state owned shares can be transferred and/or sold to state owned and/or privatecorporations, which become legal person shares.34 The high IPO price in the partial SIP caused problems because it was the consideration for state ownedshareholders’ promise to lock in their shareholdings. If these shareholdings are unlocked, minorityshareholders will demand compensation for breaching the promise. CSRC once attempted the reform in 1999and 2001, but stopped in 2002 because of the extremely negative response from the market.35 A three year or five year lock in period will be applied to these newly floated shares in order to limit theirimpact on the secondary market in the short term.

30.0

37.3

19.8 20.422.8 23.9

26.528.2 28.7

30.332.6 31.4 31.3

29.7 30.3 31.5

26.8 27.9

36.7

61.6

72.3

76.2

0

10

20

30

40

50

60

70

80

1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011

61

completed SSSR. As per Figure 3.4, themarket value of negotiable shares has increased from

around 20% in 1992 to more than 75% of the total market capitalization.36

Last but not least was the achievement of the floatation of commercial banks,

especially the big four state owned banks.37 Before the economic reform, Chinese state

owned banks were assigned the task to supply financial resources to themanufacturing and

service sectors, but the poor efficiency of these two sectors resulted in a huge number of

non performing loans. Central government established four asset management companies

to take over 1.4 trillion CNY of bad debts in 1999; the total of the non performing loans was

officially estimated to be 2.4 trillion CNY (Lu et al., 2005). After the peeling off of the bad

debts and several years of preparations including attracting strategic foreign investors, all

big four state owned banks became listed firms by the end of 2010.38 The successful

floatation of state owned banks labels the conversion of financial resources from debt

instruments to equity instruments, and underpins the pivotal position of the equity market

in China’s financial infrastructure.

As the Chinese economic reform deepens and widens, the stock market has become

an important part of the national economy, while both market participants and regulators

face more challenges.

36 Most state owned and legal person shares were restricted by a three year lock in period when they wereconverted into negotiable shares.37 They are the Industrial Bank of China (ICBC), Agriculture Bank of China (ABC), Bank of China (BOC) and ChinaConstruction Bank (CCB).38 The Bank of China became the first state owned bank floated on the A Share main board in July 2006, afterit had been listed in the Hong Kong Stock Exchange onemonth before. Another three state owned banks havealso followed this pattern and are listed in both A share market and Hong Kong stock exchange.

62

3.2. Legal Environment of the Chinese Stock Market

The Chinese legal infrastructure is classified as a civil law system and the laws that

regulate the Chinese stock market include the Securities Law, Corporation Law, and Income

Tax Code39. The general trend in the Chinese legal infrastructure is to become more

transparent and less dominated by politics, although the global investment community is

somewhat concerned about whether the unique Chinese political system may undermine

the credibility of its legal system. A fairer legal system will encourage sophisticated foreign

investors to get more involved in the Chinese market and to monitor the corporate

governance issues of listed firms. Hereby, several important laws are selected and their

influence on the Chinese stock market will be briefly discussed.

3.2.1. Corporation Law

The Corporation Law regulates the incorporation of all businesses in China. The latest

edition of the Corporation Law came into force on 1 January 2006, in which Chapter 4 and

Chapter 5 cover the specific requirements for the infrastructure of listed firms. Different

from the unlisted firms, listed companies have to appoint independent director(s) to

monitor the decisions made by the board of directors. In the meantime, when the board of

directors deliberates on a business proposal related to entities represented by any current

director, the voting rights of those directors will be restrained. Further, any significant asset

purchasing decision or issuing guarantees on a third party’s debt must be approved with a

two thirds supermajority by general meetings of shareholders. Last but not least, several

constraints are set on the liquidation of shareholdings by start up founders, directors,

39 Income Tax Code is subdivided into Personal Income Tax Code and Corporation Income Tax Code.

63

supervisors and senior managers, with the purpose of restricting the potential self dealing

behaviours.

The Corporation Law does not include any content about the dividend policy, but it

specifies that listed firms are not allowed to repurchase outstanding common shares unless

the buyback is due to a reduction of registered capital, merger and acquisition, or an

employee stock ownership plan. Therefore, different from this widely accepted practice in

developed markets, share repurchase is not defined as a profit distribution in the Chinese

equity market, and cash dividend becomes the only means for Chinese investors to receive

financial benefits from listed companies.40

3.2.2. Securities Law

The Securities Law concentrates on the whole capital market, such as the issuance

and trading of various securities, financial institutions, stock exchanges, clearing houses,

and the regulator.

The issuance and trading of securities used to be covered by the Corporation Law until

1997 when the Securities Law was promulgated. Only firms with an outstanding

performance record for the consecutive three years are allowed to apply for an IPO, and

the CSRC has the final decision on the approval. Start up foundersmust stick to the specified

use of the IPO funds in the prospectus, and any change has to be approved by a general

meeting of the shareholders. Otherwise, the underlying firm will lose the opportunity to

raise funds through seasoned equity offerings (SEOs) later on. The amendment to the

40 Chinese individual investors are much worse off because there is no capital gains tax (CGT) at the moment,while cash and stock dividends from retained earnings are subject to income tax.

64

Securities Law in 2006 introduced the concept of an issuance sponsor in the underwriting

process in order to improve the quality of IPOs and enhance investors’ confidence.

Firms need to satisfy certain requirements in the level of paid up capital if they are

going to float on the stock exchange. A minimum 25 per cent of common shares are floated

when the total paid up capital reaches 30million CNY.41 The underlying firm has tomaintain

a robust operating performance after the floatation. If a listed firm reports a negative net

profit for three consecutive years, it will be delisted from the main board.42 Its purpose is

to stimulate listed firms to improve their performance and reward the investors, but the

unintended consequence is that many listed firms conduct accounting manipulations in

order to window dress their financial statements and maintain their listed status.

In addition, the Securities Law mandates the continuous information disclosure of

listed firms. Besides the periodic financial statements, listed firms have to disclose any

change in the name and shareholdings of the top 10 shareholders, the direct/ultimate

controller, and the shareholdings of directors, supervisors and senior managers. Any

material information, such as changes in a firm’s direct/ultimate controller, has to be made

public through the stock exchange when the firm becomes aware of its existence. These

requirements on disclosures attempt to rectify the asymmetric information between the

direct/ultimate controller and minority shareholders. Further, the Securities Law outlines

the insider and trading activities which are forbidden in the secondary market, in order to

curb insider trading and price manipulation which undermines the credibility of the stock

41 If the paid up capital exceeds 400 million CNY, the minimum ratio will decrease to 10%. This is the legalfoundation of the ‘share split structure’ and the high IPO price in the Chinese market.42 This situation is different from developed markets where the stock exchange has no justification to delist afirm because of its inferior financial results.

65

market. Insiders are defined as personnel who have access to material non public

information, such as the direct/ultimate controller, managing director and monitor of the

company, the IPO sponsor and the relevant persons in the regulatory body and securities

service agencies.

The Securities Law also details the potential illegal methods of manipulating the share

price by utilising the advantage of funds, shareholdings and information. Financial service

agencies, such as securities companies, are required to deal with the client in an honest and

transparent manner. Any dishonest behaviour, such as fabricating misleading information

or conducting transactions without appropriate authorization from clients, is regarded as a

serious breach and will be punished harshly.

3.2.3. Income Tax Code

The taxation issue is critical to the research on cash dividend policy, as there are

differences among the marginal income tax rates of various investors which leads to their

different reactions to the dividend decisions of certain firms (Elton and Gruber, 1970,

Litzenberger and Ramaswamy, 1979, Bajaj and Vijh, 1990, Allen et al., 2000, Lee et al.,

2006). The Chinese Income Tax Code was not established until the early 1980s because of

the centrally planned economic infrastructure. Since the economic reforms, the monetary

system has become independent from the fiscal system and income tax has been

introduced to motivate both enterprises and individuals to seek more profit and improve

the efficiency of how they utilize resources.

66

The Corporate Income Tax Code is separate from the Personal Income Tax Code

because personal income tax was introduced earlier than corporate income tax.43 Different

from the developed markets, the incomes of individuals in China are taxed at the source of

various incomes via a separate Pay As You Go (PAYG) system. Although individuals’ salary

incomes are subject to the marginal tax rate44, cash dividends received by individual

investors are taxed at a 20 per cent flat rate.45 Along with the lack of capital gains tax (CGT),

the taxation environment encourages Chinese investors within lower income tax brackets

to chase capital gains but neglect cash dividends.46

Stock dividends, paid from either retained earnings or capital reserves, are a popular

practice in the Chinese market and are preferred by individual investors (Chen et al., 2002,

Eun and Huang, 2007, Lin et al., 2010). Generally, stock dividends are not taxed as they are

unrealized gains. But, in China, the stock dividends from retained earnings are subject to

the same income tax rate as cash dividends, and the taxable income is based on the face

value.47 The listed company has to pay a small cash dividend, which is also taxable, in order

to offset investors’ tax liability stemming from a pure stock dividend derived from retained

earnings. This special feature distinguishes them from each other, as stock dividends from

capital reserves are classified as a transfer between two equity accounts, which are exempt

from income tax, rather than a profit distribution. On the other hand, the taxation of cash

43 Personal Income Tax Code was approved by National People’s Congress in 1980, while the income tax codeon state owned enterprises was approved by State Council in 1984.44 There are 9 levels of marginal income tax rate brackets from 5 per cent to 45 per cent, with a steady 5 percent increase from the previous level.45 As a stimulus to prop up the stock market, the concessions on cash dividend paid to individuals is frequentlyapplied and removed by Treasury Department.46 Intuitively, investors within higher income tax rate brackets would prefer cash dividends. But this effect ismarginal because the size of cash dividends is much smaller than that of capital gains.47 The accounting treatment of stock dividends is also based on the face value, which is different from theUnited States where a small stock dividend (less than 25%) is booked as its market value.

67

dividends paid to institutional investors is slightly different. Before 2008, institutional

investors needed to pay the difference between their own income tax rate and that of the

public firm they invested.48 Since 2008, institutional investors are exempt from income tax

on the cash dividends if they hold the investment for more than 12 months. In other words,

they will receive tax free cash dividends.

Different from domestic investors, foreign investors, both institutions and individuals,

are exempt from income tax in China, partly because of the bilateral taxation agreements

with some countries where dividend incomes are subject to their domestic income tax

systems. Another motivation for this regulation is to encourage foreign investors, especially

institutional investors, to make long term investments in the Chinese capital market.

3.3. Payout Practice in the Chinese Stock Market

The dividend payout practice in the Chinese market is more inclined to be subject to

other company policies, such as the investment plan, rather than the popular ‘sticky’

dividend payout policy in developed markets. The tight control on the IPO quota imposed

by CSRC makes equity funds precious to listed firms. Even after the abolition of the IPO

quota system, the scarcity of capital funds still dominates the agenda of listed companies.

Therefore, Chinese listed firms prefer to hoard their cash rather than make distributions to

shareholders, which can be regarded as a symptom of the free cash flow problem, even if

the listed company does not have an outstanding investment opportunity. In addition, the

existence of direct/ultimate controlling shareholders with a block of voting rights has made

minority shareholders unable to challenge the listed firms’ decisions to hoard the cash.

48 If institutional investors’ income tax rate is lower, there would be no tax refund.

68

In response to the growing dissatisfaction of negotiable shareholders, CSRC has

imposed several regulations, including the content covering the cash dividend distribution,

with the purpose of protecting the rights of minority shareholders and reinforcing their

confidence in the equity market.

3.3.1. Evolution of regulatory requirement on cash payout

The first issue that CSRC targeted was the wide practice of paying cash dividends to a

certain group of shareholders, but stock dividends to other shareholders in the early period

of the Chinese equity market. The motivation for this discriminative treatment could be

attributed to the fact that minority shareholders prefer stock dividends and controlling

shareholders prefer cash distributions.49 Therefore, CSRC published ‘Notice of Several Issues

about Standardizing the Practice of Listed Companies’ in 1996, which specified that this

discriminative practice should be terminated immediately and all shareholders should

receive the same type of dividends.50 Meanwhile, in the same documentation, CSRC

required listed companies to specify the source of the stock dividend, as coming from either

capital reserves or from retained earnings.

CSRC announced ‘Measures for the Administration of Share Issuance by Listed

Companies’ in 2001, which attempted to link the dividend payout with the approval of new

shares issuance. CSRC required the underwriter to prepare a due diligence report which

49 The negotiable shareholders’ preference for stock dividends might be partly due to imperfect investoreducation at the beginning of the Chinese stock market. Chinese negotiable shareholders thought they couldreceive more shares after stock dividend decisions and the high share price caused by the partial SIP madethese new shares more attractive, but they failed to realize that the share price would be diluted after thebonus shares were floated.50 To some extent, this regulation was to protect the status of state owned controlling shareholders becausethe aforementioned practice would lead the stake of state owned shares to be diluted as the externalshareholders would receive more common shares and increase their shareholding proportions.

69

includes the board of directors’ explanations if no profit distribution has been made in the

past three years. Similar rules are imposed on listed companies by CSRC via ‘Provisions to

Enhance the Protection on the Rights of Public Shareholders’ in 2004, which requires listed

firms to emphasize the payback to investors and include the general policy of profit

distribution in the corporate charter. The board of directors has to announce distribution

proposals in periodic reports unless a detailed explanation with the endorsement of

independent directors is provided. More importantly, any firm which has not made a profit

distribution in the previous three years will not be approved to raise more capital by the

SEO, rights issues or convertible bonds. Besides the three year payout history, ‘Measures of

Security Issuance by Listed Companies’, enforced in 2006, required any listed firm applying

for issuance of new securities should satisfy 20%minimum payout ratio, in the form of cash

and/or stock dividends. This minimum payout ratio for issuance of new securities was

escalated to 30% in 2008 by CSRC’s ‘Decision to Amend Provisions on Cash Dividend

Distribution of Listed Companies’51 and the type of payout was specified as cash dividends.

Meanwhile, in CSRC Announcement [2011] No. 46, the disclosure requirements were

extended to require the board of directors to report the execution of distribution plans, the

actual payout amounts and payout ratios in the past three years. The board of directors also

has to include the current profit distribution plan in the financial statement. If no

distribution is proposed but retained earnings are positive, the board of directors has to

explain the zero cash distribution decision and how the retained earnings will be used.

The above regulations suggest that the efforts of the regulatory body in China are

remarkable. Although their purpose is to enhance the payout of listed firms and to protect

51 It is the first CSRC regulation solely focusing on the topic of dividend policy.

70

the rights and interests of minority shareholders, the effect of these measures needs to be

scrutinized by reviewing the payout practice in the Chinese stock market.

3.3.2. Snapshot of payout practice in the Chinese stock market

As shown in Table 3.1, by the end of 2009, there were 1,657 companies listed or once

listed52 on the main board A Share market, of which only 284 firms, or 16.46 per cent, had

reported a continuous cash dividend history for no less than 5 years. More than half of the

Chinese listed firms make discretional cash dividend payments, while 70 firms have never

paid a cash dividend in their history.

Table 3 1 Distribution of listed firms with continuous cash dividend payout (1992 2009)

Total Never Pay Discretional 2 – 4 years 5 – 9 years 10 – 14 years 15 19 years

1,657 70 908 399 200 78 2

100% 4.2% 54.8% 24.1% 12.1% 4.7% 0.1%

Source: CSMAR

With regard to the ratio of cash payers and all listed firms, the pattern is mixed, as

shown in Figure 3.5. Between 1992 and 1994, the percentage of payers increased steadily

and peaked at 75.7 per cent in 1994. Since then, the proportion of cash payers plunged to

less than 30 per cent in 1997 and 1998. This significant movement was accompanied by an

influx of IPOs between 1995 and 1999, when the number of listed firms increased from 313

to 92453. The leap in the number of listed firms, however, did not lead to a simultaneous

increase in the quality of the listed firms. Similar to other emerging markets, the Chinese

market was also disturbed by problems such as earnings management and insider trading.

Since 2000, the growth rate in the number of listed firms has been moderate because the

52 Including delisted firms due to continuous accounting losses or mergers and acquisitions.53 The majority of the change occurred in 1996 and 1997, in which the increase in percentage was 65% and40%, respectively.

71

authority attempted to improve the quality of listed firms in order to respond to the

negative feedback from investors about their weak positions when confronting the listed

companies regarding the aforementioned problems. The proportion of listed firms which

paid cash dividends increased significantly to above 60 per cent in 2000. It is noticed that

CSRC published ‘Measures for the Administration of Share Issuance by Listed Companies’, in

which CSRC required the underwriter to include an explanation from the board of directors

in the due diligence report on issuing new shares, if the underlying firm had not made a

dividend distribution in the previous three years. It appears that the listed firms responded

positively to the signal that the CSRCmight tighten the regulations on profit distribution.

Figure 3 5 Number of listed firms and percentage of cash dividend payers (1992 2010)

Source: CSMAR

But the peak in 2000 was not sustained and the indicator slid gradually until 2005,

except for a spike in 2004. It is noteworthy that the CSRC promulgated the ‘Provisions to

Enhance the Protection on Rights of Public Shareholders’ in 2004, which included the first

set of practical means to improve profit distribution. Since 2006, when ‘Measures of

Security Issuance by Listed Companies’ was published and set the threshold of the 3 year

53 182 288 313515

722826

9241,062 1,137 1,200 1,264 1,353 1,352 1,411

1,527 1,603 1,657

1,887

52.8%

60.4%

75.7%

56.5%

30.9%28.4% 28.5%

30.7%

62.2%58.8%

51.0%47.3%

53.7%

45.5%48.6%

50.6% 51.9%53.7%

59.0%

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

0

200

400

600

800

1000

1200

1400

1600

1800

2000

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

Number of listed firms Percentage of cash payers

72

average payout to total earnings for further equity financing, the proportion of cash payers

has slightly improved and is maintained at above 50 per cent.

Figure 3 6 Average cash dividend payout ratio of cash payers (1992 2010)

Source: China Securities Regulatory Commission

Besides the proportion of cash payers, the average cash payout ratios of cash payers

by year show themagnitude of the cash distribution in the Chinese stock market. Generally,

a downward trend in the cash payout ratio has been observed in recent years, as depicted

in Figure 3.6. Since 1994, the average payout ratio of cash payers has been maintained at a

stable 50 per cent level, and peaked at 71.4 per cent in 1998, but Figure 3.5 suggests the

proportion of cash payers plunged to a record low of around 30 per cent in the same period.

Less than 30 per cent of firms made cash dividend payments but distributed a significant

portion of their net profits, which is similar to the situation in the US (DeAngelo et al., 2004,

Skinner, 2008). After 2007, the average payout ratio of cash payers remained at around 50

per cent until it dropped to 36.2 per cent in 2007. The rebound in average payout ratio, as

of 2008, may be a positive response to the CSRC‘s ‘Decision to Amend Provisions on Cash

Dividend Distribution of Listed Companies’, which reflected listed firms’ concerns about the

regulation that linked the three year average cash payout ratio with SEO qualification. The

20.4%

41.1%

55.0% 53.8% 55.6% 56.6%

71.4%

59.3%

53.5%49.3%

57.1%52.9%

56.2%53.5%

49.1%

36.2%

47.9%

41.0%

35.4%

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010

73

trend of the cash payout ratio since 2000, along with cash payers’ proportions in 2000, 2004

and 2006 shown in Figure 3.5, indicate that listed companies are subject to the significant

administrative power of the CSRC. The aforementioned regulations can stimulate the

payout behaviour in the short run, but it may need more internal motivation for the listed

firms to change their attitudes towards profit distribution.

3.4. Summary

This chapter briefly reviews the history of the Chinese stock market, relevant

legislation and the payout practices. Similar to many other emerging markets, the Chinese

stock market was part of an economic reform and has occupied a very important position

in the national economy. Compared with the situation in other regimes, the Chinese stock

market is still strongly influenced by the regulatory authorities. The CSRC has imposed

several regulations on cash dividend distributions, although the situation, compared with

the early stages of the stock market, has improved a lot. However, Chinese listed firms still

prefer to hoard cash, as shown by the trends in the cash payout ratio. It suggests that the

agency cost of free cash flow is quite significant in the Chinese market. Some previous

literature has covered this area, but the results are inconclusive. Therefore, the following

three chapters will look into the connection between dividend practice and several issues

in corporate finance and asset pricing, and attempt to substantiate the research on

emerging markets.

74

Chapter 4 Ultimate Controlling Shareholders and Dividend Payout

Policy in the Chinese Stock Market

4.1. Introduction

Agency problems caused by the conflict of interest between shareholders and

managers has been studied thoroughly under the Anglo American corporate governance

infrastructure which has dispersed ownership, where the cash dividend payout can be

viewed as a shareholders’ tool to tackle the principal agent problem (Jensen and Meckling,

1976, Rozeff, 1982, Jensen, 1986). This structure contrasts with recent studies on the

alternative type of agency problem between large and small shareholders in other

developed and emerging markets, where the ownership of shareholdings in firms is more

concentrated (Claessens et al., 2000, Gugler and Yurtoglu, 2003, Attig, 2007, Renneboog

and Trojanowski, 2007, Truong and Heaney, 2007). Faccio et al. (2001) propose that cash

dividends can be used to deal with the agency problem between large shareholders and

minority shareholders, as a higher payout reduces cash holdings held at the discretion of

large shareholders, and alleviates minority shareholders’ concern about wealth

expropriation.54 However, a high cash payout does not necessarily mitigate, but rather

exacerbates the agency problem, especially in regulated economies such as the Chinese

market where large shareholders are unable to liquidate their shareholdings easily due to

regulatory constraints (Chen et al., 2009b, Huang et al., 2011).

54 This hypothesis is based on the fact that the large shareholders are board members (insiders) and able tomanipulate the business and financial decisions. Empirical evidence fromWestern European family controlledfirms supported this argument, but a contradictory phenomenon is observed in Asian markets. Truong andHeaney (2007) also document that the magnitude of cash dividend payout is undermined when the largeshareholders are insiders.

75

This chapter investigates whether the control rights of the ultimate controlling

shareholders of a listed firm in China influence the firm’s cash dividend policy. Most

previous literature on the Chinese market uses the percentage of non negotiable

shareholdings as a proxy of ownership/control by the largest shareholder (Gul, 1999, Cheng

et al., 2009, Wei and Xiao, 2009, Huang et al., 2011). The most important feature of non

negotiable shares is that investors are unable to realize capital gains by direct selling in the

secondary market, which makes it more attractive to realize their investment return by

other means, such as receiving cash dividends or conducting tunnelling behaviours. The

empirical results of the above studies are in line with the hypothesis that non negotiable

shareholders prefer cash dividends, and negotiable shareholders prefer stock dividends.55

This situation has changed, however, as the China Securities Regulatory Commission (CSRC)

started the split share structure reform (SSSR) in 2005, aiming to make listed firms fully

floated in the secondarymarket. At the end of 2012, more than 99 per cent of Chinese listed

firms have completed the SSSR.56 As the separation between negotiable and non negotiable

shares gradually phases out, the special feature of non negotiable shares, such as the

preference to cash dividend (Cheng et al., 2009), will disappear accordingly.

On the other hand, using non negotiable shares as a proxy for ownership

concentration in the Chinese market has its limitations. The obvious reason would be that

it might lead to biased estimation of large shareholders’ actual influence, because it only

reflects the direct cash rights of the controller but ignores the indirect control through the

55 Stock dividends in the Chinese market are either from retained earnings or capital reserves, which aresimilar to bonus shares and stock splits, respectively.56 A common practice in this reform is that non negotiable shares are imposed with a lock up period from oneyear to three years. These shareholders can liquidate their shares after the lock up period expires.

76

entities affiliated with the same controller. It has been documented that investors with a

large block of shareholdings pursue economic benefits beyond the profit distribution

associated with only cash flow rights (Shleifer and Vishny, 1986, Prowse, 1992, La Porta et

al., 1999, Benjamin, 2006). When Truong and Heaney (2007) investigated the relationship

between the largest shareholder and payout policy, they used the sum of direct and indirect

shareholdings controlled by the largest shareholder as the proxy of ownership.

Since 2003, the CSRC has required listed firms to disclose their ultimate controlling

shareholders, as well as the percentage of voting (control) rights under their control, in

annual financial statements. In addition, when there is any change in the ultimate

controlling shareholders and their level of control rights, listed firms have to make

immediate disclosure via stock exchanges.57 This chapter aims to utilize the unique data to

test whether the level of control rights held by UCSs is able to explain the payout pattern in

China. It is argued that the level of control rights held byUCS is a better proxy for ownership

of control than the non negotiable shareholdings, because the level of control rights

provides a more precise estimate of the actual control from the large shareholders, similar

to the widely accepted measurement of ownership in the research on other markets.

This chapter also investigates whether different types of UCSs have an impact on the

listed firms’ payout policy. Chen et al. (2009c) examine how the type of UCS influences the

57 According to ‘Measures for the Administration of the Takeover of Listed Companies’ published by CSRC,investors will be classified as ultimate controlling shareholders if they meet any of the following four criteria:(i) investor that controls the highest percentage, individually or jointly with other investors, of listed firms’common share among all shareholders; (ii) investor that controls, individually or jointly with other investors,no less than 30% voting rights; (iii) investor that is able to nominate and determine, individually or jointly withother investors, more than half of the board of directors; (iv) investor that is able to determine the underlyinglisted firm’s financial and operating policies, as well as benefit from the underlying listed firm’s operation. Infact, the definition of ultimate controlling shareholder in China is very similar to the concept of largestshareholder in Truong and Heaney (2007).

77

operating performance of Chinese listed firms, in which they define UCSs into four

categories, Bureau of State owned Assets (SAMB),58 State owned Enterprises affiliated to

Central Government (SOECG), State owned Enterprises affiliated to Local Government

(SOELG), and Private investors (Private).59 They argue that these UCSs have different

operating targets, accessibility to financial resources and managerial expertise and,

consequently, exert divergent influence on the operating performance of controlled listed

firms. As rational investors,UCSs have their own business blueprints and attempt to achieve

return from the investment in their controlled entity. However, UCSs cannot realize capital

gains by selling their holdings in the secondary markets due to holding restrictions on their

stocks. The CSRC has also imposed several regulations forbidding controlling shareholders

to occupy listed firms’ funds, so UCSs need to use conventional measures, such as cash

dividends, to retrieve their investment return in the form of cash from listed firms.

Different types ofUCSs have divergent financial capacity and flexibility, their demands

for cash dividends from listed firms may reflect the financial constraints they face and the

flexibility in their decision making. For example, Private UCSs are deemed to have weaker

financial capacity, but aremore flexible than state ownedUCSs. Therefore, Private UCSs are

more likely to instruct the controlled public firm to make more cash distribution.60 In

contrast, SOECG UCSs are closely connected with the central government and the capital

58 The state level agency that monitors the state owned assets is named as State owned Assets Supervisionand Administration Commission (SASAC). The similar agency in other levels of government is named as Bureauof State owned Assets (SAMB). Although their names are different, their solo mission is to manage stateowned assets. So SASAC controllers are classified within the SAMB group in this chapter.59 The first three types of UCSs, SAMB, SOECG and SOELG, are further defined as state owned UCSs. Privateinvestors include both private institutions and individuals.60 Although a cash dividend is subject to income tax, institutional shareholders have an advantage as thebefore tax cash dividend will be aggregated into their revenue. However, individual investors aredisadvantaged because they receive an after tax cash dividend and the withholding tax has been paid by listedfirms.

78

market, which have sufficient financial capacity and, therefore, are deemed to have less

demand for cash dividends from controlled public firms. Therefore, this chapter aims to

differentiate the cash payout pattern of listed firms with various types of UCSs.

The contribution of this chapter consists of two aspects. This is the first study which

uses the control rights held by ultimate controlling shareholders to measure the ownership

concentration in the research on the payout pattern in the Chinese market. Instead of the

non negotiable shareholdings, control rights, measured as the investor’s aggregate direct

and indirect shareholdings, provide a better proxy of the largest shareholder’s influence on

the listed firm, as it reflects the affiliation among multiple large shareholders and is

independent from any change in regulations on the negotiability of common shares.

Second, the popular categorization based on state private investors in the Chinese market

may not be sufficient as different types of state owned UCSs have divergent operating

targets and financial capability. To differentiate, the types of ultimate controlling

shareholders will supply more insights about how Chinese listed firms’ cash payout policies

are influenced by their UCSs with divergent backgrounds and features.

With a sample of 6,386 observations from 1,231 Chinese listed firms between 2003

and 2010, this chapter shows that the level of control rights held by UCSs has a positive

influence on both the probability and magnitude of cash dividend payouts, in line with

findings of Truong andHeaney (2007). The result can be attributed toUCSs’ attempt to force

the management to distribute excessive cash balance and reduce the agency cost of free

cash flow. Since most common shares controlled by UCSs during the observation window

are either non negotiable or within a lock up period, these outcomes are consistent with

Cheng et al. (2009) who found that non negotiable shareholders prefer cash dividends.

79

The empirical results also suggest that various types of UCSs exert divergent influence

on the probability andmagnitude of cash dividends. Comparedwith firms with State owned

UCSs, firms with Private UCSs exhibit marginally higher probability and a larger cash

distribution. Within the subgroup with State owned ultimate controllers, firms controlled

by SOELG and SAMB UCSs exhibit a lower probability of making a cash dividend distribution,

while firms with SOECG UCSs report a higher probability of a cash dividend announcement,

and firms with SAMB controllers show lower cash payout magnitude. These outcomes can

be attributed to the divergent backgrounds and features of their controlling shareholders.

The remainder of this chapter is structured as follows. Section 4.2 proposes the

hypotheses. Section 4.3 introduces the data and descriptive statistics. Section 4.4 reports

and discusses the empirical results. Section 4.5 summarizes the chapter.

4.2. Hypotheses development

Although the dividend irrelevance theorem proposes that the cash dividend payout

has no influence on the firm’s valuation in a frictionless capital market (Miller and

Modigliani, 1961), the practice surrounding the cash dividend policy results in plentiful

academic research to answer the question of why public companies pay a cash dividend.

Because of the separation between ownership and management in Anglo American

markets, entrepreneurs in public firms have the motivation to make decisions in favour of

themselves but at the cost of shareholders, which is defined as the agency principal

problem (Jensen and Meckling, 1976). Hence, the cash dividend policy is conjectured to be

a corporate governancemechanism to reduce the free cash flow andmonitor themanagers

(Easterbrook, 1984, Jensen, 1986, Lang and Litzenberger, 1989, La Porta et al., 2000a).

80

However, recent studies indicate that East Asian markets are dominated by concentrated

ownership and family pyramid structures (La Porta et al., 1999, Claessens et al., 2000, Yeh

and Woidtke, 2005, Wei and Zhang, 2008), along with weak legal infrastructure and

insufficient investor protection schemes (Shleifer and Vishny, 1997). In these markets, the

traditional agency problem between shareholders andmanagers has been overtaken by the

conflict of interest between large and small shareholders, which leads to the research on

large shareholders’ wealth expropriation from the minority shareholders. Some studies,

such as Faccio et al. (2001), Bae et al. (2002), Vladimir (2005), Baek et al. (2006), and Jiang

et al. (2010), document significant expropriation activities, while other studies argue that

the practice of large shareholders in East Asian markets swings between expropriating

minority shareholders’ wealth and propping up the listed firm (Friedman et al., 2003,

Cheung et al., 2006, How et al., 2008, Cheung et al., 2009a, Chen et al., 2011).

Most studies on the cash dividend payout in the Chinese market are based on the

unique ownership structure created by the partial share issuance privatization (SIP) that

separates the non negotiable shares from negotiable shares.61 As the Chinese capital

market evolves, some sophisticated investors utilize the concentration of non negotiable

shareholdings to control listed firms with a multiple tier shareholding structure. These

investors become the so called ultimate controlling shareholders and benefit from

tunnelling resources from listed companies, such as transferring price and inter company

loan guarantees, with their dominating voting rights. The lack of minority investor

61 Initially, non negotiable shares were equivalent to state owned shares, while negotiable shares were soldto individual investors with a hefty premium to compensate the state owned investors for the restrictedtrading status of non negotiable shares. State owned non negotiable shares are gradually allowed to be soldto private institutional investors at prices higher than book value but lower than secondary market values inthe block trade system, a market between the primary and secondary markets. Therefore, non negotiableshares are no longer identical to state owned shares.

81

protectionmechanisms exacerbates the expropriation in the Chinesemarket (Mitton, 2004,

Wang et al., 2008, Jiang et al., 2009, Ng et al., 2009). Due to the tightening of CSRC

regulations on tunnelling behaviours, cash dividends are regarded as a justified means to

realize investment return and extract funds from the controlled listed firms. Intuitively, a

higher level of control rights will facilitate UCSs to increase both the propensity and the

magnitude of controlled entities’ cash dividend payouts.

Previous studies have found that non negotiable shareholdings are positively

associated with both the probability and the magnitude of the cash dividend policy, while a

positive linkage between a high percentage of negotiable shareholdings and stock dividend

announcements is observed (Gul, 1999, Cheng et al., 2009). Recent research confirms the

aforementioned positive relationship between non negotiable shareholdings and cash

dividend payouts, as well as suggests new findings that conventional factors, such as

profitability and cash availability, play important roles in the process of designing dividend

policy (Huang et al., 2011).62 Similar to Short et al. (2002), Gugler and Yurtoglu (2003) and

Truong and Heaney (2007), a positive relationship between ownership and dividend payout

is proposed in this chapter, however, the proxy of ownership is not restricted to the non

negotiable shareholdings controlled by any single large shareholder, but is extended to the

control rights held byUCSs, because it provides amore precise of the influence of the actual

controller of the firm. Therefore, the first hypothesis is derived as following:

62 One of the findings in Huang et al. (2011) is that controlling shareholders are unable to exploit listed firms’cash by requiring excessive cash dividends when earnings experience a significant decrease. It suggests thecontrolling shareholders have to consider the feedback from minority shareholders and compromise if theminority shareholders’ claim is reasonable.

82

Hypothesis 4.1 The level of control rights held by UCSs positively influences both the

propensity and magnitude of the cash dividend.

Besides the level of control rights, the type of ultimate controlling shareholders is

another important factor which may influence the cash payout propensity and magnitude.

In practice, different types of UCSs possess unique operational targets, management styles

and financial capacities, which influence the UCSs’ monitoring of the operation of the listed

firms (Chen et al., 2009c). As an extension of Chen et al. (2009c), this chapter conjectures

that different types of UCSs will influence the cash payout policy of the underlying listed

firms because the divergent financial capacity and degrees of autonomy result in different

appetites for the cash flows from controlled entities. This chapter follows the categorization

of UCSs by Chen et al. (2009c) intofour types of UCSs: SAMB, SOECG, SOELG, and Private

UCSs. SAMBs are government agencies which hold state owned shares with the mission to

maintain and increase the value of state owned assets. SOECGs include state owned

enterprises under the direct control of central government agencies, which have wide

autonomy and are able to form conglomerates by investing in other listed companies.

SOELGs are different from SOECGs as they are controlled directly bymunicipal governments

and were created to manage the spin offs from the state owned enterprises previously

owned by central government. The private investment institutions and individual investors

are defined as Private UCSs. Intuitively, Private UCSs are more profit oriented but have less

support from the state controlled financial sector, which makes them more inclined to

extract cash from the controlled listed firms. Therefore, it is likely that Private UCSs will

arrange for the listed firms to pay more cash dividends and redirect the funds to other

controlled entities. In the meantime, not all state owned UCSs have equal financial capacity

83

and commitments beyond economic profits. As SAMB UCSs are effectively government

agencies and the directors appointed by SAMB UCSs are public servants, they have the least

interest in exploiting the economic benefit from the listed firms. Instead, they are more

committed to support the government to administrate the society. Dissimilarly, SOECG and

SOELG UCSs are incorporated legal persons which are more focused on the profit and loss

of the controlled entities. This difference has led to divergence in the operating

performance, among which SOECG UCSs exhibit far better profitability and efficiency than

firms with SAMB UCSs do (Chen et al., 2009c). Consequently, the cash dividend policy of

firms with different UCSs may exhibit variance because of the strong association between

cash payout and retained earnings. Furthermore, not only does the close connection

between SOECG UCSs and central government entitle them to greater financial resources,

but it also makes them more influenced by the securities regulatory agency, CSRC, which

has been encouraging listed firms to increase the cash distribution in recent years. As the

firms controlled by SOECG UCSs are the best performers (Chen et al., 2009c), it is likely

SOECG UCSs are under pressure to organise the controlled listed firms to distribute more

cash dividends. Hence, the second hypothesis is proposed as below:

Hypothesis 4.2 The types of UCSs exert divergent influences on both the propensity and

magnitude of the cash dividend. Listed firms with Private and SOECG UCSs are more likely

to announce cash dividends and pay more cash dividends.

4.3. Data and descriptive statistics

The firm level financial data of Chinese listed firms is derived from the China Stock

Market and Accounting Research (CSMAR) database. The observation window commences

84

in 2003 and ends in 2010. Several filtering criteria are applied to the dataset, including: (i)

samples will include all firms except those belonging to the financial industry, as financial

companies have special features in their financial statements, especially size and leverage;

(ii) firms with missing financial information, such as total assets, control rights and types of

UCSs, will not be included in the sample; (iii) companies that report non positive equity will

also be excluded from the sample; (iv) observations with negative earnings are excluded;

(v) firms which pay stock dividends from retained earnings are excluded;63 (vi) firms which

have been delisted or have had special treatment (ST) and/or particular treatment (PT)

imposed, are excluded. After the filtering process, the sample contains 6,386 observations

from 1,231 firms.64 Next, the top and bottom 1 per cent of the dependent and independent

variables are winsorized.

4.3.1. Proxies for cash dividend policy, control rights and type of UCSs

Cash dividend policy has two aspects, the propensity and magnitude. First, the

propensity of the cash dividend payout is proxied by a dummy variable, DUMMY_CASHDIV,

which takes the value of 1 when a cash dividend is announced in the underlying period, and

0 otherwise. Second, the magnitude of the cash dividend payout is measured by the cash

63 As per Chinese accounting standards, the stock dividend from retained earnings (SDRE) is subject to incometax which is withheld and paid by the firm. The firm which announces stock dividends from retained earningshas to pay a small amount of cash dividend simultaneously to offset shareholders’ withholding tax liabilities.So a cash dividend is endogenous with a stock dividend from retained earnings. A stock dividend from capitalreserve (SDCR) is exempt from withholding income tax because it is not classified as a distribution of retainedearnings.64 The filtering process is inclined to maintain firms with better operating performance and financial capacity,due to the exclusion of firms which are imposed ST and/or PT. These ST and/or PT firms are generally labelledas firms with financial irregularity and/or consecutive years with negative profits and less likely to makedividend distribution. Therefore, this criterion alleviates the divergence among observations and makes theresearch focus on ‘normal’ Chinese listed firms.

85

payout ratio, CASH_PAYOUT, which is a cash dividend per share scaled by earnings per

share as of any specific year.

As mentioned, the first purpose of this chapter is to test for an association between

cash dividend policy and control rights held by UCSs. The level of voting rights, which is

measured as the aggregate direct and indirect shareholdings, controlled by UCSs,

CONTROL_RIGHTS, is introduced as a key independent variable that measures the

influential power of the UCSs on the underlying listed firms.65 A significant positive

coefficient on CONTROL_RIGHTS may suggest that UCSs utilize their advantaged positions

to extract more cash returns from listed firms, while a significant negative coefficient on

CONTROL_RIGHTSmay indicate UCSs prefer hoarding the cash balance within the firm, also

known as the agency cost of free cash flow. Meanwhile, a dummy variable,

Dummy_LowControl, is included as a parameter for low control rights, which equals to 1 if

CONTROL_RIGHTS is less than 20 per cent, and 0 otherwise. According to CSMAR database,

some firms report ultimate controlling shareholders with lower percentage of control

rights. The inclusion of Dummy_LowControl is to control for observation with low level

control rights and its coefficient is expected to be opposite to that of CONTROL_RIGHTS.

A set of dummy variables, UCS_Dummy, including DPrivate, DSOECG, DSOELG and

DSAMB, are generated to label listed firms with various types of UCSs in any specific

financial year. These dummy variables take the value of 1 when the UCS of underlying listed

firm belongs to a certain UCS type, and 0 otherwise. The coefficients on UCS_Dummy are

expected to exhibit divergent signs and significances, which indicate how various types of

UCSs influence the payout patterns of controlled public firms. The identification of UCS

65 CASH_RIGHTS, which is the direct shareholding owned by UCSs, are also included in the robustness check.

86

types follows the ‘Classification Standard for Related Party Nature of Enterprise’ of the

‘China listed firm’s shareholder research database’ published by CSMAR. UCSs with

classification codes 2111 (State owned Assets Supervision Administration Commission),

2121 (Bureau of State owned Assets, directly under the Provincial and Regional

Government), 2131 (Bureau of State owned Assets, directly under the Municipal

Government), and 2141 (Bureau of State owned Assets, directly under the County level

Government), are defined as SAMB. UCSs with classification codes in category 2110 (State

Council, except 2111) are defined as SOECG.UCSs with classification codes in category 2120

(Provincial & Regional Government, except 2121), 2130 (Municipal Government, except

2131) are defined as SOELG. UCSs with classification codes in categories 1200 (Private

Enterprises) and 3000 (Natural Person) are defined as Private.66 Table 4.1 presents the

distribution of observations grouped by types of UCS. The whole sample contains 6,386

observations, among which 4,243 (66.4%) make cash dividend payouts.67 Observations with

Private and SAMB UCSs have the majority of the weight in the whole (cash payer) sample,

which occupy 33.6% (33.5%) and 43.7% (44.5%), respectively.

Table 4 1 Sample distribution by types of UCSs

Private SOECG SOELG SAMB

No. % No. % No. % No. %

Panel A: Whole sample

6,386 2,146 33.6 442 6.9 1,010 15.8 2,788 43.7

Panel B: Cash payer

4,243 1,421 33.5 318 7.5 616 14.5 1,888 44.5

This table provides a summary of sample firms by years and types of UCSs, between 2003 and 2010. Panel Apresents the distribution of listed firms, while Panel B reports the distribution of listed firms which make acash distribution. % is the proportion of each UCS category to the total number of firms.

66 759 observations with ambiguous classification code 1100 (State owned enterprise) are investigatedmanually by looking into their annual reports andUCSs’ websites, amongwhich 295 observations are classifiedas with SOECG UCSs and 464 observations are determined as with SOELG UCSs.67 Among these 6,386 observations, 726 firm years announce a SDCR with simultaneous cash dividend, and220 firm years announce a SDCR without simultaneous cash dividend.

87

4.3.2. Control variables

The variables used in this chapter are presented in Table 4.2. Several explanatory

variables are introduced to control firm specific features. Firm size, Size, is included as the

control for size effect because the dividend payout is positively correlated with the firm’s

asset base (Gul, 1999, Mitton, 2004, Blau and Fuller, 2008, DeAngelo et al., 2004, Zhang,

2008). As larger firms are more likely to make dividend payouts and pay more cash

dividends, it is expected that the coefficient on SIZE will be positive. The Price_to_Book

ratio, PRICE_TO_BOOK, is included as the control for the growth opportunity and is

expected to show a negative coefficient, in that, past literature concludes that firms with a

higher Price_to_Book ratio are deemed to have more investment opportunity and become

reluctant to distribute cash dividends (Kim et al., 1998, Chang et al., 2007, Tim et al., 1999,

Truong and Heaney, 2007).68 More growth opportunity may well imply an over investment

problem caused by an abundant free cash flow (Morck et al., 1988, Lang and Litzenberger,

1989, Gordon and Myers, 1998, Berkman et al., 2009, Hamadi, 2010). Return on equity,

ROE, is introduced as a proxy for profitability, because previous studies have confirmed the

positive connection between earnings and cash dividends (Conroy et al., 2000, Brockman

and Unlu, 2009, Huang et al., 2009, Lintner, 1956, Fama and French, 2001, Ibbotson and

Chen, 2003, Truong and Heaney, 2007). Besides the profitability, a firm’s liquidity situation

is controlled by FREE_CF, the ratio of free cash flow to equity scaled by year end total

equity.69 Chinese listed firms are inclined to announce annual financial statements earlier

68 A higher market valuation caused by more growth opportunity may also be linked with an over investmentproblem due to an abundant free cash flow.69 Free cash flow to equity is the product of the year end number of shares and the free cash flow per share,published in ‘China Stock Market Financial Database – Financial Ratios’ by CSMAR. Cheng et al. (2009) usedcash equivalent to total assets as the proxy for liquidity, but it is arguable that free cash flow to equity reflectsthe capacity to pay a cash dividend.

88

when operating results are satisfactory, but defer announcements when performance is

poor (Haw et al., 2005, Cheng et al., 2009, Chen et al., 2005a). This timing effect is controlled

by RPTTIMING, the logarithm of the gap in days between the financial statement date and

announcement date, which is anticipated to show a negative coefficient.

Table 4 2 Variable definition

DUMMY_CASHDIV dummy variable which equals to 1 when there is a cash dividend payout in year t, and 0 otherwiseCASH_PAYOUT cash payout ratio which is cash dividend per share scaled by earnings per share as of year t

CONTROL_RIGHTS the aggregate direct and indirect shareholdings controlled by ultimate controlling shareholders

Dummy_LowControl dummy variable which equals to 1 when CONTROL_RIGHTS is less than 20% in year t, and 0 otherwise

UCS_Dummy dummy variables which include DPrivate, DSOECG, DSOELG and DSAMB defined below

DPrivate dummy variable which equals to 1 when the ultimate controlling shareholder is a privately owned enterprise ornatural person in year t, and 0 otherwise

DSOECG dummy variable which equals to 1 when the ultimate controlling shareholder is state owned enterprises affiliatedwith central government in year t, and 0 otherwise

DSOELG dummy variable which equals to 1 when the ultimate controlling shareholder is state owned enterprises affiliatedwith local government in year t, and 0 otherwise

DSAMB dummy variable which equals to 1 when the ultimate controlling shareholder is state owned asset managementbureau in year t, and 0 otherwise

CASH_RIGHTS the direct shareholdings (cash rights) owned by ultimate controlling shareholders

CONTROL_CASH_GAP the difference between CONTROL_RIGHTS and CASH_RIGHTS

SIZE the firm size which is the logarithm of the total assets as of year t

PRICE_TO_BOOK the price to book ratio as of year t, which is defined as market value of equity scaled by book value of equity

FREE_CF the free cash flow to equity scaled by total equity as of year t

ROE return on equity as of year t

LEVERAGE total interest bearing liabilities scaled by total equity as of year t

RPTTIMING the logarithm of the number of days between the financial year end and financial report and dividend decisionannouncement date as of year t

SDCR dummy variable which equals to 1 when there is a stock dividend from capital reserve in year t, and 0 otherwise

Industry Dummy a set of dummy variable indicating the industry of Utilities, Properties, Conglomerate and Industry

Year Dummy a set of dummy variables indicating the financial years from 2003 to 2009

Because a higher debt burden undermines the probability and magnitude of cash

dividends (Crutchley and Jensen, 1999, Chen et al., 2005b, Eije and Megginson, 2008,

Pattenden and Twite, 2008),70 financial leverage, LEVERAGE, defined as total interest

bearing liabilities71 scaled by year end total equity, is included as the proxy for financial risk

with an expected negative sign. Finally, a dummy variable indicating a simultaneous stock

70 Debt is also considered as a substitute to deal with the agency principal problem. But due to the closeconnection between the banking sector and state owned enterprises, and the unique features of the Chineseequity market, debt finance is not considered to be a corporate governance tool in Chinese market.71 Interest bearing liabilities are short term borrowing, notes payable, long term debts, and bonds payable.

89

dividend from capital reserve (SDCR), SDCR, is introduced as a control for the simultaneous

SDCR, as it is reported that negotiable shareholders prefer stock dividends (Cheng et al.,

2009). Year Dummies and Industry Dummies are included.72

4.3.3. Descriptive statistics

Table 4.3 presents the descriptive statistics of dependent variables and key

independent variables used in this chapter. As shown in Panel A, the average (median)

payout ratio of cash payers is 38.80 per cent (35.29%), slightly lower than the results of

other markets (Truong and Heaney, 2007). The average control rights held by UCSs,

CONTROL_RIGHTS, is 40.19 per cent, which is much higher than the average level in

developed markets (Truong and Heaney, 2007). The variance in CONTROL_RIGHTS is also

significant because the third quartile of CONTROL_RIGHTS (51.89%) is about twice that of

the first quartile (27.65%). With regard to the control variables, the positive skewness of

PRICE_TO_BOOK (mean 3.1812 vs. median 2.4848) suggests investors are moderately

optimistic about growth opportunity. The mean LEVERAGE is about 60 per cent while the

mean ROE is less than 10 per cent with negative skewness, which indicates that the Chinese

listed firms’ profitability does not benefit from financial leverage.

72 Based on Tier A definition of CSRC’s Guidance on the Industry Classification of Listed Companies, fiveindustries, Utilities, Properties, Conglomerate, Industry and Commerce, are defined.

90

Table 4 3 Descriptive statistics

Panel A: Whole sample

Variable Mean S.D. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.

CASH_PAYOUT (cash payers) 0.3880 0.2170 0.1363 0.2180 0.3529 0.5184 0.7091CASH_PAYOUT 0.2578 0.2546 0.0000 0.0000 0.2160 0.4294 0.6297

CONTROL_RIGHTS 0.4019 0.1567 0.1997 0.2765 0.3959 0.5189 0.6156

CASH_RIGHTS 0.3443 0.1778 0.1176 0.2017 0.3228 0.4806 0.6009

PRICE_TO_BOOK 3.1812 2.2635 1.1495 1.5885 2.4848 4.0678 6.1039

FREE_CF 0.3948 0.5446 1.1520 0.6716 0.2474 0.0000 0.1220

ROE 0.0916 0.0635 0.0188 0.0436 0.0812 0.1254 0.1760

LEVERAGE 0.5966 0.5328 0.0367 0.1911 0.4644 0.8584 1.3155

Private

CASH_PAYOUT (cash payers) 0.3743 0.2189 0.1219 0.2003 0.3346 0.5105 0.6996CASH_PAYOUT 0.2479 0.2512 0.0000 0.0000 0.1970 0.4173 0.6294

CONTROL_RIGHTS 0.3522 0.1473 0.1750 0.2386 0.3248 0.4500 0.5703

CASH_RIGHTS 0.2567 0.1577 0.0771 0.1289 0.2308 0.3569 0.4890

PRICE_TO_BOOK 3.7173 2.4652 1.2871 1.8719 3.0845 4.8389 7.0276

FREE_CF 0.4136 0.5445 1.1815 0.7132 0.2651 0.0000 0.1158

ROE 0.0972 0.0608 0.0249 0.0540 0.0885 0.1309 0.1720

LEVERAGE 0.5270 0.4750 0.0233 0.1649 0.4223 0.7640 1.1497

Panel B: subgroups with certain type of UCSSOECG

CASH_PAYOUT (cash payers) 0.3891 0.2329 0.1158 0.1978 0.3537 0.5107 0.7498CASH_PAYOUT 0.2799 0.2639 0.0000 0.0000 0.2281 0.4452 0.6812

CONTROL_RIGHTS 0.4071 0.1769 0.1739 0.2525 0.4013 0.5429 0.6500

CASH_RIGHTS 0.3570 0.1935 0.1209 0.1972 0.3178 0.5012 0.6456

PRICE_TO_BOOK 3.1894 2.1917 1.3062 1.7324 2.4587 3.8777 5.8776

FREE_CF 0.3556 0.5084 1.0162 0.6128 0.2401 0.0000 0.1132

ROE 0.0901 0.0639 0.0219 0.0466 0.0746 0.1213 0.1760

LEVERAGE 0.5416 0.5111 0.0139 0.1774 0.4252 0.7502 1.0763

SOELGCASH_PAYOUT (cash payers) 0.4448 0.2216 0.1748 0.2661 0.4106 0.5982 0.7868CASH_PAYOUT 0.2713 0.2776 0.0000 0.0000 0.2310 0.4698 0.6744

CONTROL_RIGHTS 0.4201 0.1568 0.2267 0.2981 0.4195 0.5455 0.6312

CASH_RIGHTS 0.3893 0.1711 0.1638 0.2563 0.3861 0.5234 0.6302

PRICE_TO_BOOK 2.7067 1.7675 1.1703 1.5210 2.1664 3.4038 4.7166

FREE_CF 0.3748 0.5148 1.0484 0.6287 0.2417 0.0000 0.1214

ROE 0.0781 0.0557 0.0142 0.0366 0.0683 0.1078 0.1511

LEVERAGE 0.6147 0.4898 0.0653 0.2353 0.5098 0.8990 1.2978

SAMB

CASH_PAYOUT (cash payers) 0.3795 0.2083 0.1401 0.2162 0.3469 0.5029 0.6823CASH_PAYOUT 0.2570 0.2467 0.0000 0.0000 0.2226 0.4237 0.6046

CONTROL_RIGHTS 0.4328 0.1508 0.2240 0.3155 0.4400 0.5448 0.6310

CASH_RIGHTS 0.3934 0.1663 0.1817 0.2607 0.3812 0.5260 0.6212

PRICE_TO_BOOK 2.9393 2.1927 1.0391 1.4403 2.2535 3.6955 5.6696

FREE_CF 0.3939 0.5603 1.1869 0.6600 0.2325 0.0000 0.1283

ROE 0.0925 0.0674 0.0168 0.0399 0.0788 0.1292 0.1890

LEVERAGE 0.6523 0.5840 0.0411 0.2040 0.4939 0.9591 1.4718

This table contains descriptive statistics for some variables of interest that are used in the following analysisof the 1,231 firms between 2003 and 2010. Refer to Table 4.2 for variable definition.

Panel B of Table 4.3 reports the descriptive statistics of observations with certain type

of UCSs. Cash payers with SOELG (Private) UCSs exhibit the highest (lowest) mean

CASH_PAYOUT, but the pattern turns upside down in the comparison of ROE and

91

PRICE_TO_BOOK, in which firms with SOELG (Private) UCSs report the lowest (highest)

mean ROE73 and mean PRICE_TO_BOOK. These statistics may imply that SOELG (Private)

controlled firms are more likely to be in a mature (growth) stage. In terms of financial risk,

SAMB controlled firms report the highest average LEVERAGE, 65.23%, followed by SOELG

controlled firms. Further, state owned UCSs show higher levels of CONTROL_RIGHTS, and

SAMB (Private) controlled firms exhibit the highest (lowest) average CONTROL_RIGHTS at

43.28% (37.43%). Not only can this be attributed to the legacy of the partial SIP which

entitled state owned enterprise UCSs to more non negotiable shareholdings, but it is also

consistent with the general consensus that Private UCSs are more active in the corporate

control market.

The Pearson’s correlation matrix of non dummy dependent variables and

independent variables is shown in Table 4.4. The correlation coefficient between

CASH_PAYOUT and CONTROL_RIGHTS is significantly positive, indicating that the co

movement between control rights and cash payout ratio is in the same direction. The

correlation coefficients between CASH_PAYOUT and control variables are in line with

expectation and will provide some guidance to the expected signs on the regression

coefficients in the empirical models. The positive correlation coefficient between

CASH_PAYOUT and ROE underpins the traditional relationship between earnings and

dividend, as reported in Lintner (1956). Meanwhile, the negative correlation coefficient

between CASH_PAYOUT and LEVERAGE implies that higher leverage will dampen the level

of the cash dividend. Further, the negative correlation coefficient between CASH_PAYOUT

73 This finding is contradictory to Chen et al. (2009b) that SOECG controlled firms perform the best, followedby firms with SOELG, while Private and SAMB controlled firms perform the worst.

92

and PRICE_TO_BOOK suggests that growth opportunity undermine the magnitude of cash

distribution as the firm needs capital funds to support the growth, while the positive

coefficient between CASH_PAYOUT and FREE_CF is in line with the conventional thought

that the management considers the liquidity situation when designing the payout policy.74

Table 4 4 Pearson's correlation matrix

N = 6,384 CASH_PAYOUT CONTROL_RIGHTS SIZE PRICE_TO_BOOK FREE_CF ROE RPTTIMING

CONTROL_RIGHTS 0.1580*

SIZE 0.0171 0.1983*

PRICE_TO_BOOK 0.0472* 0.0032 0.1746*

FREE_CF 0.1345* 0.0956* 0.1216* 0.0610*

ROE 0.0831* 0.1134* 0.2365* 0.3558* 0.1612*

RPTTIMING 0.0528* 0.0597* 0.0532* 0.0631* 0.0473* 0.1639*

LEVERAGE 0.1582 0.0597* 0.3175* 0.1250* 0.5626* 0.1092* 0.0055

This table contains Pearson’s correlation coefficients matrix for selected variables used in this analysis for thesample. Refer to Table 4.2 for variable definitions. *denotes significance at 1% level.

4.4. Empirical Analysis

4.4.1. Cash dividend policy, control rights and cash rights

A linear relation between control rights and dividend policy is proposed,75 following

the procedures of Cheng et al. (2009) with some alterations in the regression models. A

probit model, Eq.4.1, is developed to test the associations between the propensity for cash

dividends and the level of control rights, while a tobit model, Eq.4.2, is used to test the

linear association between themagnitude of cash dividends and the level of control rights.76

74 Among the control variables, there are three pairs reporting significant correlation coefficients, which areSIZE and LEVERAGE(0.3175), PRICE_TO_BOOK and ROE (0.3558), and, FREE_CF and LEVERAGE( 0.5626). Thepotential multicollinearity will be tested in the robustness check.75 Because the estimates produced by the probit model are z scores, the marginal effects at means arereported along with Eq.1 and to measure the economic significance.76 Although some literature proposes a quadratic relationship between the largest shareholders’ stake andthe propensity of a cash dividend decision (Crutchley and Jensen, 1999, Khan, 2006), this quadraticrelationship is largely found to be significant among common law countries, but not among civil law countries,such as China (Truong and Heaney, 2007).

93

The standard errors of both models are adjusted for heteroscedasticity (White, 1980). The

regression results are presented in Table 4.5 and Table 4.6, respectively.77

Column (4) in Table 4.5 shows the coefficient on CONTROL_RIGHTS is positive and

signficant at 1% level, supporting a positive contribution from control rights on the

likelihood of cash dividends. The marginal effect at the means, dy/dx, indicates a 10%

increase (decrease) in CONTROL_RIGHTS is expected to result in a 2.3% increase (decrease)

in the likelihood of a cash dividend announcement. Meanwhile, the coefficient on

Dummy_LowControl is aslo significantly negative at 1% level, which means firms controlled

by UCSs which have more competence in a potential proxy contest are less likely to

distribute cash dividends.

77 Both Eq.4.1 and Eq.4.2 are also estimated with lagged control rights and lagged control variables asregressors, which report similar results as the model specification of Eq.4.1 and Eq.4.2. The results areavailable on request.

94

Table 4 5 Control rights and the likelihood of cash dividends

DUMMY_CASHDIV ExpectedSigns (1) p value (2) p value (3) p value (4) p value

Intercept 0.116** (0.010) 0.100* (0.071) 3.661*** (0.000) 3.248*** (0.000)

CONTROL_RIGHTS + 1.368*** (0.000) 1.337*** (0.000) 0.822*** (0.000) 0.624*** (0.000)

0.497 0.485 0.285 0.216

Dummy_LowControl 0.029 (0.639) 0.177*** (0.009)

0.011 0.061

Size + 0.199*** (0.000) 0.203*** (0.000)

0.069 0.070

PRICE_TO_BOOK 0.094*** (0.000) 0.093*** (0.000)

0.033 0.032

FREE_CF + 0.026 (0.510) 0.028 (0.467)

0.090 0.010

ROE + 8.414*** (0.000) 8.405*** (0.000)

2.914 2.911

RPTTIMING 0.208*** (0.000) 0.208*** (0.001)

0.072 0.072

LEVERAGE 0.424*** (0.000) 0.431*** (0.000)

0.147 0.149

SDCR ? 0.273*** (0.000) 0.272*** (0.000)

0.095 0.094

Year Dummy No No Yes Yes

Industry Dummy No No Yes Yes

Pseudo R2 0.021 0.021 0.171 0.1722 165.6 166.3 918.6 929.7

This table reports the estimation results of probit model Eq.4.1. The error term is adjusted forheteroscedasticity (White, 1980). Refer to Table 4.2 for variable definitions. p values are reported inparentheses. Marginal effects at means are reported beneath the coefficients. *, ** and *** denote significanceat 10%, 5% and 1% levels, respectively.

Moreover, the regression results of Eq.4.2, shown in Table 4.6, also display a positive

impact from control rights on the magnitude of a cash payout. Column (4) of Table 4.6

suggests every 10 per cent increase in the level of control rights will improve the cash

payout ratio by 2.2% and the estimate is significant at 1% level. Similar to Table 4.4, the

payout ratio will decrease by 5.2 per cent if UCSs’ control rights are less than 20 per cent.

Along with the descriptive statistics, the above outcomes support Hypothesis 4.1, that the

level of control rights held by UCSs is positively associated with both the propensity and

magnitude of the cash dividend decision, and are consistent with the existing literature (Gul,

1999, Truong and Heaney, 2007, Cheng et al., 2009, Huang et al., 2011).

95

Table 4 6 Control rights and the magnitude of cash dividends

CASH_PAYOUT ExpectedSigns (1) p value (2) p value (3) p value (4) p value

Intercept 0.019 (0.144) 0.031* (0.059) 0.193 (0.100) 0.197* (0.097)

CONTROL_RIGHTS + 0.402*** (0.000) 0.380*** (0.000) 0.267*** (0.000) 0.216*** (0.000)

Dummy_LowControl 0.023 (0.240) 0.050*** (0.007)

Size + 0.023*** (0.000) 0.024*** (0.000)

PRICE_TO_BOOK 0.012*** (0.000) 0.012*** (0.000)

FREE_CF + 0.017 (0.110) 0.017 (0.102)

ROE + 0.846*** (0.000) 0.844*** (0.000)

RPTTIMING 0.047*** (0.001) 0.047*** (0.001)

LEVERAGE 0.108*** (0.000) 0.110*** (0.000)

SDCR ? 0.035*** (0.004) 0.035*** (0.004)

Year Dummy No No Yes Yes

Industry Dummy No No Yes Yes

Pseudo R2 0.026 0.026 0.096 0.097

F test 187.4 93.6 37.4 36.0

This table reports the estimation results of tobit model Eq.4.2. The error term is adjusted forheteroscedasticity (White, 1980). Refer to Table 4.2 for variable definitions. p values are reported inparentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

The positive association between control rights and dividend payouts can be

attributed to UCSs’ attempts to force the directors of listed firms to distribute more cash

andmitigate the agency cost of free cash flow.78 AlthoughUCSs have the power to nominate

directors and some of themmight beUCSs’ senior managers as well, these directors are still

agents who have the discretion and incentive to conduct self dealing behaviours.79 On the

other hand, since UCSs take responsibility for monitoring the managers of controlled public

firms,80 it provides the minority shareholders with an option of free riding, and the

dissipative taxation cost on dividends can be viewed as a part of the premium for this

option.81 As the majority of voting rights which UCSs are entitled to come from the non

78 It is also possible that this positive relationship is a kind of wealth expropriation as some institutional UCSshave tax advantages, because they receive before tax dividends while individual shareholders are subject toa flat withholding tax rate. However, the income tax on cash and stock dividends has been halted and resumedby CSRC several times during the observation window, which makes its impact inconsistent.79 The severity of agency costs depends on the separation of ownership and management, which is related tothe types of UCSs.80 If necessary, UCSs will provide aid to listed firms in financial distress.81 The remaining portion of the premium is relevant to whether UCSs are insiders. If the listed firm is closelyaffiliated with its UCS, the majority of the premium is the discretion of using a listed firm’s assets.

96

negotiable shares,82 cash dividends have become an ideal tool for UCSs to monitor the

managers and obtain an investment return when maintaining their positions as controlling

shareholders. A higher level of control rights will enhance this mechanism and facilitate

UCSs to enforce closer surveillance on managerial behaviors.

Consistent with previous literature that growth firms are reluctant to distribute cash

(Kim et al., 1998, Chang et al., 2007, Tim et al., 1999, Truong and Heaney, 2007, Faccio et

al., 2001), the significantly negative coefficients on PRICE_TO_BOOK in both probit and

tobit models advise that stronger growth opportunity undermines both the likelihood and

the magnitude of cash dividends. It is also consistent with Huang et al.’s (2011) findings that

controlling shareholders self regulate expropriation activities after considering the listed

firm’s situation and minority shareholders’ response. RPTTIMING and LEVERAGE display

negative coefficients with 1% level significance. Financial risk and earnings management

have negative influence on the propensity and magnitude of cash dividend payouts. In

contrast, the coefficients on SIZE and ROE are significantly positive, in line with previous

literature such as Gul (1999), Blau and Fuller (2008), Eije and Megginson (2008) and Zhang

(2008). The significant positive coefficient on SDCR suggests the stock dividends from a

capital reserve announcement are linked with a higher likelihood and larger magnitude of

cash distribution, which coincides with the retained earnings hypothesis that stock

dividends contain information about the growth opportunity (Grinblatt et al., 1984, Elgers

and Murray, 1985, McNichols and Dravid, 1990).

82 Some of them have been negotiable, but are still locked up in the observation window of this chapter.

97

4.4.2. Cash dividend policy and types of UCSs

Eq.4.3 is based on Eq.4.1 and incorporates the cross effect betweenUCS_Dummy and

CONTROL_RIGHTS to approach the association between types of UCSs and the likelihood of

dividend policy. Standard errors are also adjusted for heteroscedasticity (White, 1980).

Similarly, Eq.4.4, with heteroscedasticity robust standard error (White, 1980), is

developed from Eq.4.2 to pursue the association between types of UCSs and the magnitude

of cash dividend.83

83 Both Eq.4.3 and Eq.4.4 are also estimated with lagged control rights and lagged control variables asregressors, which report similar results as the model specification of Eq.4.1 and Eq.4.2. The results areavailable on request.

98

The estimation results of Eq.4.3 are reported in Table 4.7. Column (1) of Table 4.7

indicates that, in contrast to firms with Private UCSs, firms with SOELG or SAMB UCSs are

less likely to announce cash dividends. It is confirmed by column (3) which suggests the

coefficient on the cross effect between DSOECG and CONTROL_RIGHTS is positive and

significant at 1%. SOECG UCSs are deemed to have the strongest financial capacity because

they are more closely connected with the central government.84 The higher propensity for

cash dividends from these firms might be attributed to SOECG UCSs’ collaboration with the

securities regulatory agency. As most Chinese listed firms have not customarily provided

adequate cash returns, CSRC attempts to improve cash dividend distribution in the Chinese

equity market by establishing models that maintain a robust cash dividend payout history.

Column (2) of Table 4.7 indicates the coefficients on DPrivate, and its cross effects

with CONTROL_RIGHTS are positive but only marginally significant at the 10% level, which

merely supports the conjecture that Private UCSs influence the controlled public firms to

increase the likelihood of cash dividend announcements. By contrast, the coefficient on

DSOELG is negative and significant at the 5% level, as shown in column (4). SOELG

controlled firms are less likely to pay cash dividends because SOELG UCSs have the least

autonomy and firms under their control are of inferior quality (Chen et al., 2009c), which

undermines their chance to increase the propensity of cash distribution.85 Finally, firmswith

SAMB UCSs have neither advantages nor disadvantages in the likelihood of cash dividend

announcements, as shown by column (5) of Table 4.7.

84 As per Table 4.3, SOECG controlled firms have better liquidity than firms with other types of UCSs.85 Panel B of Table 4.2 indicates SOELG controlled firms have the worst profitability.

99

Table 4 7 Types of UCSs and the likelihood of cash dividends

DUMMY_CASHDIV (1) p value (2) p value (3) p value (4) p value (5) p value

Intercept 3.438*** (0.000) 3.437*** (0.000) 3.239*** (0.000) 3.214*** (0.000) 3.404*** (0.000)

DSOECG 0.118 (0.129)

DSOELG 0.162*** (0.005)

DSAMB 0.110** (0.016)

DPrivate × CONTROL_RIGHTS 0.186* (0.087)

DSOECG × CONTROL_RIGHTS 0.613*** (0.000)

DSOELG × CONTROL_RIGHTS 0.285** (0.013)

DSAMB × CONTROL_RIGHTS 0.140 (0.128)

CONTROL_RIGHTS 0.687*** (0.000) 0.601*** (0.000) 0.587*** (0.000) 0.683*** (0.000) 0.709*** (0.000)

Dummy_LowControl 0.186** (0.006) 0.172** (0.011) 0.180*** (0.008) 0.176*** (0.009) 0.174** (0.011)

SIZE 0.214*** (0.000) 0.212*** (0.000) 0.202*** (0.000) 0.202*** (0.000) 0.210*** (0.000)

PRICE_TO_BOOK 0.095*** (0.000) 0.093*** (0.000) 0.095*** (0.000) 0.093*** (0.000) 0.093*** (0.000)

FREE_CF 0.035 (0.371) 0.032 (0.407) 0.029 (0.457) 0.031 (0.429) 0.030 (0.437)

ROE 8.325*** (0.000) 8.330*** (0.000) 8.458*** (0.000) 8.384*** (0.000) 8.371*** (0.000)

RPTTIMING 0.205*** (0.001) 0.209*** (0.001) 0.205*** (0.001) 0.211*** (0.000) 0.207*** (0.001)

LEVERAGE 0.424*** (0.000) 0.429*** (0.000) 0.427*** (0.000) 0.427*** (0.000) 0.430*** (0.000)

SDCR 0.251*** (0.000) 0.254*** (0.000) 0.279*** (0.000) 0.264*** (0.000) 0.264*** (0.000)

Year Dummy Yes Yes Yes Yes Yes

Industry Dummy Yes Yes Yes Yes Yes

Pseudo R2 0.174 0.173 0.174 0.173 0.1732 945.5 934.1 933.9 936.7 931.6

This table reports the estimation results of probit model Eq.4.3. The error term is adjusted for heteroscedasticity (White, 1980). Refer to Table 4.2 for variable definitions.p values are reported in parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

100

The estimation results of Eq.4.4 are reported in Table 4.8. The association between

types of UCSs and the magnitude of the cash dividend is slightly different from the findings

in Table 4.7. Column (1) of Table 4.8 indicates firms with SAMB UCSs pay less cash dividend

than thosewith Private UCSs, while column (5) confirms these firms have lower payout ratio

than firms with other types of UCSs. As aforementioned, SAMB UCSs are effectively

government agencies with commitment beyond economic profits. The managers of SAMB

controlled firms are generally government officials, which minimizes the agency cost

between large shareholders and managers. There is no need for SAMB UCSs to mitigate the

agency cost. Instead, SAMB UCSs intend to make the managers hoard the cash balance.

Both factors lowers the magnitude of cash payout by SAMB controlled firms.

By contrast, column (2) suggests firms with Private UCSs pay more cash dividends,

different from Table 4.7 which shows little evidence that they have an outstanding

likelihood of a cash dividend. The result can be, however, attributed to the separation of

ownership and management in these firms, as Private UCSs are more accustomed to

employing professionals as the managers of controlled entities. As per Table 4.3, Private

controlled firms exhibit the best growth opportunity but the largest gap between

CONTROL_RIGHTS and CASH_RIGHTS, which implies more potential for over investment

problems and less incentive to use cash dividends to tunnel the financial resource from the

underlying listed firms. Therefore, a plausible explanation to this outcome is that Private

UCSs use cash dividends to discipline the managers to avoid the abuse of excessive cash.

101

Table 4 8 Types of UCSs and the magnitude of cash dividends

CASH_PAYOUT (1) p value (2) p value (3) p value (4) p value (5) p value

Intercept 0.340*** (0.000) 0.265** (0.028) 0.188 (0.109) 0.194* (0.098) 0.259** (0.031)

DSOECG 0.000 (0.986)

DSOELG 0.029* (0.064)

DSAMB 0.033*** (0.004)

DPrivate × CONTROL_RIGHTS 0.062** (0.019)

DSOECG × CONTROL_RIGHTS 0.061 (0.121)

DSOELG × CONTROL_RIGHTS 0.018 (0.571)

DSAMB × CONTROL_RIGHTS 0.054** (0.015)

CONTROL_RIGHTS 0.232*** (0.000) 0.209*** (0.000) 0.211*** (0.000) 0.219*** (0.000) 0.250*** (0.000)

Dummy_LowControl 0.052*** (0.006) 0.048*** (0.010) 0.051*** (0.007) 0.050*** (0.007) 0.049*** (0.009)

SIZE 0.028*** (0.000) 0.027*** (0.000) 0.024*** (0.000) 0.024*** (0.000) 0.027*** (0.000)

PRICE_TO_BOOK 0.012*** (0.000) 0.012*** (0.000) 0.012*** (0.000) 0.012*** (0.000) 0.012*** (0.000)

FREE_CF 0.019* (0.068) 0.019* (0.078) 0.018* (0.096) 0.017* (0.099) 0.018* (0.083)

ROE 0.823*** (0.000) 0.825*** (0.000) 0.848*** (0.000) 0.842*** (0.000) 0.837*** (0.000)

RPTTIMING 0.047*** (0.001) 0.047*** (0.001) 0.047*** (0.001) 0.047*** (0.001) 0.047*** (0.001)

LEVERAGE 0.108*** (0.000) 0.109*** (0.000) 0.109*** (0.000) 0.109*** (0.000) 0.109*** (0.000)

SDCR 0.028** (0.020) 0.029** (0.020) 0.035*** (0.003) 0.034*** (0.004) 0.031*** (0.009)

Year Dummy Yes Yes Yes Yes YesIndustry Dummy Yes Yes Yes Yes Yes

Pseudo R2 0.099 0.098 0.098 0.097 0.098

F test 31.7 34.4 34.6 34.4 34.4

This table reports the estimation results of tobit model Eq.4.4. The error term is adjusted for heteroscedasticity (White, 1980). Refer to Table 4.2 for variable definitions.p values are reported in parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

102

With regard to firmswith SOECG and SOELGUCSs, the coefficients on the cross effects

between UCS_Dummy and CONTROL_RIGHTS are both insignificant, as per column (3) and

(4), indicating the payout ratio of these firms are indifferent from their peers. For SOECG

controlled firms, the result underpins the explanation that their cash dividend

announcements are more likely to be window dressing activities for to establish a model

firm with a continuous cash return to investors, because they have a higher propensity for

cash dividend announcements but no advantage in the cash payout ratio.

4.4.3. Robustness check

This section describes further analysis which was undertaken to evaluate the

robustness of the reported results. First, the key independent variable, CONTROL_RIGHTS,

is replaced by CONTROL_CASH_GAP, which indicates the difference between

CONTROL_RIGHTS and CASH_RIGHTS.86 This chapter differs from Cheng et al (2009)

because it uses control rights as the proxy of influence from the large shareholders, while

the latter utilize the cash rights. To justify that control rights is a better alternative to cash

rights, it is necessary to showwhether the difference between control rights and cash rights

makes positive contribution to the likelihood and magnitude dividend announcement.

Table 4.9 supplies a detailed investigation into CONTROL_CASH_GAP for the whole

sample and each UCS category. Although only 37.5 per cent, or 2,563 out of 6,836, of the

whole sample has a positive CONTROL_CASH_GAP, the difference, 14.30 per cent,87 is

86 Some observations report a positive CONTROL_CASH_GAP but less than 1%. They are classified as zeroCONTROL_CASH_GAP. All non dummy variables of the positive CONTROL_CASH_GAP sample are winsorizedat the top and bottom 1%.87 Even observations with zero CONTROL_CASH_GAP are included in the whole sample or UCSs sub group,the mean of CONTROL_CASH_GAP is still significantly different from zero.

103

significantly different from zero at 1% level. The result suggests that a complex pyramid

structure has emerged in the Chinese equity market.

Table 4 9 Difference between control rights and cash rights

N CONTROL_RIGHTS CASH_RIGHTS CONTROL_CASH_GAP t stat

Whole sample 2,563 0.3790 0.2360 0.1430* 94.93(0.0029) (0.0028) (0.0015)

Private 1,466 0.3561 0.2167 0.1394* 66.76

(0.0038) (0.0038) (0.0020)

SOECG 160 0.3730 0.2354 0.1376* 23.29

(0.0123) (0.0113) (0.0059)

SOELG 230 0.3839 0.2494 0.1345* 28.09

(0.0100) (0.0090) (0.0021)

SAMB 707 0.4267 0.2724 0.1543* 60.00

(0.0048) (0.0047) (0.0026)

This table reports the means of CONTROL_RIGHTS, CASH_RIGHTS and CONTROL_CASH_GAP. Refer to Table4.2 for variable definitions. Standard errors are reported in parentheses. * denotes significance at 1% level.

In line with expectations, more than two thirds of Private controlled firms, 1,466 out

of 2,146, have positive CONTROL_CASH_GAP, while the majority of state owned UCSs hold

control rights identical to cash rights. Private UCSs often have weaker financial capacities

than state ownedUCSs, which leads them to pursue pyramid structures with limited capital.

Consequently, they are more likely to be exposed to the traditional agency principal

problem as themultiple level shareholding structuremay well exacerbate the change in the

managers’ self dealing behaviours. On the other hand, although SAMB controlled firms

exhibit the largest mean CONTROL_CASH_GAP, the majority of firms controlled by state

owned UCSs, approximately 75 per cent, report a zero CONTROL_CASH_GAP, which

suggests state owned UCSs are relatively conservative in corporate controlled markets.88

Table 4.10 and Table 4.11 repeat the regressions in Sections 4.4.1 and 4.4.2 with the

key independent variable set as CONTROL_CASH_GAP. Table 4.9 indicates that

88 When zero CONTROL_CASH_GAP are included, Private controlled firms report the largest meanCONTROL_CASH_GAP at 9.55%, while the mean CONTROL_CASH_GAPs of groups with various state ownedUCSs are between 3% and 5%.

104

CONTROL_CASH_GAP has strong power to explain the variation in both propensity and

magnitude of cash dividend announcements. Further, Table 4.11 affirms the results of Table

4.8. Private controlled firms have no advantage in payout propensity, but lead in payout

magnitude (with marginal significance), because they are subject to greater agency costs of

free cash flow. SOECG controlled firms lead in the payout propensity but have no superior

advantage in payoutmagnitude, which underpins the ‘window dressing’ hypothesis. Finally,

SAMB controlled firms lag behind other groups in both propensity and magnitude of cash

dividend that confirms the severity of the agency problem between large and small

shareholders and the resultant expropriation.

Table 4 10 Robustness check: dividend policy and the gap between control rights and cash rights

Panel A Likelihood of dividend payout Panel B Magnitude of dividend payout(1) p value (2) p value (1) p value (2) p value

Intercept 4.553*** (0.000) 4.458*** (0.000) 0.508** (0.012) 0.480** (0.017)CONTROL_CASH_GAP 1.304*** (0.000) 1.012*** (0.008) 0.517*** (0.000) 0.413*** (0.000)

Dummy_LowControl 0.250*** (0.004) 0.095*** (0.000)

SIZE 0.258*** (0.000) 0.258*** (0.000) 0.038*** (0.000) 0.038*** (0.000)

PRICE_TO_BOOK 0.081*** (0.000) 0.084*** (0.000) 0.007 (0.115) 0.008* (0.062)

FREE_CF 0.110* (0.060) 0.110* (0.058) 0.051*** (0.003) 0.051*** (0.003)

ROE 7.436*** (0.000) 7.420*** (0.000) 0.667*** (0.000) 0.674*** (0.000)

RPTTIMING 0.179* (0.057) 0.183* (0.051) 0.047** (0.039) 0.048** (0.035)

LEVERAGE 0.352*** (0.000) 0.354*** (0.000) 0.088*** (0.000) 0.089*** (0.000)

SDCR 0.289*** (0.000) 0.286*** (0.000) 0.045** (0.017) 0.044** (0.020)

Year Dummy Yes Yes Yes Yes

Industry Dummy Yes Yes Yes YesN 2,563 2,563 2,563 2,563

Pseudo R2 0.153 0.169 0.089 0.094

X2 337.3 344.9

F test 13.2 13.9

This table reports the estimation results of probit model Eq.4.1 and tobit model Eq.4.2 withCONTROL_CASH_GAP as the key independent variable, in Panel A and Panel B, respectively. The error termsare adjusted for heteroscedasticity (White, 1980). Refer to Table 4.2 for variable definition. p values arereported in parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

105

Table 4 11 Robustness check: dividend policy, types of UCSs and the gap between control rightsand cash rights

Panel A Likelihood of cash dividendDUMMY_CASHDIV (1) DPrivate p value (2) DSOECG p value (3) DSOELG p value (4) DSAMB p value

Intercept 4.729*** (0.000) 4.506*** (0.000) 4.550*** (0.000) 4.937*** (0.000)

UCS_Dummy× CONTROL_CASH_GAP 0.416 (0.259) 1.801** (0.035) 0.137 (0.836) 0.905** (0.022)

CONTROL_CASH_GAP 1.048** (0.014) 1.244*** (0.001) 1.311*** (0.000) 1.563*** (0.000)

SIZE 0.267*** (0.000) 0.255*** (0.000) 0.258*** (0.000) 0.275*** (0.000)

PRICE_TO_BOOK 0.081*** (0.000) 0.083*** (0.000) 0.081*** (0.000) 0.082*** (0.000)

FREE_CF 0.114** (0.050) 0.105* (0.070) 0.110* (0.099) 0.117** (0.044)

ROE 7.405*** (0.000) 7.468*** (0.000) 7.438*** (0.000) 7.371*** (0.000)

RPTTIMING 0.178* (0.058) 0.178* (0.058) 0.180* (0.056) 0.170* (0.073)

LEVERAGE 0.351*** (0.000) 0.355*** (0.000) 0.351*** (0.000) 0.355*** (0.000)

SDCR 0.275*** (0.001) 0.298*** (0.000) 0.288*** (0.000) 0.268*** (0.001)

Year Dummy Yes Yes Yes Yes

Industry Dummy Yes Yes Yes YesPseudo R2 0.153 0.154 0.153 0.1542 337.4 338.9 339.5 334.8

Panel B Magnitude of cash dividendCASH_PAYOUT (1) DPrivate p value (2) DSOECG p value (3) DSOELG p value (4) DSAMB p value

Intercept 0.578*** (0.005) 0.506** (0.012) 0.507** (0.012) 0.583*** (0.004)

UCS_Dummy× CONTROL_CASH_GAP 0.160* (0.086) 0.076 (0.593) 0.087 (0.572) 0.176* (0.079)

CONTROL_CASH_GAP 0.420*** (0.000) 0.513*** (0.000) 0.522*** (0.000) 0.566*** (0.000)

SIZE 0.042*** (0.000) 0.038*** (0.000) 0.038*** (0.000) 0.042*** (0.000)

PRICE_TO_BOOK 0.007 (0.117) 0.007 (0.113) 0.007 (0.119) 0.007 (0.105)

FREE_CF 0.053*** (0.002) 0.051*** (0. 003) 0.052*** (0.003) 0.053*** (0.002)

ROE 0.659*** (0.000) 0.667*** (0.000) 0.668*** (0.000) 0.656*** (0.000)

RPTTIMING 0.046** (0.042) 0.047** (0.039) 0.048** (0.036) 0.045* (0.051)

LEVERAGE 0.088*** (0.000) 0.088*** (0.000) 0.088*** (0.000) 0.089*** (0.000)

SDCR 0.039** (0.039) 0.045** (0.017) 0.044** (0.019) 0.040** (0.034)

Year Dummy Yes Yes Yes Yes

Industry Dummy Yes Yes Yes Yes

Pseudo R2 0.090 0.089 0.089 0.090

F test 12.6 12.7 12.5 12.6

This table reports the estimation results of probit model Eq.4.3 and tobit model Eq.4.4 withCONTROL_CASH_GAP as the key independent variable, in Panel A and Panel B, respectively. The error term isadjusted for heteroscedasticity (White, 1980). Refer to Table 4.2 for variable definitions. p values are reportedin parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

Second, regression models Eq.4.1 and Eq.4.2 are estimated with various

methodologies, including ordinary least squares (OLS), Fama Macbeth two stage procedure

(Fama and MacBeth, 1973), the random effect model and the fixed effect model. These

results are generally consistent with those of the original analysis (refer to Appendix 4.1).

Third, high correlation coefficients among ROE, PRICE_TO_BOOK, LEVERAGE and FREE_CF

106

may result in biases in the regression results. Variance inflated factors (VIF) and tolerance

(1/VIF) are reported in Appendix 4.2, which suggest multicollinearity is insignificant.

Last but not least, the selection bias might be a problem as the higher level of control

rights and certain types of UCSs probably stems from investors’ preference to firms with

higher dividend payout. Heckman selectionmodel is introduced to investigate this potential

selection bias, as shown in Appendix 4.3. Step 1 assumes DUMMY_CASHDIV to be

determined by all control variables, while step 2 regress CONTROL_RIGHTS (UCS_Dummy)

on lagged SIZE, lagged LEVERAGE and lagged PRICE_TO_BOOK. Most Mill’s Lambdas in both

Panel A and B are not statistically significant. The only exceptional case is SOECG UCS, In

general, the selection bias is not a significant issue in this research.

4.5. Summary

This chapter examines the interaction between the ultimate controlling shareholders

(UCSs) and the dividend policy, with a unique data set from the Chinese market between

2003 and 2010. It examines whether the cash dividend payout in the Chinese market is an

expropriation of minority shareholders’ wealth, or a tool to deal with the agency cost of

free cash flow. UCSs’ control rights are proposed as an alternative measurement of

corporate ownership in the Chinese market as non negotiable shares are phased out. The

types of UCSs are introduced to reflect their backgrounds and nature.

The empirical evidence indicates that UCSs’ control rights are positively associated

with both the propensity and the magnitude of cash dividends in the Chinese stock market.

The results are assigned to UCSs’ efforts to reduce the agency costs of free cash flow by

coercing the management to pay cash dividends, as well as realize the investment return

107

when the liquidation of the shareholding is restricted by regulation. Consistent with

previous studies (Tim et al., 1999, Fama and French, 2002, Truong and Heaney, 2007), lower

growth opportunity, lower financial leverage and higher profitability appear to encourage

the firms to announce cash dividends and make more cash distributions.

Different types of UCSs lead to divergent influences on the propensity and magnitude

of the cash dividend policy. Contrary to past studies about non state owned firms’

inclination to hoard cash, this chapter shows that Private controlled firms lead in the

magnitude of cash distribution, which is ascribed to their attempts to reduce the agency

cost of free cash flow, as the separation of ownership and management is more popular

among these firms. Of the firms with various types of state owned UCSs, SOECG controlled

firms have some advantages leading to the likelihood of cash payouts, but their cash payout

magnitude is similar to firms with other types of state owned UCSs. The phenomenon is

attributed to the manipulation of their continuous cash dividend history as a result of their

closer relationship with the CSRC. In contrast, external investors in SAMB controlled firms

are disadvantaged because such firms pay fewer cash dividends, which indicates the agency

cost between large and small shareholders as the managers of these firms are closely

affiliated with the ultimate controlling shareholders. Generally, this chapter suggests both

types of agency costs exist in the Chinese market and their relative significance depends on

the ultimate controlling shareholders.

108

Chapter 5 Dividend Announcement Effect and Ultimate Controlling

Shareholders in the Chinese Stock Market

5.1. Introduction

This chapter investigates stock abnormal returns following simultaneous dividends

and earnings announcements,89 and the availability of information on ultimate controlling

shareholders, in the Chinese equity market. Theoretically, the valuation of a firm is not

affected by the dividend policy in a frictionless capital market, because future capital gain

from the growth rate will be offset by a current dividend (Miller and Modigliani, 1961).

However, with market inefficiencies, such as transaction costs and income tax, unexpected

changes in cash dividends may be taken as a signal reflecting the managers’ opinions on

future operating performance, and may affect the market capitalization due to the

existence of income tax and asymmetric information between the external shareholders

and the managers (Lintner, 1956, Pettit, 1972, Charest, 1978a, Bhattacharya, 1979, Divecha

and Morse, 1983, Miller and Rock, 1985, Ghosh and Woolridge, 1988). A number of

empirical studies have been conducted on the existence and significance of the cash

dividend announcement effect in both developed markets (Aharony and Swary, 1980, Bar

Yosef and Huffman, 1986, Williams, 1988, John and Lang, 1991, Noe and Rebello, 1996, Eije

and Megginson, 2008) and emerging markets (Twu, 2010, Hussin et al., 2010, Yeh et al.,

2011, Altiok Yilmaz and Akben Selcuk, 2010). Most of the studies support the existence of

a significant dividend announcement effect which is attributed to market imperfection.

89 As per the regulations of the Chinese Security Regulatory Commission (CSRC), Chinese listed firms have topublish annual financial statements and dividend decisions on the same day.

109

Intuitively, unless the operating performance changes significantly, managers prefer to

adopt a sticky dividend policy and maintain the flexibility of a payout policy with other

distribution methods, such as share repurchase, to avoid the impact on investors’

expectations and the firm’s market valuation. Further, compared with the synchronous

earnings announcement, the dividend announcement, especially a change in the cash

dividend, has more influence on the stock abnormal return (Healy and Palepu, 1988,

Woolridge, 1982, Aharony et al., 1988, Lee, 1995, Docking and Koch, 2005, Skinner, 2008).90

Being one of the largest transitional economies, China has been in the spotlight and

aroused researchers’ interest in recent years. As a typical government sponsored equity

market, the wealth effect caused by partial share issuance privatization (SIP) and

significantly positive IPO underpricing (Zhou and Zhou, 2010) results in negotiable

shareholders’ preference for capital gain. In contrast, cash dividend yields are relatively

trivial and negligible.91 Previous studies on investors’ reactions to dividend announcements

in the Chinese market report mixed results.92 Chen et al. (2009a) document that the stock

abnormal return is positively associated with both an increase and a decrease in cash

dividends, which implies investors welcome cash dividends and are indifferent to a change

in dividend.93 Other studies, however, conclude that a cash dividend announcement makes

little contribution to the stock abnormal return (Chen et al., 2002, Cheng et al., 2009,

90 Some other literature, such as Grullon et al. (2002), argues that the dividend announcement effect can beattributed to the updated systematic risk level, as firms approaching a mature stage are likely to distributemore cash due to fewer investment opportunities.91 Another fact is that a cash dividend is subject to income tax, as yet, while capital gains tax has not beenimplemented in China.92 Trading activity has been dominated by minority shareholders until recently due to the previous share splitstructure which separated the negotiable shares from non negotiable shares. In general, negotiable sharesare held by individual investors and non negotiable shares are owned by institutional investors, which is alegacy of the partial share issuance privatization.93 Chen et al. (2009a) do not incorporate the influence of synchronised earnings announcements.

110

Anderson et al., 2011). Therefore, the first contribution of this chapter is to supply updated

empirical evidence on whether Chineseminority shareholders have become sensitive to the

cash dividend announcements, because resilient announcement effects indicate that the

continuous efforts of the China Securities Regulatory Commission (CSRC) to encourage cash

dividends payout94 have achieved the desired outcomes, and Chineseminority shareholders

regard cash dividends as an important form of investment return.

Meanwhile, although stock dividends do not provide any direct economic benefit to

investors, some studies indicate that they contain private information about the managers’

confidence in future performance.95 Some previous studies report that Chinese investors

prefer stock dividends to cash dividends (Yi et al., 2007, Cheng et al., 2009, Wei and Xiao,

2009). It is noticed that these studies use the term ‘stock dividends’ but do not specify the

types of stock dividends. In fact, there are two kinds of stock dividends in the Chinese

market, stock dividends from retained earnings (SDRE) and stock dividends from capital

reserves (SDCR), which are similar to bonus shares and stock splits in developed market,

respectively. Therefore, the second contribution of this chapter is to differentiate the SDRE

from SDCR and investigate their impacts on the stock abnormal return separately.

Aside from the information content and signalling hypothesis, cash dividends are also

affiliated with the conflicts of interest between stakeholders within the corporation. The

asymmetric information and agency cost of free cash flowmaymake shareholders push the

94 The CSRC has announced several measures to improve the cash dividend magnitude in recent years. Forexample, listed firms have to disclose their investment plans to justify their zero dividend announcements;listed firms need to maintain a robust dividend payout history before they apply for further equity issuance.95 According to the so called ‘retained earnings hypothesis’ based on the accounting principles in the US,because a small stock dividend, no more than 25%, requires the firm to deduct the retained earnings accountby the market value of the stock dividend, the announcement of such a stock dividend is deemed todemonstrate managers’ confidence to restore the retained earnings balance with future profits (Elgers andMurray, 1985, McNichols and Dravid, 1990).

111

management to distribute the excessive cash balance and restrict potential self dealing

behaviours (Jensen, 1986, Miller and Rock, 1985). This argument is supported by some

empirical evidence (Lang and Litzenberger, 1989, Graham et al., 2006, Cheng et al., 2007,

John and Lang, 1991, Banerjee et al., 2007). Whereas, in markets other than Anglo

American countries where the ownership of shareholdings in firms is more concentrated,

the agency problem between large and small shareholders is more crucial (Claessens et al.,

2000, Gugler and Yurtoglu, 2003, Attig, 2007, Renneboog and Trojanowski, 2007, Truong

and Heaney, 2007). Cash dividends can also be used to deal with this type of agency

problem, as higher payouts restrict large shareholders’ discretion to abuse the cash holdings

and, thus, mitigate the propensity of wealth expropriation onminority shareholders (Faccio

et al., 2001). Higher cash payouts, however, can also be a form of wealth expropriation,

especially when the large shareholders are less able to liquidate their shareholdings

because of regulatory issues (How et al., 2008). In the meantime, some empirical studies in

China focus on the relationship between the dividend payout practice and the ownership

structure because of the large block of non negotiable shares and the existence of

direct/ultimate controlling shareholders (Gul, 1999, Cheng et al., 2009, Wei and Xiao, 2009,

Huang et al., 2011). Chen et al. (2009c) report that some types of UCSs are associated with

influencing the operating performance because of the divergence in their backgrounds and

natures. Although a strong UCS may exacerbate the tunnelling activities, such as transfer

pricing and inter company loan guarantees, they may well provide high quality corporate

governance as well as necessary support in cases of business or financial distress, which is

also known as free rider opportunity. The disclosure of UCSs and their voting rights may

influence minority shareholders’ trading activity, as they are caught in a dilemma between

112

potential wealth expropriation concerns and free rider opportunity. Hence, the last

contribution of this chapter is to explore investors’ reactions toward the disclosure of UCS

information, as well as the unexpected cash dividend announcements from firms controlled

by different types of UCSs.

Based on data sets between 1999 and 2010, the regression results suggest that stock

dividend payers exhibit significant positive abnormal returns within the [ 10, +10] window

around the announcement day. Specifically, stock dividends from capital reserves (SDCR)

are preferred by investors, while stock dividends from retained earnings (SDRE) only make

a marginal contribution to the announcement effect. The difference can be explained by

their divergent backgrounds and tax treatments. Meanwhile, an unexpected cash dividend

makes a marginal contribution to the stock abnormal return around announcement day. In

contrast, investors respond negatively to earnings shock, which can be attributed to the

popular earnings management in the Chinese market. These findings are consistent with

the literature that finds Chinese negotiable shareholders prefer stock dividends to cash

dividends, but are inconsistent with the literature that earnings announcements have more

influence on stock abnormal returns than the nearly simultaneous cash dividend

announcements (Chen et al., 2009a, Cheng et al., 2009, Truong, 2011). The relationship

between dividend announcement and stock abnormal return is robust to controls for firm

specific factors, such as firm size, idiosyncratic risk and financial report timing effect. Lastly,

consistent with the findings of Chen et al. (2009c), that types of ultimate controlling

shareholders (UCSs) influence the operating performance of Chinese listed firms, investors

react discriminatorily to cash dividend shocks from firms with different types of UCSs, but

are uninterested in the UCSs’ ownership which is measured by their voting rights. This

113

phenomenon can be explained by the divergent features ofUCSs andminority shareholders

adaption to UCSs’ wealth expropriation.

The rest of this chapter is organized as follows. Section 5.2 presents the development

of hypotheses. Section 5.3 explains the data and descriptive statistics. Empirical results are

reported in Section 5.4. Section 5.5 summarizes the chapter.

5.2. Hypotheses development

The stock dividend puzzle in the Chinese market has been examined by many studies,

which report mixed results. Chen et al.(2002) document that stock dividend

announcements work as a catalyst to earnings announcement, in that, they enhance

(dampen) stock returns when earnings news is positive (negative). Su (2005) finds a

significant decline in stock abnormal return after the stock dividend announcement because

most of these firms havemore government control and fewer institutional investors. Cheng

et al. (2009) document that a positive change in stock dividend is preferred by negotiable

shareholders and leads to stronger stock abnormal return around the announcement day.

Wei and Xiao (2009) also point out that negotiable shareholders with negotiable shares

prefer stock dividends to cash dividends. Anderson et al. (2011) report a positive association

between stock dividends and stock abnormal returns due to the low profitability and/or

poor cash availability of the underlying firms.96

The above studies did not specify whether stock dividends are defined as either stock

dividends from capital reserves or stock dividends from retained earnings, or both. In fact,

96 The negotiable shareholders’ preference for stock dividends is partly due to the significant IPO underpricingduring the partial SIP. There was a commonmisunderstanding that more shares would generate more wealth,but negotiable shareholders, especially individual investors, ignored the fact that the common shares afterstock splits and/or bonus shares have been diluted in value.

114

SDCR and SDRE are different from each other in many aspects, such as accounting

treatments and taxation issues. As per the Chinese Corporation Law, capital reserves can

only be transferred to paid in capital accounts, while retained earnings have more

functions, such as making up losses carried forward, increasing paid in capital and paying

cash dividends. In general, SDCR is regarded as an accounting practice to substantiate the

equity capital, while SDRE is treated more like a profit distribution, although no cash is

involved. Consequently, as per the Chinese Taxation Code, SDCR is exempt from income tax,

but SDRE is subject to income tax as it is part of profit distribution and creates tax liability

on investors.97 Intuitively, neither stock splits nor bonus shares change investors’ wealth in

developed markets, but Chinese investors will be disadvantaged if an SDRE is announced,

because their wealth will be reduced by the amount of the corresponding income tax.

Moreover, stock dividends from retained earningsmay suggest inferior earnings quality and

poor liquidity stemming from the popular earnings management practice in the Chinese

market (Haw et al., 2005, Chen et al., 2009d, Chen et al., 2011, Anderson et al., 2011), while

SDCR is irrelevant to profitability and more suitable as a tool for management and the UCSs

behind the management98 to share the growth opportunity with the investors. Therefore,

the first hypothesis is based on the aforementioned elements and is outlined as below.

Hypothesis 5.1 The announcement effect of stock dividends is significantly positive; stock

dividends from capital reserves lead to a stronger announcement effect than stock dividends

from retained earnings.

97 This practice is different from developed markets where stock dividends are income tax free because thecapital gain is unrealized.98 The controlling shareholders have sufficient voting rights to appoint the managers and exercise control viathe management team. The managers in Chinese listed firms are not necessarily professional entrepreneurs,but more likely to be someone affiliated with large/controlling shareholders.

115

Empirical studies indicate that stock prices often react resiliently to the unexpected

changes in cash dividends, and the impact is magnified by the market momentum (Lee,

1995, Woolridge, 1982, Ghosh and Woolridge, 1988, Ghosh and Woolridge, 1991, Docking

and Koch, 2005, Yeh et al., 2011). But the existing literature on the cash dividend

announcement effect in the Chinese market is different from the general consensus. For

example, Chen et al. (2002) conclude few connections exist between cash dividends and

stock abnormal return, in line with the aforementioned negotiable shareholders’

preference for stock dividends because the cash dividend is overwhelmed by the capital

gain. Cheng et al. (2009)measure the information content of cash the dividend and earnings

announcement with expectation models, and their results indicate that an earnings shock,

instead of cash dividend shock, leads to significant announcement effects. Anderson et al.

(2011) document that cash dividend plays a supplementary role in the dividend payout

policy of Chinese listed firms.99 But Yi et al. (2007) argue that the measurement of return

should be differentiated between negotiable shareholders and non negotiable

shareholders, as the latter has less flexibility to realize the capital gain and achieves more

return from a cash dividend announcement.

Different from the above studies, this chapter posits a positive relation between cash

dividend shock and announcement effect. To some extent, Chinese minority shareholders

have adapted to their disadvantaged positions in the stock market in which the controlling

shareholders have dominant power in the underlying listed firm. The dividend shock will be

regarded as an additional benefit by the listed companies, which is ‘better than nothing’.

99 The results in Anderson et al. (2011) may be influenced by the tax liability of stock dividend from retainedearnings, which is offset by a simultaneous small amount of cash dividend.

116

Further, from the minority shareholders’ viewpoint, unexpected cash dividends are much

better than having the cash balance abused by the managers and controlling shareholders,

which is akin to ‘the bird in the hand’ argument.100

Meanwhile, the studies on the cash dividend announcement effect often mention

encountering noise from synchronous earnings announcements. Empirical evidence from

the US suggests that earnings announcements following the dividend announcement only

lead to marginal stock abnormal return because investors have reacted to the information

content embedded in the dividend announcement (Healy and Palepu, 1988). This issue of

simultaneous announcements is more significant in the Chinese market as it is mandatory

for listed firms to declare the annual financial statements and dividend decision on the same

day, except in some special cases. Some empirical research suggests the unexpected

earnings have a positive influence on the stock abnormal return (Chen et al., 2002, Cheng

et al., 2009). This chapter conjectures, however, that the earnings shock makes little

contribution to stock abnormal return because of the poor corporate governance

environment and popular earnings management practices in the Chinese market (Shen and

Chih, 2007, Bhaumik and Gregoriou, 2010, Haw et al., 2005, Chen et al., 2011).

Hypothesis 5.2 Unexpected cash dividends, instead of unexpected earnings, positively

influence the abnormal stock returns during the announcement period.

The agency cost between large and small shareholders in an emerging market has

beenmore significant than the traditional agency cost betweenmanagers and shareholders

(Burkart et al., 1997, Faccio et al., 2001). The unique ownership structure of most Chinese

100 In Anglo American markets, ‘the bird in the hand’ argument usually means shareholders actively pushmanagers to distribute cash but, in the Chinese market, investors passively wait for the cash dividends.

117

listed firms and the existence of ultimate controlling shareholders are the legacy of partial

SIP, thatmagnifies the possibility of self dealing activities at the cost of the externalminority

shareholders (Cheung et al., 2009b). The situation is further exacerbated by the weak legal

protection for minority investors in China (Mitton, 2004, Wang et al., 2008). Since 2003,

Chinese listed firms are required to publish information about ultimate controlling

shareholders.101 It is posited that these disclosures, along with the earnings and dividend

announcements, may well be associated with stock abnormal return. Intuitively, UCSs with

more control rights will raise minority shareholders’ concerns about wealth expropriation,

which results in negative stock abnormal return.

Besides the control rights, the types of UCSs are also introduced because Chinese

minority shareholders often deem the strength of UCSs to be an important factor for the

prosperity of the underlying listed company. A strong UCS may imply potential support in

case of financial distress, whereas a weak UCS may exacerbate the expropriation. This

chapter adopts the categorization defined by Chen et al. (2009c) State owned Asset

Management Bureau (SAMB), State Owned Enterprises (SOE) affiliated with central

government (SOECG), SOEs affiliated with local government (SOELG), and Private investors

(Private). These UCSs have different backgrounds, natures and financial capacity.102 Chen

et al. (2009c) document divergent operating performance among firms controlled by

various types of UCSs.103 Here it is conjectured that unexpected cash dividends from SAMB

or Private controlled firms are positively correlated with stock abnormal returns. Private

101 The information includes their background, affiliation with different levels of government, directshareholdings (cash rights) and, direct and indirect shareholdings (control rights).102 For details of the differences among various types of UCSs, refer to Chapter 4.103 In the previous chapter, it is reported that listed firms with Private UCSs are inclined to pay more cashdividends, but listed firms with SAMB or SOELG UCSs are likely to pay a lower cash distribution.

118

UCSs are often considered to bemore efficient than state ownedUCSs. These investorsmay

force the managers to pay out more cash to mitigate the agency cost of free cash flow,

which will be taken as a positive signal by minority shareholders. Meanwhile, SAMB UCSs

are effectively government agencies and deemed to be the least efficient of those types of

UCSs. Therefore, cash dividend shocks from SAMB controlled firms are welcomed by

external investors as they receive additional return which leads to lower cash balances held

by these firms.104

Hypothesis 5.3 Level of control rights has negative influence on dividend announcement

effects; unexpected cash dividends from firms with SAMB or Private UCSs lead to positive

stock abnormal return.

5.3. Data and descriptive statistics

5.3.1. Sample construction

The data are abstracted from the Chinese Security Market and Accounting Research

(CSMAR) database and the observation period is from 1999 to 2010. The criteria for the

data filtering are: (i) samples will include all firms except those belonging to the financial

industry; (ii) firms with missing financial information will not be included in the sample; (iii)

companies which report non positive equity are excluded from the sample; (iv) firms with

a total payouts ratio105 larger than one are omitted; (v) observations are excluded if there

104 As per Chen et al. (2009b), SAMBs are government agencies which operate state owned assets. SOECGsare state owned enterprises directly controlled by central government. The difference between SOELGs andSOECGs is that SOELGs were created to manage the spin offs from the state owned enterprises previouslyowned by central government. Different from those three state owned UCSs, Private UCSs include bothprivate investment institutions and individual investors.105 The total payout ratio is defined as the sum of cash dividend and stock dividend from retained earningsscaled by net after tax profit.

119

is an equity financing announcement, such as seasoned equity offering and rights issues,

within the window between 30 days before and 10 days after the earnings/dividend

announcement. After applying the above filtering criteria, there are 9,405 observations

from 1,203 firms across the observation window between 1999 and 2010, of which 4,806

observations, about 51 per cent, make cash only dividend announcements. Besides the

cash only payers, another 1,265 observations make cash dividends payout with

simultaneous stock dividends. The numbers of observations, grouped by stock and cash

payout decisions, are reported in Table 5.1.106

Table 5 1 Distribution of observations by the combination of stock and cash dividends

Year Cash onlypayer

Cash withSDRE

Cash withSDCR

Cash with bothSDRE and SDCR

Non cashSDRE

Non cashSDCR

Non cash SDREand SDCR

Nonpayer

Total 4,806 244 685 336 50 257 47 2,980

This table provides the annual numbers of observations grouped by the combination of cash and stockdividend decisions. SDRE refers to stock dividends from retained earnings. SDCR refers to stock dividends fromcapital reserve.

5.3.2. Dependent variable

The dependent variable is the cumulative abnormal return (CAR) around the dividend

and financial statement announcement day. The daily return on individual stock is regressed

on the return on market107 to estimate the intercept and slope of the capital

asset pricing model, over the 200 day window before the announcement day.108

106 The empirical analysis will focus on the cash only payer group, and cash payer group, which includes cashonly payers, cash with SDRE, cash with SDCR and cash with both SDRE and SDCR.107 SSE A Share Index and SZSE A Share Index are set as market index for stocks listed in two exchanges,respectively.108 Fama French 3 factor model was considered but not used because of the segmentation of negotiable andnon negotiable shares. Although the non negotiable shares are gradually phased out during the observationwindow, many shares held by large shareholders are actually locked up, which inflate the market value offirms with large block of locked up negotiable share. HML and SMB factors in FF 3 factor model relies on theportfolio construction based on the market equity value. It might not be proper to use the price of fullynegotiable shares to proxy the locked up common shares.

120

Daily abnormal return is defined as follows:

Four CARwindows are generated, including one pre event window [ 10, 2], and three

event windows, [ 1, +1], [ 3, +3], and [ 10, +10], defined as short, medium and long event

windows, respectively.109 The pre event window is to test whether the market reacts to

dividend shock and/or earnings shock in anticipation of the announcement, while the three

event windows are to test the market reaction to dividend shocks, earnings shocks, and

stock dividend announcements.

5.3.3. Independent variables

5.3.3.1 Construction of expectation models for event study

Expectation models are developed to generate earnings and dividend forecasts.110

Past literature on developed markets suggests annual earnings follow a stochastic process

with a growth component (Watts and Leftwich, 1977, Bamber, 1987). Therefore, an

expectation proxy to model market anticipated earnings and cash dividends is generated

based on the levels of prior years and an industry111 adjustment factor, which is the growth

109 One examiner points out that themarketmodel to estimate the intercept and slopemay cause overlappingproblem, because it is based on observation window [ 200, 1]. As a response, the intercept and slope are reestimated using window [ 210, 11]. When the methodologies in the following sections are applied on CARsand other relevant variables generated from the updated intercept and slope, the results are indifferent fromthose based on window [ 200, 1].110 Although the CSMAR database collects the earnings and dividend estimates from local securities analysts,these forecasts were not systematic until recently and are insufficient for this research.111 CSMAR has two industry categorizations, Code A and Code B. Cheng et al. (2009) use Industry Code B, whichhas a narrower definition of individual industry. This research uses Industry Code A, which only contains sixcategories: financial, utilities, properties, conglomerates, industrials and commerce. The use of Industry CodeA is to avoid the overestimation of unexpected earnings and cash dividend because of narrow industrydefinition.

121

rate of mean earnings (dividends) for the firms in industry j112 of current year (year t) from

the previous year (year t 1). The expectation models for earnings and dividends are:

where is the expected earnings per share of firm i in year t; is the

expected cash dividend per share of firm i in year t; IFEjt is the industry adjustment factor

for earnings per share (the percentage change of the mean earnings per share for all firms

of industry j between year t and year t 1);113 IFDjt is the industry adjustment factor for cash

dividend per share (the percentage change of the mean dividend per share for all firms of

industry j between year t and year t 1);114 EPSit is the earnings per share for firm i in year t;

DPSit is the cash dividend per share of firm i in year t; t represents the fiscal year; j represents

the industry category.

5.3.3.2 Unexpected cash dividend, unexpected earnings, and UCSs information

The expectations models specified above are used to estimate the unexpected

earnings, ‘earnings shock’, UNEXPECTED_EPSi,t, and the unexpected cash dividend,

‘dividend shock’, UNEXPECTED_DPSi,t. UNEXPECTED_EPSi,t is the deviation of reported

earnings per share, EPSit, from expected earnings per share, , in percentage, while

the unexpected cash dividend per share,UNEXPECTED_DPSi,t, is the difference between the

actual cash dividend per share (DPSit) and the expected cash dividend per share

112 Earnings management is quite common in China because listed firms attempt to avoid being delisted dueto reporting net losses for three consecutive years. The growth forecasts generated from industry average aremuch smoother than those based on individual firm’s historic earnings.113 The industry means of EPSs are based on all observations’ EPS of industry j in any specific year.114 The industry means of DPSs are based on the cash payer’s DPS of industry j in any specific year.

122

scaled by the share price of announcement day (SPit), in order to avoid the influence of zero

dividends in the prior fiscal year. UNEXPECTED_DPSi,t and UNEXPECTED_EPSi,t are the key

independent variables in explaining the abnormal return during an announcement period:

Besides the unexpected dividends and unexpected earnings, two dummy variables,

SDCRit and SDREit, are generated as proxies of stock dividend from capital reserve and stock

dividend from retained earnings, respectively. With regard to the information on UCSs, the

level of control rights, CONTROL_RIGHTS, is included to reflect the UCSs’ ownership in the

underlying listed firm, as well as four dummy variables, DPrivate, DSOECG, DSOELG and

DSAMB, indicating various types of UCSs are introduced to detect their divergent impact on

dividend announcement effects.

5.3.3.3 Other control variables

Several firm level control variables are introduced. Chinese listed firms are

accustomed to announcing annual financial statements earlier when operating results are

satisfactory, but defer announcements when performance is poor (Haw et al., 2005, Cheng

et al., 2009, Chen et al., 2005a). Therefore, the logarithm of the number of days between

financial year end and financial report announcement day, RPTTIMING, is introduced to

control the timing of financial statements. Firm size may also exert an influence on the

abnormal return around announcement date, so the logarithm of total assets, SIZE, is

included. The idiosyncratic risk, IDIO_RISK, is included which is computed as the standard

123

deviation of residuals generated from market model Eq.2 between 145 and 35 days prior

to the announcement date. The proxy of financial risk, LEVERAGE, is defined as total

interest bearing liabilities scaled by total equity. Idiosyncratic risk and financial leverage are

expected to negatively associate with CAR. Year dummies are also included. Table 5.2

presents the variables used in this chapter.

Table 5 2 Variable definition

CAR_PRE cumulative abnormal return generated from market model over the event window of [ 10, 2]CAR_SHORT cumulative abnormal return generated from market model over the event window of [ 1, +1]

CAR_MEDIUM cumulative abnormal return generated from market model over the event window of [ 3, +3]

CAR_LONG cumulative abnormal return generated from market model over the event window of [ 10, +10]

UNEXPECTED_DPS unexpected dividend per share generated from expectation model specified in Eq. 5.3

UNEXPECTED_EPS unexpected earnings per share generated from expectation model specified in Eq. 5.4

SDCR dummy variable which equals to 1 when there is stock dividend from capital reserve in year t, and 0 otherwise

SDRE dummy variable which equals to 1 when there is stock dividend from retained earnings in year t, and 0 otherwise

DPrivate dummy variable which equals to 1 when the ultimate controlling shareholder is privately owned or natural person inyear t, and 0 otherwise

DSOECG dummy variable which equals to 1 when the ultimate controlling shareholder is state owned enterprises affiliated withcentral government in year t, and 0 otherwise

DSOELG dummy variable which equals to 1 when the ultimate controlling shareholder is state owned enterprises affiliated withlocal government in year t, and 0 otherwise

DSAMB dummy variable which equals to 1 when the ultimate controlling shareholder is state owned asset managementbureau in year t, and 0 otherwise

CONTROL_RIGHTS the direct and indirect shareholdings controlled by ultimate controlling shareholders

RPTTIMING the logarithm of the number of days between the financial year end and financial report and dividend decisionannouncement date as of year t

SIZE the logarithm of the total assets as of year t

IDIO_RISK the standard deviation of residuals generated from Eq.4.1 between 145 and 35 days prior to the announcement date

LEVERAGE total interest bearing liabilities scaled by total equity as of year t

Year Dummy a set of dummy variables indicating the financial years from 1999 to 2009

5.3.4. Descriptive statistics

Table 5.3 presents the descriptive statistics of dependent and independent variables

associated with the cash payer, cash only payer and cash only payer with UCSs115 samples,

respectively. As shown in every panel, all CARs, except CAR_PRE, exhibits negative mean.

The comparison of mean CARs between Panel A and Panel B suggests the observations with

stock dividend have better CARs in all four windows. The comparison of mean CARs

between Panel B and Panel C indicates some deterioration in recent periods.

115 The CSMAR database provides information on UCSs since 2003, so cash only payer with UCSs sample,shown in Panel C is from 2003 to 2010, while the other two cover the observation window from 1999 to 2010.

124

Table 5 3 Descriptive statistics

Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.

Panel A: Cash payer sample (1999 2010, N = 6,071)CAR_PRE 0.0070 0.0644 0.0650 0.0310 0.0014 0.0391 0.0885CAR_SHORT 0.0052 0.0488 0.0634 0.0325 0.0069 0.0195 0.0541

CAR_MEDIUM 0.0025 0.0656 0.0771 0.0400 0.0066 0.0310 0.0766

CAR_LONG 0.0011 0.1035 0.1227 0.0614 0.0064 0.0532 0.1292

UNEXPECTED_DPS 0.0010 0.0092 0.0083 0.0023 0.0005 0.0047 0.0109

UNEXPECTED_EPS 0.1796 1.7040 0.6938 0.4273 0.0676 0.4570 1.2971

IDIO_RISK 0.0210 0.0074 0.0117 0.0152 0.0205 0.0263 0.0311

SIZE 21.7129 1.1163 20.4264 20.9047 21.5493 22.3433 23.1978

LEVERAGE 0.5296 0.4849 0.4043 2.2984 0.0000 1.36 4.87

Panel B: Cash only payer sample (1999 – 2010, N = 4,806)

CAR_PRE 0.0034 0.0608 0.0643 0.0322 0.0012 0.0330 0.0793CAR_SHORT 0.0095 0.0457 0.0655 0.0341 0.0097 0.0147 0.0455

CAR_MEDIUM 0.0083 0.0614 0.0790 0.0426 0.0106 0.0238 0.0635

CAR_LONG 0.0074 0.0995 0.1238 0.0636 0.0109 0.0447 0.1146

UNEXPECTED_DPS 0.0014 0.0095 0.0081 0.0020 0.0007 0.0053 0.0120

UNEXPECTED_EPS 0.1384 1.8569 0.7158 0.4392 0.0674 0.4571 1.2965

IDIO_RISK 0.0206 0.0074 0.0114 0.0148 0.0200 0.0258 0.0310

SIZE 21.7362 1.1279 20.4641 20.9277 21.5653 22.3572 23.2234

LEVERAGE 0.5302 0.4833 0.0298 0.1621 0.4055 0.7635 1.1848

Panel C: Cash only payer sample with control rights (2003 – 2010, N = 3,436)

CAR_PRE 0.0041 0.0671 0.0726 0.0374 0.0000 0.0388 0.0864CAR_SHORT 0.0117 0.0496 0.0726 0.0395 0.0116 0.0152 0.0480

CAR_MEDIUM 0.0103 0.0669 0.0887 0.0496 0.0135 0.0267 0.0699

CAR_LONG 0.0094 0.1085 0.1393 0.0744 0.0132 0.0492 0.1272

UNEXPECTED_DPS 0.0010 0.0101 0.0088 0.0021 0.0004 0.0049 0.0120

UNEXPECTED_EPS 0.0295 1.9366 0.7392 0.4967 0.1479 0.3950 1.2079

CONTROL_RIGHTS 0.4199 0.1559 0.2121 0.2975 0.4184 0.5359 0.6310

IDIO_RISK 0.0229 0.0068 0.0141 0.0177 0.0226 0.0277 0.0320

SIZE 21.9204 1.1796 20.5829 21.0794 21.7333 22.5929 23.5000

LEVERAGE 0.5760 0.5132 0.0316 0.1810 0.4487 0.8285 1.2719

Panel D: Cash only payer sample with Private UCSs (N = 922)

CAR_PRE 0.0060 0.0686 0.0692 0.0383 0.0001 0.0399 0.0938

CAR_SHORT 0.0133 0.0511 0.0774 0.0448 0.013 0.0166 0.0486

CAR_MEDIUM 0.0103 0.0686 0.0897 0.0522 0.0138 0.032 0.077

CAR_LONG 0.0107 0.1086 0.1393 0.0739 0.0168 0.05 0.126

UNEXPECTED_DPS 0.0002 0.0101 0.0097 0.0029 0.0000 0.004 0.011

UNEXPECTED_EPS 0.0048 1.474 0.7323 0.5123 0.2122 0.2874 1.0134

CONTROL_RIGHTS 0.346 0.1389 0.1812 0.2452 0.3203 0.4255 0.555

IDIO_RISK 0.0237 0.0063 0.0157 0.0192 0.0234 0.0283 0.0317

SIZE 21.4445 0.9000 20.3555 20.8053 21.3298 22.0374 22.6544

LEVERAGE 0.5519 0.4721 0.0333 0.1877 0.4552 0.8031 1.1819

Panel E: Cash only payer sample with SOECG UCSs (N = 280)

CAR_PRE 0.0025 0.0679 0.0636 0.0379 0.0013 0.0328 0.0771

CAR_SHORT 0.0088 0.0489 0.0656 0.0375 0.0094 0.018 0.0581

CAR_MEDIUM 0.0113 0.0662 0.0932 0.0470 0.0108 0.0281 0.0633

CAR_LONG 0.0133 0.114 0.1388 0.0710 0.0138 0.0404 0.1347

UNEXPECTED_DPS 0.0008 0.0099 0.0090 0.0020 0.0002 0.004 0.0106

UNEXPECTED_EPS 0.0658 2.3292 0.7768 0.5553 0.1889 0.4773 1.4331

CONTROL_RIGHTS 0.433 0.1823 0.1832 0.2632 0.4494 0.5717 0.6993

IDIO_RISK 0.0226 0.0071 0.0141 0.0174 0.0216 0.0277 0.0325

SIZE 21.9618 1.3056 20.5906 21.0957 21.7351 22.558 23.7786

LEVERAGE 0.531 0.4809 0.0003 0.1773 0.4531 0.7613 1.0424

125

Table 5.3 Descriptive statistics (Cont.)Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.

Panel F: Cash only payer sample with SOELG UCSs (N = 550)

CAR_PRE 0.0088 0.063 0.0652 0.0316 0.0029 0.0441 0.0886

CAR_SHORT 0.0124 0.0476 0.0684 0.0372 0.0128 0.0125 0.0449

CAR_MEDIUM 0.0095 0.0648 0.0789 0.0455 0.0123 0.0214 0.0611

CAR_LONG 0.0029 0.0985 0.1262 0.0666 0.0033 0.0533 0.1134

UNEXPECTED_DPS 0.0015 0.0109 0.0095 0.0024 0.0002 0.0062 0.0142

UNEXPECTED_EPS 0.209 2.5397 0.8096 0.5275 0.1397 0.4087 1.2075

CONTROL_RIGHTS 0.4386 0.1492 0.25 0.3207 0.4323 0.554 0.6332

IDIO_RISK 0.0209 0.0068 0.0127 0.0155 0.02 0.025 0.031

SIZE 21.668 0.9618 20.5693 20.9558 21.53 22.2959 22.9539

LEVERAGE 0.5472 0.4829 0.0333 0.1568 0.4264 0.8033 1.2171

Panel G: Cash only payer sample with SAMB UCSs (N = 1,684)

CAR_PRE 0.0017 0.0674 0.0776 0.0384 0.0014 0.0376 0.0836

CAR_SHORT 0.0111 0.0495 0.0703 0.0386 0.0111 0.0146 0.0476

CAR_MEDIUM 0.0104 0.0667 0.0889 0.0504 0.0139 0.0256 0.0693

CAR_LONG 0.0102 0.1106 0.1443 0.0776 0.0147 0.0478 0.1326

UNEXPECTED_DPS 0.0012 0.0099 0.0083 0.0018 0.0007 0.0051 0.0119

UNEXPECTED_EPS 0.0164 1.857 0.7215 0.4827 0.1076 0.4632 1.3296

CONTROL_RIGHTS 0.4521 0.1487 0.2487 0.3428 0.4649 0.5587 0.6441

IDIO_RISK 0.0233 0.007 0.0142 0.0179 0.023 0.0278 0.0325

SIZE 22.2565 1.2467 20.8303 21.3617 22.0634 23.0022 23.9618

LEVERAGE 0.6061 0.547 0.0334 0.1875 0.4584 0.8708 1.3873

This table contains descriptive statistics for variables of interest that are used in the following analysis. PanelA presents the descriptive statistics of cash payer sample; Panel B presents the descriptive statistics of cashonly payer sample; Panel C presents the descriptive statistics of cash only payer sample with UCSs’information. Panel D, Panel E, Panel F and Panel G present the descriptive statistics of cash only payer withPrivate, SOECG, SOELG and SAMB UCSs, respectively. Refer to Table 5.2 for variable definitions.

Both UNEXPECTED_DPS and UNEXPECTED_EPS show positive means and large

standard deviations. The low value of UNEXPECTED_DPS is due to the use of share price as

the denominator. Themagnitude of SDCR is larger than that of SDRE, in terms of both mean

and median values. This result is not surprising given that SDRE is restricted by the current

net profit, while SDCR depends on the accumulated capital reserve. When the sample with

UCSs information is subdivided based on the types of UCSs, the summary statistics are

reported in Panel D to Panel G of Table 5.3. SOECG controlled firms report the highest

average CAR_SHORT and SOELG controlled firms lead in CAR_LONG. Meanwhile, four

categories exhibit similar average UNEXPECTED_DPS, but different mean

UNEXPECTED_EPS, amongwhich SOELG controlled firms report worstmean earnings shock.

126

Table 5 4 Pearson's correlation matrix

Panel A

N = 6,071 UNEXPECTED_DPS UNEXPECTED_EPS IDIO_RISK Size LEVERAGE

CAR_PRE 0.0116 0.0070 0.0489* 0.0064 0.001

CAR_SHORT 0.0162 0.0132 0.0814* 0.0286 0.0044

CAR_MEDIUM 0.0096 0.0155 0.0897* 0.0076 0.0109

CAR_LONG 0.0115 0.0223 0.1197* 0.0001 0.0025

UNEXPECTED_DPS 0.1659* 0.0136 0.0278 0.0229

UNEXPECTED_EPS 0.0457* 0.0211 0.0294

IDIO_RISK 0.0408* 0.1238*

Size 0.3296*

Panel B

N = 4,806 UNEXPECTED_DPS UNEXPECTED_EPS IDIO_RISK Size LEVERAGE

CAR_PRE 0.0152 0.0073 0.0591* 0.0015 0.0124

CAR_SHORT 0.0406* 0.0041 0.1048* 0.0562* 0.0063

CAR_MEDIUM 0.0268 0.0202 0.1127* 0.028 0.0212

CAR_LONG 0.02000 0.0278 0.1302* 0.0113 0.0044

UNEXPECTED_DPS 0.1574* 0.0245 0.0398* 0.0341

UNEXPECTED_EPS 0.0364 0.0272 0.0205

IDIO_RISK 0.0490* 0.1336*

Size 0.3136*

Panel C

N = 4,425 UNEXPECTED_DPS UNEXPECTED_EPS IDIO_RISK Size LEVERAGE CONTROL_RIGHTS

CAR_PRE 0.0141 0.004 0.0923* 0.0099 0.0232 0.0003

CAR_SHORT 0.0117 0.012 0.0954* 0.0717* 0.0162 0.0334*

CAR_MEDIUM 0.0077 0.0417 0.1295* 0.0438 0.0174 0.0170

CAR_LONG 0.0086 0.0475* 0.1586* 0.0198 0.0080 0.0173

UNEXPECTED_DPS 0.1411* 0.0049 0.0303 0.0287 0.0142

UNEXPECTED_EPS 0.0126 0.0074 0.0399 0.0101

IDIO_RISK 0.0992* 0.0583* 0.1184*

Size 0.2992* 0.2378*

LEVERAGE 0.0702*

This table contains Pearson’s correlation coefficients matrix for selected variables that are used in thefollowing analysis. Panel A presents the descriptive statistics of cash payer sample; Panel B presents thedescriptive statistics of cash only payer sample; Panel C presents the descriptive statistics of cash payersample with control rights since 2003. Refer to Table 5.2 for variable definitions. * denotes statisticalsignificance at 1% level.

The Pearson’s correlation coefficients matrix is reported in Table 5.4. In the cash

payer sample, unexpected dividends have no significant correlation with the cumulative

abnormal return of all pre event and event windows. But in the cash only payer sample,

unexpected dividend reports a significant positive correlation with CAR_SHORT and a

marginally positive correlation with CAR_MEDIUM. But when observation window is from

2003 to 2010, as shown in Panel C, dividend shock has little correlation with CARs, but

127

earnings shock has a significant negative correlation, although only marginally, with the

CAR_MEDIUM. Inconsistent with expectation, the correlation coefficients between

unexpected earnings and CARs are either insignificant or significantly negative.

The correlation coefficients between CARs and IDIO_RISK are all negative and

significant at 1% level, which indicates that higher volatility in the share price is associated

with lower CARs. SIZE has little association with abnormal returns, except for CAR_SHORT

in cash only payer and cash payer with CONTROL_RIGHTS samples. The announcement

timing effect, RPTTIMING, is negatively correlated with CAR_PRE and CAR_LONG, in line

with the expectation that investors are concerned about delayed announcements.

5.4. Empirical results

5.4.1. Univariate analysis

Table 5.5 reports the mean CARs by grouping observations based on the sign of

UNEXPECTED_EPS and UNEXPECTED_DPS, and the presence of a simultaneous stock

dividend. Obviously, the stock dividend (either or both of SDCR and SDRE) payers report

superior CARs across all observation windows, no matter whether the coefficients on

UNEXPECTED_EPS and UNEXPECTED_DPS are positive or negative, consistent with the

previous literature that Chinese investors respond positively to the stock dividend

announcement (Chen et al., 2002, Yi et al., 2007, Cheng et al., 2009).

128

Table 5 5 Average CARs of different combinations based on earnings shock, dividend shock andsimultaneous stock dividend

N CAR_PRE CAR_SHORT CAR_MEDIUM CAR_LONGPositive UNEXPECTED_EPS 4,061

Cash payerWithout Stock dividend

Positive UNEXPECTED_DPS 1,488 0.0020 0.0069 0.0075 0.0103(1.25) ( 5.45)*** ( 4.43)*** ( 3.86)***

Negative UNEXPECTED_DPS 703 0.0007 0.0083 0.0090 0.0136

( 0.32) ( 5.02)*** ( 4.04)*** ( 4.18)***

With stock dividend

Positive UNEXPECTED_DPS 311 0.0244 0.0143 0.0189 0.0210(5.45)*** (4.21)*** (4.43)*** (3.20)***

Negative UNEXPECTED_DPS 249 0.0184 0.0132 0.0233 0.0221

(4.11)*** (3.49)*** (5.01)*** (3.21)***

Cash non payer

Without stock dividend 1,150 0.0035 0.0073 0.0065 0.0036(1.62) ( 4.80)*** ( 2.95)*** ( 1.04)

With stock dividend 160 0.0254 0.0193 0.0219 0.0232

(3.74)*** (3.43)*** (3.21)*** (2.26)**

Positive stock dividend 720 0.0226 0.0150 0.0211 0.0219(7.78)*** (6.46)*** (7.26)*** (5.04)***

Zero stock dividend 3,341 0.0020 0.0073 0.0075 0.0087

(1.74)* ( 8.68)*** ( 6.40)*** ( 4.78)***

Negative UNEXPECTED_EPS 5,344

Cash payerWithout Stock dividend

Positive UNEXPECTED_DPS 1,201 0.0069 0.0095 0.0062 0.0011(3.41)*** ( 6.31)*** ( 2.95)*** ( 0.34)

Negative UNEXPECTED_DPS 1,414 0.0041 0.0121 0.0097 0.0049

(2.32)** ( 9.22)*** ( 5.48)*** ( 1.73)*

With stock dividend

Positive UNEXPECTED_DPS 284 0.0243 0.0112 0.0249 0.0295(4.73)*** (3.29)*** (5.03)*** (3.84)***

Negative UNEXPECTED_DPS 421 0.0182 0.0090 0.0157 0.0229

(4.85)*** (3.04)*** (3.78)*** (3.72)***

Cash non payer

Without stock dividend 1,830 0.0026 0.0071 0.0056 0.0040(1.21) ( 5.47)*** ( 3.20)*** (1.29)

With stock dividend 194 0.0205 0.0107 0.0198 0.0337

(3.55)*** (2.01)** (2.79)*** (3.65)***

Positive stock dividend 899 0.0206 0.0101 0.0195 0.0273(7.65)*** (4.80)*** (6.66)*** (6.41)***

Zero stock dividend 4,445 0.0042 0.0093 0.0071 0.0002

(3.58)*** ( 11.81)*** ( 6.55)*** ( 0.14)

This table reports average CARs of different combinations based on earnings shock, dividend shock andsimultaneous stock dividend. Refer to Table 5.2 for variable definitions.

The number of cash only payers with both positive UNEXPECTED_DPS and

UNEXPECTED_EPS (1,488) is twice those with positive UNEXPECTED_EPS but negative

UNEXPECTED_DPS (703), while there are more cash only payers with both negative

UNEXPECTED_DPS and UNEXPECTED_EPS (1,414) than those with negative

129

UNEXPECTED_EPS but positive UNEXPECTED_DPS (1,201). The outcome suggests a positive

association between UNEXPECTED_DPS and UNEXPECTED_EPS. Meanwhile, CARs in both

Panel A and Panel B indicate, without stock dividends, the impact from UNEXPECTED_EPS

seems to be consistently negative, no matter whether it is positive or negative, while a

positive UNEXPECTED_DPS slightly alleviates the impact from the UNEXPECTED_EPS.

Figure 5 1 CAAR trends of cash only payers grouped by the sign of UCEXPECTED_DPS andUNEXPECTED_EPS

Figure 5.1116 presents the trend of cash only payers’ cumulative average abnormal

return (CAAR) from day 10 to day +10, grouped by the signs of UNEXPECTED_DPS and

UNEXPECTED_EPS. Similar to Table 5.5, when the signs on UNEXPECTED_EPS

(UNEXPECTED_DPS) are identical, CAAR of groups with positive (negative)

UNEXPECTED_DPS (UNEXPECTED_EPS) is better than that with negative (positive)

UNEXPECTED_DPS (UNEXPECTED_EPS). The group with positive (negative)

UNEXPECTED_DPS and negative (positive) UNEXPECTED_EPS report the best (worst) CAAR

performance. The figure suggests that the UNEXPECTED_DPS (UNEXPECTED_EPS) has

positive (negative) influence on the stock abnormal return around the announcement day.

116 As the influence of stock dividend is quite obvious, the remaining univariate analysis focuses on the cashonly payers.

0.015

0.010

0.005

0.000

0.005

0.010

T 10 T 9 T 8 T 7 T 6 T 5 T 4 T 3 T 2 T 1 T T+1 T+2 T+3 T+4 T+5 T+6 T+7 T+8 T+9 T+10

UDPS>0 UEPS>0 UDPS>0 UEPS<0 UDPS<0 UEPS>0 UDPS<0 UEPS<0

130

It can be explained that the external investors take the dividend shock as an incremental

economic benefit and a signal of reduced free cash flow, but the earnings shock might be

suspicious because of the earnings management practice.

The CAAR trends of cash only payers grouped by types UCSs, depicted in Figure 5.2,

suggest the divergence among the four categories is quite significant. All four groups exhibit

negative CAARs after the event day, amongwhich the observationswith SOELGUCSs exhibit

the best CAAR performance, while the group with SAMB UCSs report stable but inferior

CAAR performance. Without controls for firm specific factors, this preliminary result is

slightly against Hypothesis 5.3.

Figure 5 2 CAAR trends of cash only payers grouped by the types of UCSs

5.4.2. Multivariate analysis

5.4.2.1 Stock dividend and CAR

The regression models to investigate Hypothesis 5.1 and Hypothesis 5.2 are

constructed based on Cheng et al. (2009) with some amendments in the control variables.

CAR refers to the cumulative abnormal return across the [ 10, 2], [ 1, +1], [ 3, +3] and [ 10,

+10] windows around the dividend announcement day. Eq.5.7 is estimated using ordinary

0.020

0.015

0.010

0.005

0.000

0.005

0.010

0.015

T 10 T 9 T 8 T 7 T 6 T 5 T 4 T 3 T 2 T 1 T T+1 T+2 T+3 T+4 T+5 T+6 T+7 T+8 T+9 T+10

SOECG SOELG SAMB Private

131

least squares (OLS) regression adjusted for heteroscedasticity (White, 1980), and estimation

results are reported in Table 5.6.

Consistent with Table 5.5 and findings in past studies that Chinese negotiable

shareholders respond positively to stock dividends (Cheng et al., 2009, Anderson et al.,

2011), both coefficients on SDCR and SDRE are significantly positive, as per Panel B of Table

5.6. The presence of SDCR is expected to improve CARs by 2.00%, 2.00%, 2.82% and 3.51%,

respectively, after controlling for firm specific parameters. The contribution from SDRE is

relatively weaker, which is expected to lift the abnormal returns by 0.50%, 0.90%, 1.00%

and 1.00%, respectively. These outcomes support Hypothesis 5.1 that investors react

positively to stock dividend announcements and SDCR is more welcome.

Table 5 6 Stock dividend announcement and stock abnormal returns

Expected signs CAR_PRE p value CAR_SHORT p value CAR_MEDIUM p value CAR_LONG p value

Intercept 0.1414*** (0.000) 0.0493*** (0.003) 0.0008 (0.970) 0.0937*** (0.007)

UNEXPECTED_DPS + 0.1535* (0.072) 0.1438** (0.031) 0.1900** (0.032) 0.2939** (0.026)

UNEXPECTED_EPS 0.0011** (0.040) 0.0001 (0.856) 0.0006 (0.280) 0.0009 (0.316)

SDCR + 0.0210*** (0.000) 0.0210*** (0.000) 0.0304*** (0.000) 0.0384*** (0.000)

SDRE + 0.0047 (0.147) 0.0095*** (0.000) 0.0099*** (0.002) 0.0098** (0.046)

RPTTIMING 0.0216*** (0.000) 0.0014 (0.518) 0.0015 (0.574) 0.0120*** (0.004)

IDIO_RISK 1.4057*** (0.000) 0.5276*** (0.000) 1.1459*** (0.000) 2.4693*** (0.000)

Size + 0.0010 (0.211) 0.0023*** (0.000) 0.0013 (0.127) 0.0001 (0.994)

LEVERAGE 0.0011 (0.547) 0.0002 (0.877) 0.006 (0.726) 0.0022 (0.437)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.063 0.047 0.062 0.084

F test 17.02 12.82 16.94 21.96

This table reports cross sectional analysis on market reaction to the dividend announcement. Eq.5.7 isestimated using ordinary least squares (OLS) regression. Refer to Table 5.2 for variable definitions. Yeardummies are included. The sample period is from 1999 to 2010. Figures in parentheses are p values based onWhite’s heteroscedasticity robust standard error (White, 1980). *, ** and *** denote significance at 10%, 5%and 1% levels, respectively.

132

Among the control variables, RPTTIMING exhibits a negative sign in both pre event

and long event windows, suggesting delayed announcements are deemed to be more

doubtful. Consistent with expectation that higher volatility undermines the announcement

effect, idiosyncratic risk, IDIO_RISK, shows significant negative coefficients in all

observation windows. At the same time, firm size, Size, has a marginal influence on

CAR_SHORT, which indicates larger firms have stronger abnormal returns close to the

announcement day. Minority shareholders that focus on growth do not prefer holding

shares in firms with a larger asset base. Different from expectation, financial leverage,

LEVERAGE, has little impact on CARs.

5.4.2.2 Unexpected cash dividend, unexpected earnings and CAR

Eq.5.7, excluding SDCR and SDRE, is applied on the cash only payer sample and the

results are reported in Table 5.7. Panel A indicates that, without firm level controls, the

influence ofUNEXPECTED_DPS is significant in three event windows, while the contribution

from UNEXPECTED_EPS is insignificant in all four windows. When firm level controls are

incorporated, as in Panel B, the coefficients on UNEXPECTED_DPS are significantly positive

at the 1% level in CAR_SHORT column, and positive but marginally significant at 10% level

in CAR_MEDIUM column. Meanwhile, the coefficients on UNEXPECTED_EPS remain

insignificant, except CAR_PRE column in which a marginally positive coefficient on

UNEXPECTED_EPS is observed.

133

Table 5 7 Unexpected earnings, unexpected dividend and stock abnormal returns

Expected Sign CAR_PRE p value CAR_SHORT p value CAR_MEDIUM p value CAR_LONG p value

Intercept 0.1361*** (0.000) 0.0747*** (0.000) 0.0242 (0.302) 0.0682* (0.071)

UNEXPECTED_DPS + 0.0729 (0.412) 0.1778*** (0.009) 0.1665* (0.076) 0.2067 (0.138)

UNEXPECTED_EPS 0.0010* (0.068) 0.0003 (0.472) 0.0006 (0.284) 0.0010 (0.270)

RPTTIMING 0.0195*** (0.000) 0.0005 (0.832) 0.0029 (0.314) 0.0108** (0.015)

IDIO_RISK 1.3140*** (0.000) 0.4079*** (0.004) 0.9552*** (0.000) 2.2132*** (0.000)

Size + 0.0009 (0.298) 0.0035*** (0.000) 0.0026*** (0.003) 0.0011 (0.435)

LEVERAGE 0.0009 (0.650) 0.0002 (0.962) 0.0024 (0.220) 0.0014 (0.655)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.053 0.023 0.039 0.080

F test 12.51 7.47 9.42 17.54

This table reports cross sectional analysis on market reaction to the dividend announcement. Eq.5.7,excluding SDCR and SDRE, is estimated using ordinary least squares (OLS) regression. Refer to Table 5.2 forvariable definitions. The sample period is from 1999 to 2010. Figures in parentheses are p values based onWhite’s heteroscedasticity robust standard error (White, 1980). *, ** and *** denote significance at 10%, 5%and 1% levels, respectively.

These results indicate investors react positively to dividend shocks but sluggishly to

earnings shocks, which is contradictory to previous studies (Chen et al., 2002, Cheng et al.,

2009, Anderson et al., 2011) and can be attributed to several reasons. First, the earnings

management practice in the Chinese market has undermined the credibility of a positive

earnings shock (Haw et al., 2005, Wang et al., 2008, Chen et al., 2009d, Shafer and Wang,

2011). An outstanding earnings announcement is unable to stir up the minority

shareholders. Second, although cash dividends paid to individual investors are subject to

income tax, an unexpected cash dividend increase is a positive signal, not only because it is

a real economic benefit paid to investors, but it is also viewed as a symptom of robust

operating performance and strong earnings. Third, the unexpected cash dividend payout

will reduce the cash balance controlled by the firm, which mitigates the agency cost of free

cash flow. Chinese minority shareholders are disadvantaged in both taxation and corporate

governance, so an unexpected increase in cash dividends is better than letting the

management, and the controlling shareholders behind the management, hoard the cash

balance and facilitate their over investment and empire building activities. In other words,

134

it can be viewed as an alternative version of ‘the bird in the hand’ argument117 that Chinese

minority shareholders favour cash flows from listed firms more than the managers’ growth

story. Last but not least, although the unexpected cash dividend may be another kind of

tunnelling behaviour by the controlling shareholders, Chinese minority shareholders expect

the large shareholders will compensate by providing more support in the future as both

expropriation and propping up exists in the Chinese markets (Cheung et al., 2009a).

5.4.2.3 Level of control rights, types of UCSs and CAR

Eq.5.8 is developed from Eq.5.7 to approach the association between level of voting

rights and CARs is examined first and the estimation results are reported in Table 5.8.

Table 5.8 shows neither CONTROL_RIGHTS nor UNEXPECTED_DPS has an impact on

the CARs of pre event and event windows,118 which can be explained in two tiers. One

reason is that the information about UCSs’ voting rights are available during the financial

year if investors follow firms’ periodic disclosures and special disclosures after share block

trading. In other words, these information has been incorporated into the pricing and

decreases its impact on CARs around the announcement day. The other reason is that

minority shareholders are only concerned about who the UCS is, rather than how strongly

117 The classic ‘bird in the hand’ argument refers to investors preferring cash flow with certainty to growthopportunity with variability.118 UNEXPECTED_EPS has a marginal negative influence on the CAR_LONG.

135

the firm is controlled. At the same time, the insignificant coefficients on UNEXPECTED_DPS

shown in Table 5.8 indicate that the general positive market reaction to unexpected cash

dividends has faded in recent years. Instead, there is a possibility that minority shareholders

respond differently to cash dividends shock from firms with certain types of UCSs.

Table 5 8 Control rights and stock abnormal returns

Expected sign CAR_PRE CAR_SHORT CAR_MEDIUM CAR_LONG

Intercept 0.1525*** (0.000) 0.0807*** (0.000) 0.0324 (0.233) 0.0749* (0.091)

CONTROL_RIGHTS 0.0009 (0.901) 0.0021 (0.704) 0.0021 (0.780) 0.0026 (0.822)

UNEXPECTED_DPS + 0.0936 (0.375) 0.0781 (0.325) 0.0926 (0.394) 0.1620 (0.323)

UNEXPECTED_EPS 0.0005 (0.391) 0.0003 (0.467) 0.0009 (0.154) 0.0019* (0.060)

RPTTIMING 0.0215*** (0.000) 0.0032 (0.278) 0.0012 (0.737) 0.0123** (0.033)

IDIO_RISK 1.4436*** (0.000) 0.4840*** (0.003) 1.1237*** (0.000) 2.3429*** (0.000)

Size + 0.0012 (0.288) 0.0030*** (0.000) 0.0026** (0.011) 0.0013 (0.452)

LEVERAGE 0.0016 (0.485) 0.0001 (0.943) 0.0030 (0.190) 0.0001 (0.967)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.057 0.018 0.043 0.089

F test 11.92 3.96 9.50 18.56

This table reports cross sectional analysis on market reaction to the dividend announcement. Eq5.8 isestimated without and with control variables using ordinary least squares (OLS) regression as shown in PanelA and Panel B, respectively. Refer to Table 5.2 for variable definitions. The sample period is from 1999 to 2010.p values based on White’s heteroscedasticity robust standard error (White, 1980) are reported inparentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

Eq.5.9 is another derivative from Eq.5.7 to pursue the association between

UNEXPECTED_DPS from types of UCSs and CARs.119 The cross effects between

UNEXPECTED_DPS and UCSs dummy variables, DSOECG, DSOELG, DSAMB and DPrivate,

are introduced to investigate which UCSs’ cash dividend shock affects the stock abnormal

return and the direction of the influence. The estimation results are presented in Table 5.9.

119 As the regression results of Eq.5.8 suggest there is little impact from CONTROL_RIGHTS on CARs,CONTROL_RIGHTS is not included as a firm specific control variable in Eq.9. In the meantime, the types ofUCSs, along with CONTROL_RIGHTS, are available to investors. Therefore, investors will not react to the typesof UCSs during announcement period. Instead, this research investigates the conditional contribution fromtypes of UCSs, that is, the unexpected cash dividend from firms with which type of UCS will lead to superioror inferior stock abnormal return during announcement period.

136

As per Table 5.9, cash dividend shocks from different types of UCSs have divergent

impacts on the CARs. The cross effect between DSAMB and UNEXPECTED_DPS reports is

significantly positive (at the 5% level) in CAR_SHORT and CAR_MEDIUM regressions at the

1% and 5% significance levels, respectively. SAMBUCSs are effectively government agencies

with various tasks beyond economic profits, and are less concerned about the demand for

cash dividends by minority shareholders in the controlled public firms, and are more likely

to hoard cash.120 Therefore, the cash dividend shock from SAMB controlled listed firms is

interpreted as a positive signal because minority shareholders welcome the additional cash

flow from the firms which are reluctant to make distributions and are deemed to havemore

agency costs from free cash flow. Moreover, the unexpected cash dividend may also hint at

the controlling shareholders’ confidence in the operation and its financial capacity.

Further, the cross effect between DPrivate and UNEXPECTED_DPS shows marginal

positive coefficients in CAR_SHORT and CAR_MEDIUM regressions at the 5% and 10% levels

of significance, respectively. Private UCSs have greater efficiency than the state owned

UCSs as they have strong motivation to monitor the listed firms. As it has become more

popular for Chinese private investors to recruit entrepreneurs to manage the operation of

120 Another potential reason why SAMB controlled firms are inclined to retain the cash is because of theirhigher financial leverage, as per the descriptive statistics shown in Table 5.3.

137

listed firms, the conflict of interest between them emerges. Therefore, cash dividends

gradually become a tool to cope with the agency cost. External investors welcome the cash

dividend shock from Private controlled firms, as they expect an improvement in operation

efficiency if the large cash distribution can reduce the agency cost of free cash flow.

Table 5.9: Type of UCSs and stock abnormal returnsCAR_PRE CAR_SHORT CAR_MEDIUM CAR_LONG

Intercept 0.1518*** (0.000) 0.0806*** (0.000) 0.0307 (0.255) 0.0751* (0.089)

DSOECG X UNEXPECTED_DPS 0.4364 (0.268) 0.5634 (0.112) 0.5858 (0.193) 0.3533 (0.505)

DSOELG X UNEXPECTED_DPS 0.2202 (0.416) 0.0786 (0.724) 0.0700 (0.777) 0.1741 (0.699)

DSAMB X UNEXPECTED_DPS 0.2127 (0.496) 0.5405** (0.021) 0.7626*** (0.005) 0.5758 (0.271)

DPrivate X UNEXPECTED_DPS 0.2868 (0.423) 0.5701** (0.047) 0.7195* (0.053) 0.3804 (0.559)

UNEXPECTED_DPS 0.3725 (0.264) 0.4348 (0.104) 0.5776* (0.075) 0.2404 (0.672)

UNEXPECTED_EPS 0.0005 (0.397) 0.0003 (0.481) 0.0009 (0.167) 0.0020* (0.060)

RPTTIMING 0.0215*** (0.000) 0.0033 (0.256) 0.0009 (0.802) 0.0122** (0.035)

IDIO_RISK 1.4434*** (0.000) 0.4910*** (0.003) 1.1362*** (0.000) 2.3580*** (0.000)

Size 0.0011 (0.291) 0.0031*** (0.000) 0.0025** (0.013) 0.0013 (0.407)

LEVERAGE 0.0017 (0.449) 0.0002 (0.926) 0.0028 (0.224) 0.0002 (0.963)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.056 0.018 0.045 0.089

F test 9.93 4.24 8.85 15.60

This table reports cross sectional analysis on market reaction to the dividend announcement. Eq5.9 isestimated using ordinary least squares (OLS) regression. Refer to Table 5.2 for variable definitions. Yeardummies are included. The sample period is from 1999 to 2010. p values based onWhite’s heteroscedasticityrobust standard error (White, 1980) are reported in parentheses. *, ** and *** denote significance at 10%, 5%and 1% levels, respectively.

5.4.3. Robustness Check

Several procedures are applied to the study as robustness checks. First, it is argued

the winsorization process may alter the fundamental pattern of the dataset. Therefore,

Eq.5.7 is estimated based on the samples before winsorization (refer to Appendix 5.1) and

the results are generally consistent with those from winsorized sample. Second, variance

inflated factors (VIF) and tolerance (1/VIF) are included in Appendix 5.2, which indicate

multicollinearity is not a significant issue. Third, Fama Macbeth two stage procedure (Fama

andMacBeth, 1973) is used to estimate Eq.5.7 (refer to Appendix 5.3). The outcomes are in

line with what has been reported in sections using OLS regressions.

138

5.5. Summary

This chapter investigates two empirical questions in the Chinese equity market, how

the investors react to the stock dividend decision, unexpected cash dividend and

unexpected earnings, and whether the type of ultimate controller, as well as their voting

rights, influences the stock abnormal returns around announcement day.

The ordinary least squares regression results suggest that both types of stock dividend

have positive contributions to the dividend announcement effect, which is consistent with

past literature. More specifically, stock dividends from capital reserve are more influential

than stock dividends from retained earnings, due to their different accounting and tax

treatments. Contradictory to past literature, Chinese minority shareholders respond

positively to unexpected cash dividends. This can be taken as the result of imbalanced

corporate governance infrastructure. Chinese minority shareholders passively regard the

additional cash dividends as an incremental economic benefit and a positive signal of

decrease in free cash flow. In contrast, they react negatively to earnings shock because the

earnings management practice is quite significant in Chinese market.

Finally, although there is little reaction on the information of voting rights held by

ultimate controlling shareholders, the unexpected dividend from firms with various types

of ultimate controlling shareholders has a dissimilar influence on the CAR. The dividend

shock from SOELG controlled firms is treated as negative, while the unexpected cash

dividend from SAMB controlled firms is welcomed by minority shareholders. This

phenomenon can be attributed to UCSs’ divergent backgrounds and operating style.

139

Chapter 6 Dividend Payout Policy and Future Earnings Growth in

the Chinese Stock Market

6.1. Introduction

According to conventional theory, the distribution of cash dividends will reduce the

internally generated cash balance and force a firm to raise external capital at a higher cost

when investment opportunities emerge. Because a firm’s value is inversely related to the

cost of capital, more retention will increase the firm’s value by decreasing the funding cost.

Financial managers may deviate from the optimal capital structure and follow a ‘pecking

order’ of raising capital that ranks the retained earnings as the cheapest source (Gordon,

1962, Myers, 1984). Some empirical studies support this proposed inverse relationship

between cash payout and future earnings growth (Rozeff, 1982, Fama and French, 2002,

Ibbotson and Chen, 2003).121

As one of the largest emerging markets, the Chinese equity market is strictly

monitored by the China Securities Regulatory Commission (CSRC). Listed firms have to apply

for approvals from the CSRC if they are going to raise funds through capital markets. This

scarcity of access to capital markets makes the retained earnings more precious as a source

of capital. Chinese firms are, therefore, more likely to retain the cash and prepare a future

investment plan. As shown in the previous two chapters, Chinese listed firms do not follow

the popular ‘sticky’ dividend policy, partly due to their investment demands and the high

121 Benartzi et al. (1997) argue that a change in cash dividends reflects achieved performance rather thanfuture earnings. Meanwhile, other studies contend that a change in cash dividends conveys privateinformation from the managers as they know more than the external shareholders do (Aharony and Swary,1980, Asquith andMullins, 1983, Brickley, 1983, Aharony and Dotan, 1994, Yoon and Starks, 1995, Nissim andZiv, 2001, Harada and Nguyen, 2005).

140

thresholds of the capital markets. Therefore, the relationship between the cash dividend

payout and future earnings growth is likely to be inverse in the Chinese market.

Although the cost of raising capital is lower as cash retention increases, investment

return is not guaranteed because the managers may be over optimistic. Further, the

agency principal problem caused by the separation between ownership and control may

facilitate entrepreneurs to abuse the cash holdings and conduct empire building activities,

which lead to the free cash flow problem. Working as a corporate governance mechanism,

cash dividends reduce the cash holdings controlled by themanagement and subject the firm

to the microscope of the financial markets when it is in need of further capital. Intuitively,

this monitoring mechanism enhances the managers’ commitment to the company and

improves their performance. In other words, cash dividend payouts may well be positively

related to future earnings growth (Jensen, 1986, La Porta et al., 2000a, Arnott and Asness,

2003, Zhou and Ruland, 2006, Gwilym et al., 2006, Vermeulen and Smit, 2011).

Corporate governance and investor protection in emerging markets are often weaker

leaving shareholders in those markets more likely to be exposed to agency principal

problems (La Porta et al., 1998, La Porta et al., 2000b). The Chinese equity market used to

be dominated by state owned enterprises and the managers of listed firms were appointed

as government officials, but this situation has changed and the diversification of

shareholders’ backgrounds has improved. Some studies suggest that Chinese listed firms

have become more accustomed to hiring professionals as corporate directors and their

remuneration packages are now closely linked with operating performance (Xu et al., 2005,

Adithipyangkul et al., 2011). Hence, besides the well known agency problem existing

141

between large and small shareholders documented in past studies, agency problems

between shareholders and managers have emerged as a new issue in the Chinese market.

In the previous two chapters, the relationship between ownership and dividend

policy, as well as the dividend announcement effect, in the Chinese stock market has been

investigated. Similar to other studies (Gao and Kling, 2008, Zou et al., 2008, Chen et al.,

2009b), the results observed in those two chapters can be explained by the free cash flow

problem in Chinese listed firms. On the one hand, ultimate controlling shareholders (UCSs)

utilize their control rights to force managers to increase the propensity and magnitude of

cash dividends, as they are concerned about potential abuse of cash holdings by the

management. On the other hand, minority shareholders respond differently towards the

unexpected cash dividends from firms with various types of UCSs, partly due to the

divergent levels of free cash flow problems. Both chapters point out the existence and

importance of the agency principal problem in the Chinese equity market.

Following the results of the previous two chapters, this chapter will pursue the

influence of current dividend decisions, including both cash dividends and stock dividends,

on future earnings growth in the Chinese stock market. In addition, this influence will be

scrutinized in firms with different levels of agency costs, proxied by Tobin’s Q, an indicator

of growth opportunity. Empirical results indicate a positive association between current

cash distribution and future earnings growth, but the influence is diminishing as the payout

ratio increases. Not only does it imply the validity of the explanation that cash dividends

work as a disciplinary function to mitigate the agency cost, but it also incorporates the

consideration of capital cost, also known as pecking order. Moreover, the positive impact

from cash payouts is more significant in the observations with greater Tobin’s Q, which

142

supports the free cash flow hypothesis that shareholders utilize cash payouts to cope with

the agency costs in firms with greater over investment potential. Finally, although the

empirical evidence in Chapter 5 suggests minority shareholders react positively to stock

dividend announcements, a negative association between stock dividend decisions and

future earnings growth is observed. Anderson et al. (2011) report a stock dividend is a

choice when a firm is experiencing poor liquidity or weak performance. This chapter shows

this negative implication is not limited to the current period, but also stretched into the

operating results in the following financial year(s). It supplies further evidence that stock

dividends in the Chinese equity market are signals of earnings management.

One contribution of this chapter is that it extends the studies on the traditional

agency principal problem to an emerging market where ownership and management is

being separated gradually. Although the concentration of ownership changes slowly, the

trend of selecting competent candidates as directors has become more popular in the

Chinese market. This chapter evaluates the capability of a cash dividend policy being used

as a corporate governance mechanism to alleviate agency costs as shareholdings become

more dispersed and professionals, rather than government officials or family members, are

appointed asmanagers. The second contribution of this chapter is to provide an explanation

of the stock dividend puzzle in the Chinese market from the angle of agency costs. In

general, stock dividends are inclined to suggest inferior future performance.

The remainder of this chapter is organized as follows. Section 6.2 proposes the

hypotheses, followed by Section 6.3 which introduces the sample construction and

descriptive statistics. Section 6.4 presents empirical results and the robustness test. Finally,

Section 6.5 summarizes the chapter.

143

6.2. Hypotheses development

As per the above discussion, Chinese listed firms encounter both the barriers to

capital markets set by the CSRC and agency costs between shareholders and managers. It is

conjectured that the latter has become more significant than the former, because the

evolution and openness of the Chinese financial markets is prompting the deregulation of

the financial sector. The competition among financial institutions also provides listed firms

with more flexibility to raise capital through various channels, which challenges the validity

of the pecking order theory in the Chinese market.

Instead, the agency cost between shareholders and managers is a relatively new

phenomenon in the Chinese market as the affiliation between managers and large

shareholders is becoming weaker. Conflicts between shareholders andmanagers in Chinese

firms, reflected by proxy contests, have been under the spotlight in recent years, and

entrepreneurs have become more and more independent from the large (founding)

shareholders.122 Compared with the pecking order theory, the free cash flow hypothesis

stemmed from agency costs which were attracting increased interest as a corporate

governance issue in the Chinese market.

Following previous studies which document mixed empirical evidence about the

relationship between cash payout and future earnings growth, this chapter proposes a

diminishing positive association between cash dividend payouts on future earnings growth

in the Chinese market. When the payout magnitude is relatively low, a positive relationship

between current cash dividend payouts and future earnings growth is expected, as cash

122 Two prominent cases are GOME Electrical Appliances Holding Limited (0493.HK) and NVC LightingTechnology Corporation (2222.HK).

144

dividends are employed as the tool tomitigate the shareholders’ concerns about the agency

cost of free cash flow. As the payout ratio increases, the marginal contribution to reducing

agency cost decreases and the cost of capital will overtake it as a more important issue.

Tobin’s Q is often used as a proxy for an over investment problem (Lang and

Litzenberger, 1989, Denis et al., 1994, McConnell and Servaes, 1995). Firms with higher

Tobin’s Q are deemed to have more investment opportunity but are also subject to more

severe agency costs of free cash flow. Over optimism may lead the management to indulge

in investment sprees and pick up projects with negative net present value, while

shareholders will bear the cost of the managers’ imprudent investment decisions.

Intuitively, the over investment will result in inferior future performance when the losses

from unprofitable investment projects are realized. Similar to Zhou and Ruland (2006), it is

conjectured that a positive association between cash payouts and future earnings growth

is more significant among firms with higher Tobin’s Q as they have more potential agency

costs from free cash flow. The first hypothesis, therefore, is outlined below:

Hypothesis 6.1 Cash payout ratio is positively associated with future earnings growth, but

this relationship diminishes gradually as the cash payout ratio increases; higher cash payout

will improve the earnings growth of firms with more agency cost of free cash flow.

As per the retained earnings hypothesis based on the US Generally Accepted

Accounting Principles (GAAP), bonus shares have the function of signalling future

performance. The GAAP require the firm to reduce the retained earnings account by the

market value of the stock dividend if it is classified as a small stock dividend (no more than

25%). Hence, the decisions about stock dividends indicate management’s confidence in

recovering the retained earnings account from future operating profit (Elgers and Murray,

145

1985, Lakonishok and Lev, 1987, McNichols and Dravid, 1990). Huang et al. (2009) propose

the ‘balanced dividend hypothesis’ that the positive contribution from payout decisions and

future earnings growth is significant among firms announcing a balanced dual dividend.

More specifically, the cash dividend component works as a mechanism to deal with the

agency cost of the free cash flow problem, while the stock dividend component is to signal

management’s confidence in the future performance.

As afore mentioned, there are two types of stock dividends in the Chinese market,

stock dividends from retained earnings (SDRE) and stock dividends from capital reserves

(SDCR). Although neither of them generates any immediate economic benefit to

shareholders, they are quite popular and preferred by Chinese minority shareholders, as

shown in Chapter 5 and other relevant studies (Su, 2005, Cheng et al., 2009). Since the

Chinese accounting principles only require firms to deduct the face value of stock dividends,

the retained earnings hypothesis can hardly explain the 'stock dividend puzzle’ of Chinese

listed firms. On the contrary, because stock dividends do not require any cash balance, it

may facilitate and exacerbate the agency cost of free cash flow if managers avoid

distributing excessive cash holdings to shareholders. Moreover, the stock dividend

announcement might be associated with the managers’ opportunism to benefit from

positive market feedback in cases of inferior performance or poor liquidity (Anderson et al.,

2011). Therefore, the stock dividend announcement is posited to be inversely associated

with future profitability.

Hypothesis 6.2 The decision to issue a stock dividend, from either capital reserves or retained

earnings, will imply inferior future earnings growth.

146

6.3. Data and descriptive statistics

6.3.1. Sample construction

The data used in this research were collected from the China Stock Market and

Accounting Research (CSMAR) database, which includes firms listed on both the Shanghai

and Shenzhen stock exchanges from 2001 to 2010. The observations are clustered into five

groups, which are cash only payers, cash payers with SDCR only, cash payers with SDRE

only, cash payers with both SDCR and SDRE, and others.123

The data filtering criteria are: (i) samples will include all firms except those belonging

to the financial industry, as financial companies have special features in their financial

statements, especially size and leverage; (ii) firms withmissing financial information will not

be included in the sample; (iii) companies which report non positive equity will also be

excluded from the sample; (iv) firms with a total payouts ratio124 larger than one are

omitted. After these filtering criteria are applied, 1,022 firms remain and there are 8,485

observations across the ten year window. Based on the various combinations of dividend

announcements, these observations are divided into six groups, presented in Table 6.1.125

Table 6 1 Sample distribution number of firms (% of total)

Total Cash only Cash with SDCR only Cash with SDRE only Cash with both SDRE and SDCR Others

8,475 4,277 (50.5%) 670 (7.9%) 192 (2.3%) 241 (2.8%) 3,095 (36.5%)

This table provides a summary of sample firms by the type of dividend policy between 2001 and 2010. Thenumber of firms is the total number of observations in a specific year. The number of each category is theobservations which announced specific payout decisions. % is the proportion of each category to the totalnumber of firms.

123 Different from Huang et al. (2009), the dual dividend defined in this chapter contains observations withsimultaneous cash dividends and stock dividends from retained earnings. The combination of third and fourthgroups is the sub sample of dual dividend payers.124 The total payout ratio is defined as the sum of cash dividend and stock dividend from retained earningsscaled by net after tax profit. This criterion reduces the sample size by 581 observations.125 The distribution of various payout decisions underpins the findings of previous studies that non negotiableshareholders prefer cash dividends. This chapter will focus on the first four categories.

147

6.3.2. Dependent variables – proxies for earnings growth

The variables used in this chapter are presented in Table 6.2. Two proxies are

introduced to measure earnings growth. The first proxy is the future growth in earnings per

share (EPS), FUTURE_EPSG, defined as the difference between EPSt+1 and EPSt then scaled

by the absolute value of EPSt.126 EPS is one of the most highlighted financial indicators with

regard to operating performance and is frequently used by the investment industry. Strong

FUTURE_EPSG indicates managers not only achieve efficiency in operation but also

successfully utilize financial leverage to manage the cost of capital and tax payable.

The second parameter is the future growth in earnings before interest and tax (EBIT),

FUTURE_EBITG, defined as the difference between EBITt+1 and EBITt then scaled by the

absolute value of EBITt. Different from EPS, EBIT measures the operating profitability

without the influence of financial decisions. Firms with high EBIT growth exhibit a strong

capability of properly choosing investment projects. Hence, strong FUTURE_EBITG is

expected to be positively associated with a lower agency cost of free cash flow.

6.3.3. Independent variables

The key independent variable is the cash dividend payout ratio, CASH_PAYOUT,

defined as the cash dividend per share scaled by earnings per share. According to the

discussion in the hypothesis development, large cash dividend distribution is expected to

suppress the agency cost of free cash flow and positively influence FUTURE_EPSG and

FUTURE_EBITG. Because it is also conjectured that such positive contribution will decrease

as CASH_PAYOUT increases, the squared cash payout, CASH_PAYOUT2, will be included in

126 The absolute value is to cope with firms which reported negative earnings in the previous year.

148

the regression model. Additionally, the change in cash payout ratio, CASH_PAYOUT, is

introduced as a supplementary measurement of cash dividend policy because the level of

cash payout ratio may not be stationary. The coefficients on CASH_PAYOUT and

CASH_PAYOUT are anticipated to be positive, while CASH_PAYOUT2 is expected to exhibit

a negative coefficient. According to the China Securities Law, dividend payout, including

both cash dividends and stock dividends from retained earnings, should not exceed the net

profit of the current financial year. Therefore, stock dividend payout is proxied by two

variables. One is a dummy variable, SDRE, assigned to label the presence of a stock dividend

from retained earnings; the other is the stock payout ratio, STOCK_PAYOUT, measured by

the stock dividend per share scaled by earnings per share. In contrast, stock dividends from

capital reserves are irrelevant to the current net profit, so their existence is only

represented by a dummy variable, SDCR. According toHypothesis 6.2, coefficients on SDRE,

STOCK_PAYOUT and SDCR are anticipated to be negative.

Several control variables are introduced. First is the Tobin’s Q, TOBIN’S_Q, which is a

measurement of the over investment problem.127 Higher TOBIN’S_Q implies a higher

agency cost of free cash flow, so it is expected to show a negative sign. Firm size, SIZE,

defined as the logarithm of total assets as of year t, is expected to have a positive influence

on the probability and magnitude of dividends, that is, larger firms are inclined to pay more

dividends (DeAngelo et al., 2004). In the meantime, the expected profitability,

EXPECTED_ROA, defined as the return on assets as of the first quarter of year t+1, is

included as a control for firms’ awareness of future earnings, which may affect their payout

127 TOBIN’S_Q is derived from the Financial Ratio Database, a module in the CSMAR database.

149

decision.128 Future asset growth, FUTURE_AG, is introduced as a control with an expected

positive sign on the coefficient, because the effective management of more assets will lead

to economies of scale and improved profitability. Dividend yield, DIVYIELD, is included as

the control for clientele effect, because firms with a high dividend yield may well be

neglected by sophisticated investors who prefer growth opportunity (Graham and Kumar,

2006). Lack of sophisticated investors means the firmmay be exposed to more agency cost,

so the coefficient on DIVYIELD is anticipated to be negative. Finally, potential mean

reversion in earnings growth is taken into account, and current EPS or EBIT growth,

CURRENT_EPSG or CURRUNT_EBITG, is included in the regressionmodel (Zhou and Ruland,

2006, Huang et al., 2009). Table 6.2 contains definitions of the variables.

Table 6 2 Variable definitions

FUTURE_EPSG the future earnings growth rate which is (EPSt+1 EPSt)/|EPSt|; EPSt+1 and EPSt are earnings per share as of year t+1 and year t

FUTURE_EBITG the future EBIT growth rate which is (EBITt+1 EBITt)/|EBITt|; EBITt+1 and EBITt are earnings before interest and tax as of year t+1 andyear t

CASH_PAYOOUT cash payout ratio which is cash dividend per share scaled by earnings per share

CASH_PAYOOUT change in cash payout ratio between year t and year t 1

STOCK_PAYOUT stock dividend payout ratio which is stock dividend from retained earnings per share scaled by earnings per share

SIZE the firm size which is the logarithm of the total assets as of year t

EXPECTED_ROA expected return on assets which is return on assets as of the first quarter of year t+1

DIVYIELD cash dividend yields which is cash dividends per share scaled by closing share price as of year t

TOBIN’S_Q Tobin’s Q, which is the market value of the firm (including non negotiable shares) scaled by the book value of the firm

CURRENT_EPSG the current earnings growth which is the lagged FUTURE_EGt

CURRENT_EBITG the current EBIT growth which is the lagged FUTURE_EBITGt

FUTURE_AG future asset growth which is total assets as of year t+1 scaled by total assets as of year t

SDCR dummy variable which equals 1 when there is stock dividend from capital reserves in year t, and 0 otherwise

SDRE dummy variable which equals 1 when there is stock dividend from retained earnings in year t, and 0 otherwise

Year Dummy129 a set of dummy variables indicating the financial years from 2001 to 2009

128 Chinese listed firms have to disclose the annual reports of the previous year before April, which issynchronically the same as the release of the first quarter’s financial reports.129 Industry dummy variables are not introduced in the research design, as a fixed effect model will be usedas the main methodology, except when CASH_PAYOOUT is taken as the key independent variable.

150

6.3.4. Descriptive statistics

Table 6.3 reports the FUTURE_EPSG (FUTURE_EBITG) by various combinations of

payout decisions. As per Panel A, the cash only payers achieve best mean FUTURE_EPSG

(12.90%) and cash payers with both SDCR and SDRE show the worst mean FUTURE_EPSG (

23.96%), while the other two groups also report negative mean FUTURE_EPSG. The t stats

and p values suggest these mean statistics are significantly different from zero. The

preliminary result indicates observations with stock dividends are associated with worse

future EPS performance.

Table 6 3 Summary statistics of FUTURE_EPSG and FUTURE_EBITG by payout decision

Panel A: FUTURE_EPSG N Mean St. Err. t test p value

Cash only 4,277 0.1290 0.0121 10.663 0.0000Cash with SDCR only 670 0.1701 0.0221 9.5701 1.0000

Cash with SDRE only 192 0.1388 0.0349 3.9787 1.0000

Cash with SDRE and SDCR 241 0.2396 0.0272 8.8021 1.0000

Panel B: FUTURE_EBITG N Mean St. Err. t test p value

Cash only 4,277 0.1952 0.0093 20.999 0.0000Cash with SDCR only 670 0.2574 0.0222 11.624 0.0000

Cash with SDRE only 192 0.1212 0.0374 3.2409 0.0007

Cash with SDRE and SDCR 241 0.2664 0.0406 6.5599 0.0000

This table provides the summary of FUTURE_EPSG and FUTURE_EBITG by payout category. Refer to Table 6.2for variable definitions. Each group is winsorized at the top and bottom 1%. One tail t test and p value areincluded to test whether the mean is larger than zero.

The statistics of FUTURE_EBITG are a bit different from those of FUTURE_EPSG. Cash

payers with simultaneous stock dividends exhibit significant divergence between mean

FUTURE_EBITG and mean FUTURE_EPSG. Specifically, cash payers with SDCR have a far

better mean FUTURE_EBITG, but cash payers with SDRE only report worse means

FUTURE_EBITG. These outcomes indicate that stock dividend payers’ net profits are more

influenced by financial cost and extraordinary one off items, even if their operating

performance is robust.

151

Table 6.4 presents the summary statistics of dependent and independent variables,

with regard to the cash only sample, dual dividend payer sample, cash payer with SDCR

sample and payer sample, respectively. According to the first three panels, cash only payers

report the highest mean CASH_PAYOUT, about 40 per cent, and the largest

CASH_PAYOUT. In the meantime, dual dividend payers’ total payout ratio is dominated by

the stock dividend portion, which is more than twice that of the cash dividend (0.4061 vs.

0.1645). Both dual dividend payers and cash dividend payers with SDCR exhibit negative

CASH_PAYOUT. Furthermore, cash only payers report the lowest mean EXPECTED_ROA,

TOBIN’S_Q and FUTURE_AG, but also the highestmeanDIVYIELD, which suggests they have

less growth opportunity.

Table 6.5 presents the Pearson’s correlationmatrix. Consistent with expectation, both

CASH_PAYOUT and EXPECTED_ROA are positively correlated with FUTURE_EPSG and

FUTURE_EBITG, in all three panels. Instead, TOBIN’S_Q reports a significant negative

correlation coefficient with FUTURE_EPSG in Panel C, but no significant correlation with

earnings growth measurements in the other two panels, which suggest higher growth

opportunity is not necessarily associatedwith resilient future earnings growth. Finally, there

is no significant correlation between FUTURE_EPSG (FUTURE_EBITG) and CURRENT_EPSG

(CURRENT_EBITG).

152

Table 6 4 Descriptive statistics

Panel A: Cash only payer sample

N = 4,277 Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.FUTURE_EPSG 0.1290 0.7915 0.5074 0.2000 0.0822 0.3582 0.8667FUTURE_EBITG 0.1952 0.6078 0.3889 0.1077 0.1400 0.4117 0.7991

CASH_PAYOUT 0.4073 0.2161 0.1477 0.2353 0.3781 0.5443 0.7330

CASH_PAYOUT 0.0333 0.2680 0.2353 0.0928 0.0057 0.1470 0.3680

SIZE 21.8190 1.1432 20.5352 20.9981 21.6422 22.4712 23.3254

EXPECTED_ROA 0.0124 0.0119 0.0013 0.0045 0.0096 0.0176 0.0284

DIVYIELD 0.0151 0.0116 0.0038 0.0065 0.0116 0.0203 0.0316

TOBIN’S_Q 2.0505 1.1862 1.0345 1.2439 1.6859 2.4480 3.5039

CURRENT_EPSG 0.3054 1.1440 0.4444 0.2000 0.0588 0.3573 1.0567

CURRENT_EBITG 0.3557 0.9215 0.2700 0.0533 0.1614 0.4396 0.9671

FUTURE_AG 0.3192 0.4097 0.0144 0.0583 0.1638 0.3915 1.0798

Panel B: Dual dividend payers sample

N = 433 Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.FUTURE_EPSG 0.1973 0.4394 0.7000 0.4595 0.2500 0.0057 0.3247FUTURE_EBITG 0.1990 0.5712 0.3859 0.1011 0.1512 0.4378 0.7316

CASH_PAYOUT 0.1645 0.1202 0.0495 0.0782 0.1301 0.2212 0.3422

CASH_PAYOUT 0.0923 0.2214 0.4065 0.1919 0.0575 0.0545 0.1354

STOCK_PAYOUT 0.4061 0.1887 0.1675 0.2636 0.3966 0.5332 0.6825

SIZE 21.9081 1.0625 20.5354 21.0702 21.8215 22.6755 23.4479

EXPECTED_ROA 0.0198 0.0178 0.0034 0.0079 0.0152 0.0260 0.0408

DIVYIELD 0.0068 0.0071 0.0013 0.0024 0.0047 0.0086 0.0146

TOBIN’S_Q 2.9526 2.1180 1.2301 1.5711 2.2741 3.4306 5.7565

CURRENT_EPSG 0.7623 2.1760 0.3132 0.0886 0.1769 0.5993 1.6807

CURRENT_EBITG 0.8388 1.8178 0.0628 0.1409 0.3643 0.7497 1.7522

FUTURE_AG 0.4223 0.4643 0.0038 0.1039 0.2452 0.5732 1.2017

Panel C: Cash payer with SDCR only sample

N = 670 Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.FUTURE_EPSG 0.1701 0.4601 0.6372 0.4444 0.2000 0.0000 0.3433FUTURE_EBITG 0.2574 0.5732 0.3335 0.0330 0.1820 0.4921 0.8497

CASH_PAYOUT 0.3441 0.2112 0.1059 0.1697 0.2901 0.4731 0.6652

CASH_PAYOUT 0.0169 0.2769 0.2879 0.1305 0.0186 0.0905 0.2958

SIZE 21.6664 1.1202 20.3113 20.8578 21.5090 22.3239 23.1747

EXPECTED_ROA 0.0150 0.0126 0.0023 0.0059 0.0130 0.0201 0.0301

DIVYIELD 0.0125 0.0106 0.0027 0.0046 0.0091 0.0170 0.0271

TOBIN’S_Q 2.6425 1.7212 1.1443 1.4155 2.1321 3.2739 4.8258

CURRENT_EPSG 0.3784 1.3189 0.4000 0.1731 0.0956 0.4706 1.1156

CURRENT_EBITG 0.5147 1.0466 0.1493 0.0582 0.2991 0.6186 1.1455

FUTURE_AG 0.4222 0.4606 0.0281 0.1015 0.2289 0.5954 1.2126

Panel D: Cash payer sample

N = 5,380 Mean St. Dev. 10th Pctl. 25th Pctl. 50th Pctl. 75th Pctl. 90th Pctl.FUTURE_EPSG 0.0637 0.7072 0.5779 0.2884 0.0237 0.3077 0.7694FUTURE_EBITG 0.2036 0.5966 0.3859 0.1009 0.1464 0.4247 0.8017

CASH_PAYOUT 0.3799 0.2198 0.1177 0.2046 0.3465 0.5184 0.7130

CASH_PAYOUT 0.0170 0.2670 0.2557 0.1035 0.0002 0.1292 0.3418

SIZE 21.8055 1.1295 20.5090 20.9843 21.6418 22.4716 23.3107

EXPECTED_ROA 0.0133 0.0126 0.0016 0.0048 0.0103 0.0188 0.0297

DIVYIELD 0.0141 0.0114 0.0032 0.0057 0.0107 0.0190 0.0301

TOBIN’S_Q 2.1931 1.3628 1.0516 1.2832 1.7615 2.6160 3.8454

CURRENT_EPSG 0.3503 1.2602 0.4286 0.1927 0.0712 0.3916 1.1056

CURRENT_EBITG 0.4140 1.0242 0.2443 0.0227 0.1885 0.4906 1.0550

FUTURE_AG 0.3403 0.4229 0.0089 0.0650 0.1760 0.4293 1.1027

This table contains descriptive statistics for variables of interest that are used in the following analysis of the1,022 firms between 2001 and 2010. Refer to Table 6.2 for variable definitions.

153

Table 6 5 Pearson's correlation matrix

Panel A: Cash only payers FUTURE_EPSG FUTURE_EBITG CASH_PAYOUT CASH_PAYOUT SIZE EXPECTED_ROA DIVYIELD TOBIN’S_Q CURRENT_EPSG CURRENT_EBITG

FUTURE_EBITG 0.8474*

CASH_PAYOUT 0.0403* 0.0494*

CASH_PAYOUT 0.0410* 0.0261* 0.3154*

SIZE 0.0179 0.0106 0.2065* 0.0522*

EXPECTED_ROA 0.1748* 0.1660* 0.0418* 0.0816* 0.0821*

DIVYIELD 0.0178 0.0244 0.4511* 0.0740* 0.1482* 0.2064*

TOBIN’S_Q 0.0201 0.0227 0.0325 0.0070 0.2844* 0.3621* 0.3347*

CURRENT_EPSG 0.0245 0.0208 0.1109* 0.0200 0.0267 0.1146* 0.0453* 0.0634*

CURRENT_EBITG 0.0049 0.0159 0.1253* 0.0163 0.0564* 0.1337* 0.0227 0.0690* 0.7448*

FUTURE_AG 0.0686* 0.1478* 0.2010* 0.0477* 0.1756* 0.1088* 0.1362* 0.1627* 0.0721* 0.0761*

Panel B: Cash payers FUTURE_EPSG FUTURE_EBITG CASH_PAYOUT CASH_PAYOUT SIZE EXPECTED_ROA DIVYIELD TOBIN’S_Q CURRENT_EPSG CURRENT_EBITG

FUTURE_EBITG 0.8193*

CASH_PAYOUT 0.0878* 0.0488*

CASH_PAYOUT 0.0592* 0.0254 0.3313*

SIZE 0.0237 0.0198 0.1945* 0.0336

EXPECTED_ROA 0.1520* 0.1764* 0.0766* 0.0877* 0.0609*

DIVYIELD 0.0181 0.0221 0.5008* 0.1087* 0.1355* 0.1478*

TOBIN’S_Q 0.0539* 0.0135 0.0346 0.0383* 0.2842* 0.3989* 0.3467*

CURRENT_EPSG 0.0020 0.0137 0.1255* 0.0349 0.0300 0.1101* 0.0590* 0.0624*

CURRENT_EBITG 0.0246 0.0130 0.1496* 0.0264 0.0548* 0.1401* 0.0533* 0.0936* 0.7533*

FUTURE_AG 0.0520* 0.1546* 0.2009* 0.0545* 0.1618* 0.1170* 0.1592* 0.1921* 0.0666* 0.0822*

This table contains Pearson’s correlation coefficients matrix for selected variables used in this analysis for the sample between 2001 and 2010. Refer to Table 6.2 forvariable definitions. * denotes significance at 1% level.

154

6.4. Empirical Analysis

6.4.1. Univariate analysis

This section investigates whether the earnings growth is affected by firm specific

features. Following Fama and French (1993), the cash payer (cash only payer) sample is

sorted in descending order of individual independent variables, and the top (bottom) three

deciles are termed ‘high’ (‘low’) sub groups.130 Tests are conducted for statistically

significant differences in the mean FUTURE_EPSG and FUTURE_EBITG across these sub

groups, which are reported in Table 6.6.

Panel A of Table 6.6 suggests several appealing results for FUTURE_EPSG. Within the

cash only payer sample, first, the high CASH_PAYOUT sub group has a higher mean

FUTURE_EPSG (0.1601) than that of the low CASH_PAYOUT sub group (0.0728), and the

difference in the mean of FUTURE_EPSG between the high and low sub groups is

statistically significant at the 1% level. Further, the higher CASH_PAYOUT sub group also

reports a higher mean FUTURE_EPSG (0.1940) than that of the low CASH_PAYOUT sub

group (0.1022), and their difference is also significant at the 1% level. Second, expected

profitability does provide some guidance to the future earnings growth, as the difference

between the high and low EXPECTED_ROA sub groups is statistically significant at the 1%

level. Third, firms with larger SIZE or faster FUTURE_AG are inclined to have better

FUTURE_EPSG. More importantly, the high TOBIN’S_Q sub group reports a worse

130 For cash only payer sample, the top and bottom 30% contains 1,283 observations. For all payer sample,the top and bottom 30% contains 1,614 observations. It was considered that the median of each independentvariable is used as the hurdle to divide the sample into two halves, but the key independent variable,CASH_PAYOUT, is restrained between 0 and 1. As per the summary statistics shown in Table 6.4 Panel A, thethird quartile of CASH_PAYOUT is slightly above 50% and the difference between the first quartile and thethird quartile is only 30%. To divide the sample by the median may cause the genuine high and low payoutsto be diluted by the observations around the median.

155

FUTURE_EPSG. The difference between the means of high and low TOBIN’S_Q sub groups

is statistically significant at the 1% level. It is in line with the expectation that the future

earnings growth is undermined by high TOBIN’S_Q because it reflects a greater agency cost

of free cash flow.

The means of the cash payer sample show a little difference. The positive difference

between high and low CASH_PAYOUT ( CASH_PAYOUT) sub groups enlarges to 0.1519,

compared with 0.0873 in the cash only payer sample. Meanwhile, the negative difference

between the high and low TOBIN’S_Q sub groups widens from 0.0830 to 0.1123. These

findings are consistent with the summary statistics shown in Table 6.4 in which the stock

dividend payers have inferior future earnings growth. In contrast, the differences in the

means of high and low SIZE, EXPECTED_ROA and FUTURE_AG maintain the same sign as

reported in cash only payer sample but shrink in the magnitude.

Panel B of Table 6.6 reports the comparison of mean FUTURE_EBITG between high

and low sub groups. Similar to Panel A, the differences in the mean of the high and low

CASH_PAYOUT subgroups, for both samples, is statistically significant at the 1% level, but

the gap between the differences of the two samples (0.0766 vs. 0.0708) is not as huge as

has been reported in Panel A. In general, the comparison results are similar to Panel A, but

the gap between the differences of the high and low sub groups is much smaller than has

been reported in Panel A. It is consistent with the expectation that EBIT is more stable and

less influenced by one off extraordinary items than EPS.

156

Table 6 6 Mean FUTURE_EPSG (FUTURE_EBITG) comparison

Mean of top 3 deciles Mean of bottom 3declies Difference in mean t statistics

(Mann Whitney) Mean of top 3 deciles Mean of bottom 3declies Difference in mean t statistics

(Mann Whitney)Cash only payer Cash payer

Panel A – Future EPS growth (FUTURE_EPSG)CASH_PAYOUT 0.1601 0.0728 0.0873 2.70*** 0.1303 0.0216 0.1519 5.98***

(2.65)*** (7.70)***

CASH_PAYOUT 0.1940 0.1022 0.0918 2.78*** 0.1414 0.0241 0.1172 4.44***

(2.76)*** (5.13)**

SIZE 0.1582 0.0862 0.0720 2.37** 0.0920 0.0247 0.0673 2.76***

(2.73)*** (2.57)**

EXPECTED_ROA 0.2769 0.0306 0.3075 9.13*** 0.1772 0.0674 0.2446 9.26***

(11.58)*** (10.72)***

DIVYIELD 0.1140 0.1583 0.0443 1.34 0.0814 0.0328 0.0486 1.90*

(0.11) (5.36)***

TOBIN’S_Q 0.0923 0.1753 0.0830 2.59*** 0.0001 0.1122 0.1123 4.45***

( 2.30)** ( 5.31)***

CURRENT_EPSG 0.1634 0.1485 0.0149 0.43 0.0654 0.0890 0.0237 0.87(1.46) ( 0.89)

FUTURE_AG 0.2182 0.0608 0.2790 8.75*** 0.1295 0.0883 0.2178 8.50***

(10.07)*** (9.21)***

Panel B – Future EBIT growth (FUTURE_EBITG)CASH_PAYOUT 0.2269 0.1503 0.0766 3.11*** 0.2411 0.1703 0.0708 3.28***

(3.28)*** (3.14)***

CASH_PAYOUT 0.2420 0.1752 0.0668 2.67*** 0.2522 0.1833 0.0689 3.11***

(2.13)** (2.55)**

SIZE 0.2085 0.1738 0.0348 1.47 0.2219 0.1790 0.0429 2.07*

(1.76)* (2.46)**

EXPECTED_ROA 0.3037 0.0828 0.2209 8.75*** 0.3205 0.0926 0.2280 10.28***

(10.07)*** (11.42)***

DIVYIELD 0.1760 0.2216 0.0456 1.83* 0.1898 0.2208 0.0310 1.45( 0.15) (0.10)

TOBIN’S_Q 0.1663 0.2200 0.0536 2.18** 0.1763 0.2156 0.0392 1.86*

( 2.41)** ( 2.18)**

EBG 0.1333 0.1302 0.0030 0.09 0.2105 0.1945 0.0160 0.70(1.28) (1.77)*

FUTURE_AG 0.3311 0.0003 0.3314 13.44*** 0.3438 0.0062 0.3376 15.62***

(14.70)*** (16.91)***

This table reports univariate analysis for different subgroups. Panel A presents the analysis results for FUTURE_EPSG and Panel B presents the analysis for FUTURE_EBITG.Refer to Table 6.2 for variable definitions. The t statistics for differences in mean assume unequal variance. The Mann Whitney test is reported in parentheses. *, ** and*** denote significance at 10%, 5% and 1% levels, respectively.

157

6.4.2. Multivariate analysis

6.4.2.1 Future earnings growth and cash dividend payout

The cash only payer sample will be used to testHypothesis 6.1. The fixed effect model

is introduced as the sample contains both cross sectional and time series effects. Following

Zhou and Ruland (2006) and Huang et al. (2009)131 with some amendments, themultivariate

models for Hypothesis 6.1 can be expressed as follows:

Where Future Earnings Growth refers either to FUTURE_EPSG or FUTURE_EBITG;

Cash Payout Measurement refers to CASH_PAYOUT and CASH_PAYOUT. Eq.6.1 is the

model which conjectures there is a potential non linear relationship between earnings

growth and cash dividend distribution. When the earnings measurement is proxied by

CASH_PAYOUT, Eq.6.1 is estimated using an ordinary least squares (OLS) model adjusted

for heteroscedasticity (White, 1980); otherwise, Eq.6.1 is estimated using a fixed effect

model adjusted for heteroscedasticity (White, 1980). Table 6.7 reports the estimation

results of Eq.6.1 for FUTURE_EPSG and FUTURE_EBITG.132

131 Both papers use Fama Macbeth (1973) two step procedure, which will be applied in the robustness checkof this research. Other econometrics models are included and compared in the robustness check.132 Columns (1) to (4) of each panel examine the influence of CASH_PAYOUT and CASH_PAYOUT on futureearnings growth, while columns (5) and (6) investigate the potential non linear relationship betweenCASH_PAYOUT and future earnings growth.

158

Table 6 7 Cash dividend and future earnings growth cash only payer sample

Panel A: FUTURE_EPSG

Expected signs (1) p value (2) p value (3) p value (4) p value (5) p value (6) p value

Intercept 0.098** (0.012) 2.629*** (0.008) 0.125*** (0.000) 0.160 (0.575) 0.228*** (0.001) 1.951* (0.050)

CASH_PAYOUT + 0.557*** (0.000) 1.330*** (0.000) 1.264*** (0.000) 2.806*** (0.000)

CASH_PAYOUT2 0.740** (0.030) 1.480*** (0.000)

CASH_PAYOUT + 0.121** (0.030) 0.216*** (0.000)

SIZE 0.139*** (0.001) 0.014 (0.265) 0.121*** (0.005)

EXPECTED_ROA + 23.761*** (0.000) 15.623*** (0.000) 24.208*** (0.000)

DIVYIELD 28.834*** (0.000) 10.596*** (0.000) 30.214*** (0.000)

TOBIN’S_Q 0.145*** (0.000) 0.084*** (0.000) 0.150*** (0.000)

CURRENT_EPSG 0.008 (0.553) 0.001 (0.914) 0.004 (0.761)

FUTURE_AG + 0.354*** (0.000) 0.492*** (0.000) 0.364*** (0.000)

Year Dummies No Yes No Yes No Yes

Adj. R2 0.014 0.185 0.001 0.133 0.016 0.192

F test 34.252 28.862 4.702 33.873 20.775 28.591

Panel B: FUTURE_EBITG

Expected signs (1) p value (2) p value (3) p value (4) p value (5) p value (6) p value

Intercept 0.019 (0.525) 0.045 (0.954) 0.193*** (0.000) 0.262 (0.203) 0.149** (0.004) 0.520 (0.507)

CASH_PAYOUT + 0.525*** (0.000) 1.113*** (0.000) 1.227*** (0.000) 2.419*** (0.000)

CASH_PAYOUT2 0.735*** (0.005) 1.309*** (0.000)

CASH_PAYOUT + 0.059 (0.152) 0.153*** (0.000)

SIZE 0.029 (0.392) 0.014 (0.118) 0.014 (0.677)

EXPECTED_ROA + 19.028*** (0.000) 11.357*** (0.000) 19.398*** (0.000)

DIVYIELD 22.779*** (0.000) 8.726*** (0.000) 24.014*** (0.000)

TOBIN’S_Q 0.142*** (0.000) 0.073*** (0.000) 0.147*** (0.000)

CURRENT_EBITG 0.022* (0.071) 0.025** (0.025) 0.018 (0.143)

FUTURE_AG + 0.595*** (0.000) 0.701*** (0.000) 0.605*** (0.000)

Year Dummies No Yes No Yes No Yes

Adj. R2 0.021 0.237 0 0.189 0.023 0.246

F test 53.089 38.353 2.053 44.217 33.721 39.566

This table reports the estimation results of Eq.6.1 for FUTURE_EPSG and FUTURE_EBITG. Equations with CASH_PAYOUT (CASH_PAYOUT2)are estimated using a fixedeffect model adjusted for heteroscedasticity (White, 1980), while equations with CASH_PAYOUT are estimated using an ordinary least squares (OLS) model adjustedfor heteroscedasticity (White, 1980).Panel A contains the results for FUTURE_EPSG while Panel B presents the results for FUTURE_EBITG. Refer to Table 6.2 for variabledefinitions. p values are reported in parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

159

As per columns (1) and (2), the coefficients on CASH_PAYOUT are positive and

significant at 1%. With firm level control, a 10% increase in cash payout ratio leads to an

expected 13.3% increase in FUTURE_EPSG. At the same time, columns (3) and (4) suggest a

moderate non linear relationship between cash payout and future earnings growth. Not

only CASH_PAYOUT, but also CASH_PAYOUT has a positive impact on FUTURE_EPSG, as

shown in rows (5) and (6), because the coefficient on CASH_PAYOUT is positive and

statistically significant at the 1% level. Further, according to rows (5) and (6), the coefficients

on CASH_PAYOUT2 are significantly negative, while the coefficients on CASH_PAYOUT

maintain a positive sign with strong significance. 133 The findings of Panel A are underpinned

by Panel B with FUTURE_EBITG as an earnings growth measurement, in which all

coefficients on key independent variables show the same signs and significance levels.

Consistent with recent studies on developed and emerging markets (Arnott and

Asness, 2003, Gwilym et al., 2006, Zhou and Ruland, 2006, Huang et al., 2009, Vermeulen

and Smit, 2011), the above results support Hypothesis 6.1 and suggest cash payout ratio

has a positive contribution to improve future earnings growth. However, as the cash payout

ratio increases, the contribution to future earnings growth declines. Therefore, the cash

dividend’s disciplinary power on the agency cost of free cash flow will be overtaken by the

concern about future financial costs.

133 Although the estimation results of quadratic equations are statistically significant, the quadratic model isrestricted by the payout practice in the Chinese market. As per column (3), the parabolic curve of futureearnings growth and cash payout peaks when the first order partial differential equation, FUTURE_EPSG’=2 ×( 0.740) × CASH_PAYOUT + 1.264, is equal to zero. It is equivalent to 88% (1.264/(2×0.740)) cash payout ratio.According to the summary statistics in Table 6.4 Panel A, the mean and median of cash payout ratio is lessthan 50%. There are only 125 observations, or 2.9% of the cash only payer sample, which report a cash payoutratio higher than 88%. Therefore, a linear version of Eq.6.1 is able to proxy the relationship between currentcash payout ratio and future earnings growth.

160

With regard to firm specific factors, the coefficient on EXPECTED_ROA is positive and

statistically significant at the 1% level in all estimation results. In line with expectation, the

nearly simultaneous announcements of the previous year’s dividend decision (financial

statements) and first quarter financial results make it possible that controlling shareholders

decide the cash distribution based on the operating performance of the following year. If

the financial performance of the first quarter is optimistic, they may well distribute more

cash dividends from the net after tax profits as of the previous financial year. In contrast,

the coefficients on DIVYIELD are significantly negative in all regression specifications, which

mean a low dividend yield is associated with superior future earnings growth. As

mentioned, the sophisticated investors exert some monitoring function on the firm’s

operation, but they prefer growth stocks. A lower dividend yield implies more sophisticated

investors are involved and they mitigate the agency cost by their superior knowledge of

corporate governance issues.134 Similarly, the coefficients on TOBIN’S_Q, the indicator of

over investment problems, are also negative and statistically significant at 1% level, which

indicates that future earnings growth is undermined by the agency cost of free cash flow.

The coefficients on SIZE are either negative or statistically insignificant, but FUTURE_AG

reports significantly positive coefficients in all regression results. The outcomes indicate

that the earnings growth benefits as the pace of growth in firm size increases, but not the

outstanding large asset base. Finally, there is no momentum effect or mean reversion in

earnings growth, as CURRENT_EPSG (CURRENT_EBITG) has no significant contribution to

FUTURE_EPSG (FUTURE EBTIT).

134 The large coefficients, but in opposite directions, on the EXPECTED_ROA and DIVYIELD raise the issue ofmulticollinearity, which will be investigated in the robustness check.

161

Based on the results in Table 6.7, two independent variables, DIVYIELD and

TOBIN’S_Q, are selected as the proxy for agency cost. Following Fama and French (1993)

and the univariate analysis, the cash only payer sample is divided into three sub samples

by the third and seventh decile of DIVYIELD and TOBIN’S_Q, respectively. Eq.6.1, without

squared cash payout measurement, will be estimated in the top 30% (high) sub sample and

bottom 30% (low) sub sample. Because the subdivision of the sample reduces the time

series effect among observations, Eq.6.1 is estimated using the Fama Macbeth (1973) two

stage procedure. The estimation results are presented in Table 6.8.

Columns (1) and (2) of Table 6.8 Panel A present a comparison of regression results

on FUTURE_EPSG between High TOBIN’S_Q and Low TOBIN’S_Q sub samples. As per

column (1), the coefficient on cash payout ratio is positive and statistically significant at the

1% level. More than half of the independent variables exhibit significant coefficients that

indicate the regression result is robust. In contrast, most explanatory variables in column

(2), including CASH_PAYOUT, display statistically insignificant coefficients. These results

support Hypothesis 6.1 that the contribution from cash dividends on future EPS growth is

more significant when the over investment problem is greater. But columns (3) and (4) do

not supply any difference between high/lowDIVYIELD sub samples, because DIVYIELD is an

indirect indicator of agency cost which relies on the involvement of sophisticated investors,

while TOBIN’S_Q is a direct indicator of the market opinion on the firms’ investment

opportunities and valuation.

162

Table 6 8 Future earnings growth and over investment problem

Panel A FUTURE_EPSG(1)High

TOBIN’S_Qp value

(2)Low

TOBIN’S_Qp value (3)

High DIVYIELD p value(4)Low

DIVYIELDp value

Intercept 1.023* (0.078) 10.109 (0.272) 0.474 (0.418) 0.190 (0.893)CASH_PAYOUT 0.932*** (0.000) 0.242 (0.877) 0.461** (0.047) 1.133*** (0.005)

SIZE 0.054** (0.029) 0.412 (0.240) 0.021 (0.417) 0.012 (0.850)

EXPECTED_ROA 15.985*** (0.000) 80.492* (0.071) 18.167*** (0.000) 21.676** (0.022)

DIVYIELD 27.833*** (0.000) 92.731 (0.478) 8.860*** (0.009) 62.236** (0.027)

TOBIN’S_Q 0.108 (0.130) 0.044 (0.944) 0.216*** (0.005) 0.161** (0.049)

CURRENT_EPSG 0.035 (0.346) 0.696 (0.513) 0.031 (0.366) 0.094 (0.163)

FUTURE_AG 0.672*** (0.003) 2.071 (0.530) 0.420*** (0.005) 0.757** (0.010)

N 1,283 1,283 1,283 1,283

R2 0.261 0.336 0.263 0.231

F test 52.997 20.712 783.91 19.4

Panel B FUTURE_EBITG(1)High

TOBIN’S_Qp value

(2)Low

TOBIN’S_Qp value (3)

High DIVYIELD p value (4)Low DIVYIELD p value

Intercept 1.108** (0.043) 5.353 (0.358) 0.363 (0.444) 0.007 (0.995)CASH_PAYOUT 0.862*** (0.000) 0.750*** (0.001) 0.554*** (0.001) 1.031*** (0.000)

SIZE 0.052** (0.012) 0.091 (0.325) 0.016 (0.451) 0.000 (0.996)

EXPECTED_ROA 13.955*** (0.001) 33.485** (0.015) 15.625*** (0.000) 20.596*** (0.001)

DIVYIELD 29.892*** (0.000) 39.536* (0.087) 10.669*** (0.000) 41.801** (0.028)

TOBIN’S_Q 0.114* (0.067) 2.677 (0.405) 0.225*** (0.000) 0.151*** (0.003)

CURRENT_EBITG 0.003 (0.937) 0.572 (0.224) 0.027 (0.378) 0.034 (0.529)

FUTURE_AG 0.899*** (0.000) 0.292 (0.812) 0.684*** (0.000) 0.820*** (0.001)

N 1,283 1,283 1,283 1,283

R2 0.34 0.359 0.304 0.261

F test 59.511 54.24 37.782 25.429

This table reports the estimation results of Eq.6.1 (linear version) for FUTURE_EPSG and FUTURE_EBITG, usingthe Fama Macbeth (1973) two step procedure. The sample is sorted by either TOBIN’S_Q or DIVYIELD,respectively, and the top 30% is defined as High, while the bottom 30% is defined as Low. Panel A containsthe results for the comparison between High and Low TOBIN’S_Q, while Panel B contains the results for thecomparison between High and Low DIVYIELD. The first two rows of each panel present the estimation resultsfor FUTURE_EPSG, while the next two rows present the estimation results for FUTURE_EBITG. The odd rowincludes the regression results of the High sub sample, while the even row reports the regression results ofthe Low sub sample. Refer to Table 6.2 for variable definitions. p values are reported in parentheses. *, ** and*** denote significance at 10%, 5% and 1% levels, respectively.

The estimation results with FUTURE_EBITG as the dependent variable, as shown in

Panel B of Table 6.8, report a similar outcome. Although the coefficients on CASH_PAYOUT

in both columns (1) and (2) of Panel B are statistically significant, only three out of seven

explanatory variables in row (4) are significant, which means this result from the

Fama_French two step procedure is not reliable. Similar to Panel A, both estimation results

on the High and Low DIYYIELD sub samples display significant positive coefficients on

CASH_PAYOUT. In a nutshell, this section provides supportive evidence to Hypothesis 6.1

163

that cash dividend payouts reduce the agency cost of free cash flow and, consequently,

improve the future operating performance. This association is more significant among the

firms with a higher potential agency cost.

6.4.2.2 Future earnings growth and stock dividend decision

This section will test Hypothesis 6.2, the association between the stock dividend

announcement and future earnings growth, based on the sample of all cash payers. Model

specification, Eq.6.2, is based on Eq.6.1 and incorporates STOCK_PAYOUT and dummy

variables SDCR and SDRE which indicate the existence of stock dividends from retained

earnings and stock dividends from capital reserve.

Where future earnings growth refers to either FUTURE_EPSG or FUTURE_EBITG; payout

measurement refers to CASH_PAYOUT, CASH_PAYOUT and STOCK_PAYOUT. Eq.6.2 is

estimated using a fixed effect model adjusted for heteroscedasticity (White, 1980), and the

results are reported in Table 6.9.

As per panel A which reports the estimation results on FUTURE_EPSG, columns (2)

indicates the coefficient on CASH_PAYOUT maintains a positive sign and statistical

significance. Every 10% increase in CASH_PAYOUT will lead to an 11.95% increase in the

future EPS growth, and the result is significant at the 1% level. But both coefficients on SDCR

and SDRE are significantly negative. With firm specific controls, the presence of a SDRE

(SDCR) is expected to result in a decrease in future EPS growth by 20.2% (27.2%), consistent

164

with the anticipation that stock dividends, from either source, undermine the future

earnings growth. When stock dividend from retained earnings is measured by

STOCK_PAYOUT, as per column (3), its coefficient is negative and statistically significant at

1% level. Finally, when CASH_PAYOUT is substituted by CASH_PAYOUT, as shown in

columns (4) and (5) of Panel A, the coefficients on SDCR and SDRE are still significantly

negative, while the coefficient on CASH_PAYOUTmaintains significantly positive.

Table 6 9 Future earnings growth and stock dividend decision

Panel A FUTURE_EPSG

(1) p value (2) p value (3) p value (4) p value (5) p value

Intercept 0.077*** (0.009) 2.874*** (0.000) 2.914*** (0.000) 0.119*** (0.000) 0.227 (0.305)CASH_PAYOUT 0.534*** (0.000) 1.195*** (0.000) 1.207*** (0.000)

CASH_PAYOUT 0.104** (0.016) 0.176*** (0.000)

SDRE 0.148*** (0.000) 0.202*** (0.000) 0.157*** (0.000) 0.251*** (0.000)

SDCR 0.295*** (0.000) 0.272*** (0.000) 0.280*** (0.000) 0.261*** (0.000) 0.275*** (0.000)

STOCK_PAYOUT 0.399*** (0.000)

SIZE 0.150*** (0.000) 0.152*** (0.000) 0.010 (0.316)

EXPECTED_ROA 18.651*** (0.000) 18.561*** (0.000) 13.132*** (0.000)

DIVYIELD 25.643*** (0.000) 25.676*** (0.000) 8.978*** (0.000)

TOBIN’S_Q 0.113*** (0.000) 0.115*** (0.000) 0.067*** (0.000)

CURRENT_EPSG 0.002 (0.848) 0.001 (0.884) 0.003 (0.735)

FUTURE_AG 0.348*** (0.000) 0.351*** (0.000) 0.456*** (0.000)

Year Dummies No Yes Yes No Yes

Adj. R2 0.053 0.221 0.22 0.031 0.163

F test 91.623 48.69 48.257 90.624 57.904

Panel B FUTURE_EBITG

(1) p value (2) p value (3) p value (4) p value (5) p value

Intercept 0.019 (0.458) 0.573 (0.409) 0.585 (0.400) 0.192*** (0.000) 0.498*** (0.005)CASH_PAYOUT 0.566*** (0.000) 1.122*** (0.000) 1.132*** (0.000)

CASH_PAYOUT 0.063* (0.084) 0.147*** (0.000)

SDRE 0.020 (0.591) 0.074** (0.028) 0.032 (0.293) 0.120*** (0.000)

SDCR 0.056** (0.036) 0.067*** (0.005) 0.061** (0.010) 0.077*** (0.001) 0.054*** (0.008)

STOCK_PAYOUT 0.091 (0.270)

SIZE 0.055* (0.069) 0.056* (0.066) 0.006 (0.438)

EXPECTED_ROA 16.585*** (0.000) 16.516*** (0.000) 10.730*** (0.000)

DIVYIELD 22.667*** (0.000) 22.621*** (0.000) 8.285*** (0.000)

TOBIN’S_Q 0.122*** (0.000) 0.123*** (0.000) 0.060*** (0.000)

CURRENT_EBITG 0.012 (0.197) 0.013 (0.183) 0.021** (0.024)

FUTURE_AG 0.619*** (0.000) 0.619*** (0.000) 0.722*** (0.000)

Year Dummies No Yes Yes No Yes

Adj. R2 0.026 0.249 0.248 0.002 0.202

F test 29.929 48.822 48.685 4.651 55.321

This table reports the estimation results of Eq.6.1 for FUTURE_EPSG and FUTURE_EBITG. Equations withCASH_PAYOUT are estimated using a fixed effect model adjusted for heteroscedasticity (White, 1980), whileequations with CASH_PAYOUT are estimated using an ordinary least squares (OLS) model adjusted forheteroscedasticity (White, 1980). Panel A contains the results for FUTURE_EPSG while Panel B presents theresults for FUTURE_EBITG. Refer to Table 6.2 for variable definitions. p values are reported in parentheses. *,** and *** denote significance at 10%, 5% and 1% levels, respectively.

165

Panel B of Table 6.9 presents the estimation results of Eq.6.2 with FUTURE_EBITG set

as the dependent variable. Differing slightly from Panel A, when the coefficient on SDRE

maintains a negative sign but with less significance, the coefficient on SDCR becomes

significantly positive, in line with the descriptive statistics shown in Table 6.4 that cash

payers with SDCR only exhibit highest mean FUTURE_EBITG. The divergence between Panel

A and Panel B proposes the different implication of SDCR and SDRE. Because it creates no

tax liabilities on the shareholders, SDCR may be used to imply improvement in the

operation, but the overall performance is undermined by the financial cost and one off

charge. Subject to income tax, SDRE indicates deteriorations in operating performance and

a worsening financial situation because the decision of non cash SDRE casts some doubt

about the quality of the accounting profit.135 Neither the ‘retained earnings hypothesis’ nor

the ‘balanced dual dividend hypothesis’ is able to explain the stock dividend practice in the

Chinese market, because managers utilize stock dividends to appease the shareholders as

a substitute for cash dividends. Although the intention of paying stock dividends may be

different depending on the source, the consequence is the same, that is, managers will

hoard the cash balance and potentially exacerbate the agency cost of free cash flow.

6.4.3. Robustness test

Several additional tests have been conducted to check the robustness of the above

results. First, the regression results indicate that the impacts from both EXPECTED_ROA and

DIVYIELD on the future earnings growth are extremely significant, but in opposite

135 Because the cash only payers occupy about 80% of the cash payer sample, these observations will beexcluded and Eq.6.2 will be applied on a filtered sample in the robustness check.

166

directions. Therefore, Eq.6.1 and Eq.6.2 are estimated with either or neither of them, and

the results are included in Appendix 6.1. The coefficients on CASH_PAYOUT in all

estimations are significantly positive. Second, various econometrics models are applied to

the cash only payer sample in order to test whether the results of the fixed effect model

are robust. Appendix 6.2 includes the regression estimates of Eq.6.1 and Eq.6.2 using the

OLS, Fama Macbeth (1973) two step procedure, random effect model and two stage least

squares (IV) regression. The coefficients on CASH_PAYOUT, SDRE and SDCR exhibit the

same signs with sufficient significance as has been reported for the fixed effect model.

Third, it is possible that the positive influence from cash payouts and changes in cash

payouts on the future earnings growth only exists in sub groups with certain cash payout

levels. Therefore, both the cash only payer sample and the cash payer sample are divided

into High/Medium/Low sub groups based on the third and seventh deciles of

CASH_PAYOUT and CASH_PAYOUT. Eq.6.1 is applied to each sub sample using the Fama

Macbeth (1973) two step procedure and the results are presented in Appendix 6.3.

According to the first three rows of each panel, the coefficients on CASH_PAYOUT are all

significantly positive and significant in all three sub groups, which underpins the

abovementioned pattern in different levels of CASH_PAYOUT. In Panel C and Panel D, the

coefficients on SDRE and SDCR show the same signs as Table 6.10, with similar significance.

But the last three rows of each panel indicate that the contribution from CASH_PAYOUT

on future earnings growth is only valid in the high CASH_PAYOUT subgroup, which

indicates the association between CASH_PAYOUT and future earnings growth is only

significant when CASH_PAYOUT is large enough.

167

Fourth, the cash only payers are omitted from the cash payer sample in order to test

whether the contribution of cash payout to future earnings growth is still valid. Due to the

sample size, Eq.6.2 is estimated using the Fama Macbeth (1973) two step procedure.

Appendix 6.4 suggests the coefficients on SDRE and SDCR report patterns similar to those

reported in Table 6.9, while all coefficients on CASH_PAYOUT are positive and statistically

significant at the 1% level, which indicate the findings in Section 6.4.2.2 are robust.

Fifth, the growth in earnings per share over a two year window is also computed and

set as a dependent variable. Eq.6.1 and Eq.6.2 are estimated using both a fixed effect model

and the Fama Macbeth (1973) two step procedure. As per Appendix 6.5, the results are

consistent with the estimation results for one year growth.

Sixth, there is potential selection bias in the causal relation between cash payout ratio

and future earnings growth, that is, the selection of cash dividend payers may lead to the

conclusion of positive relation between cash payout ratio and future earnings growth.

Therefore, the sample is mixed with zero cash payout firm years and Heckman selection

model is applied, as shown in Appendix 6.6. Step 1 assume CASH_PAYOUT is determined

by SIZE, TOBIN’S_Q and CURRENT_EPSG (CURRENT_EBITG) and step 2 follows Eq.6.1 and

Eq.6.2. The insignificant Miller’s Lambdas indicate selection bias is not serious in this study.

Finally, variance inflated factors (VIF) and tolerance (1/VIF) are included in Appendix 6.7,

that suggest multicollinearity is not a significant issue.

6.5. Summary

This chapter conducts an empirical study into the association between dividend policy

and future earnings growth, based on a sample of 5,380 observations between 1999 and

168

2010 in the Chinese equity market. The traditional theory proposes that a high cash payout

will increase the cost of capital and undermine the after tax performance, but it can also be

argued that cash dividends may supply some positive influence to future earnings growth

because the decrease in cash holding will mitigate the agency cost of free cash flow.

Consistent with Zhou and Ruland (2006) and Vermeulen and Smit (2011), the

empirical evidence presented in this chapter shows a positive relationship between current

cash payout and future earnings growth in the Chinese equity market, which is attributed

to the cash dividend’s function as a mechanism to tackle the agency costs of free cash flow

and improve the managers’ efficiency. Besides the level of cash payout, the change in the

cash payout ratio also has a strong positive association with future earnings growth.

Moreover, the above positive relationship is more significant among the observations with

greater agency costs of free cash flow, proxied by Tobin’s Q. Other firm specific features,

such as expected profitability, dividend yield and future asset growth, also exert a significant

influence on the future earnings growth, but there is no clear evidence to suggest that

earnings growth shows mean reversion.

There is also some evidence that the above documented positive relationship

weakens as the cash payout ratio increases, which can be attributed to the cost of capital.

In particular, in a highly regulated market, the Chinese listed firms are inclined to prefer the

flexibility of internally generated retained earnings as a perceived cheaper source of

funding. Therefore, managers are likely to encounter a dilemma between retaining cash for

future investment opportunities and distributing cash to alleviate shareholders’ concerns

about the agency costs. The payout ratio is determined after consideration of both factors.

169

In contrast, stock dividends in the Chinese market are negatively associated with

future earnings growth. Different from the potential signalling function in developed

markets (Crawford et al., 2005, Bechmann and Raaballe, 2007), the stock dividend practice

in the Chinese market is more likely to be used as a substitute to cash dividends and,

therefore, exacerbates the agency cost of free cash flow, as stock dividends do not require

an actual cash balance. The source of the stock dividend, to some extent, matters, because

stock dividends from capital reserves are positively associated with growth in EBIT, while

stock dividends from retained earnings have a purely negative impact on the growth in both

EPS and EBIT.

170

Chapter 7 Conclusion

7.1. Introduction

This thesis investigates the dividend payout policy and its relationship with ultimate

controlling shareholders, future earnings growth and stock abnormal return around the

announcement window in the Chinese equity market. It is widely accepted that dividend

policy is irrelevant to firm valuation in an efficient market (Miller andModigliani, 1961). But,

in emerging regimes where the equity market is far from efficient, the dividend policy and

its change may influence the market capitalization of underlying firms. This thesis focuses

on one of the largest emerging markets, China, in an attempt to document empirical

evidence of whether corporate finance theories established in developed markets explain

the practices in emerging markets.

As the Chinese economy evolves and its equity market continues to grow, the agency

cost issue raised by Jensen and Meckling (1976) has become more significant, and provides

a relevant explanation of the corporate governance issues arising in the Chinese equity

market. This thesis presents recent evidence regarding the relationship between dividend

payout policy and corporate control rights, and the interaction between stock dividends

and future earnings growth. It also supplies evidence on the relation between dividend

policy, corporate control rights and stock abnormal returns.

Dividend policy is an important component of corporate decision making, and

especially as a substitute corporate governance mechanism (La Porta et al., 2000a). When

the ownership of a firm is separated from its management, the agency cost problem arises

which undermines the operating efficiency of the underlying firm (Jensen, 1986). As a result,

171

cash dividends are introduced as a solution to the agency cost (La Porta et al., 2000a). When

there is a large shareholder with substantial shareholdings, the cash dividend becomes a

useful tool to reduce the cash hoarded by managers and mitigate the agency cost (Truong

and Heaney, 2007). As a legacy of the economic reforms in China, ultimate controlling

shareholders, similar to the largest shareholders in developed markets, have become an

important force in the Chinese equity market. These controllers exercise the function of

monitoring the listed firm, as well as supporting the controlled entity financially, if

necessary. Meanwhile, as the Chinese equity market is strictly regulated by the Chinese

authorities, listed firms need to design payout policy carefully taking into account many

factors, such as the possibility of having to raise future equity capital. The effectiveness of

cash dividends as a solution to the agency problem is reflected in the improvement in

profitability after the management has distributed cash holdings to shareholders. Further,

the dividend announcement provides new information to the secondary market and

investors to enable an assessment of whether the dividend announcement signals any

change in the firm’s operating performance. Therefore, the influence of a dividend policy is

associated with the agency principal problem and other key issues in corporate finance,

such as firm valuation and the operation’s efficiency.

This chapter is organized as follows. Section 7.2 outlines key findings generated from

the empirical research pertaining to the three main topics of research. As mentioned in

Chapter 2, the first topic is the relationship between corporate control rights, dividend

policy and subsequent equity financing. The second topic concentrates on the relationship

between dividend policy, stock abnormal return and corporate control rights. The third

topic examines the interaction between dividend policy and future earnings growth. Finally,

172

Section 7.3 summarizes the thesis, and also identifies some limitations and potential future

research topics extending from the thesis.

7.2. Key findings

7.2.1. Dividend policy, corporate control rights and subsequent equity financing

i. The results, based on the analysis of Chinese listed firms between 2003 and 2010,

are generally consistent with Truong and Heaney (2007) and Cheng et al. (2009). In

this thesis, both the propensity and magnitude of dividend policy are investigated,

in which the propensity is proxied by a dummy variable indicating the existence of a

cash dividend, and the magnitude is measured by the cash payout ratio. Both the

propensity and the magnitude of cash dividend policy vary across the observations

with various types of ultimate controlling shareholder.

ii. There is substantial support for the relationship between cash dividend policy, both

propensity and magnitude, and corporate control rights. Specifically, there is a

significant linear relationship between the propensity and magnitude of cash

dividend payouts, and the control rights held by ultimate controlling shareholders.

With a higher percentage of control rights, the underlying firm has a higher

probability of making a cash dividend distribution, as well as paying more cash

dividends out of current net profits. When the control rights are relatively low, that

is, less than 20 per cent, both the propensity and magnitude of cash dividends

decrease. The proposed explanation to this outcome is that the controlling

shareholders force the managers to pay out the excess cash balance and, thus,

reduce the agency costs of free cash flow.

173

iii. There is also some support for the argument that various ultimate controlling

shareholders exert divergent influences on the cash dividend policy, which is

generally consistent with the findings in Chen et al. (2009c). The influence is more

significant in terms of cash dividend propensity, among firms with SOECG ultimate

controllers which have a higher propensity to pay cash dividends than those with

SOELG or SAMB controllers.

iv. In terms of cash payout magnitude, less divergence is observed among the sub

groups with different ultimate controlling shareholders. Different from the

traditional viewpoint that Chinese firms controlled by Private investors are less

concerned with cash distribution, the existence of Private controllers enhances the

magnitude of cash dividend payouts, which is attributed to the effort of Private

investors to use cash dividends to tackle the agency costs of free cash flow. In

contrast, SAMB controlled firms are more likely to pay lessfewer/lower cash

dividends as their managers are closely affiliated with SAMB controllers, which

exacerbate the expropriation of funds.

v. Firm specific features, such as profitability, firm size and financial leverage, also

influence the cash dividend policy. For instance, larger firms with stronger

profitability are more likely to make cash dividend announcements and pay more

cash dividends.

174

7.2.2. Dividend policy, corporate control rights and announcement effect

7.2.2.1 Dividend policy and stock abnormal return

i. The stock abnormal return around the dividend announcement is the focus of this

chapter. The research extends the work of Cheng et al. (2009) and Chen et al.

(2009a) to test whether the traditional dividend announcement effect is significant

and robust, based on a sample from 1999 to 2010.

ii. Both the impacts from a cash dividend shock and stock dividend decisions are

pursued, where the cash dividend shock is measured by unexpected cash dividends

adjusted for industry factors, and stock dividends are represented by two dummy

variables, depending on the source of the stock dividends (capital reserves or

retained earnings).

iii. To exclude the noise from simultaneous earnings announcements, unexpected

earnings are also generated in a manner similar to unexpected cash dividends, and

included as an explanatory variable along with unexpected dividends.

iv. The univariate analysis indicates that the existence of stock dividends significantly

improves the cumulative abnormal return around the event window, which

underpins the past literature on the Chinese market which finds that investors

welcome stock dividend announcements.

v. Independent from the stock dividend announcements, cash dividend shock has a

marginally significant positive influence on the CARs. Investors react positively to

unexpected cash dividends, suggesting that they take the increase in cash

175

distribution as a positive signal, which is slightly different from the conditional

reaction reported in Chen et al. (2009a).

vi. Contrary to expectation, an earnings shock exhibits a significantly negative impact

on the CARs across some event windows, which can be attributed to the persistent

accounting manipulation by Chinese listed firms.

7.2.2.2 Corporate control rights and stock abnormal return

i. Following the suggestion in Chapter 4, the relationship between corporate control

rights and stock abnormal return is investigated. The justification is that investors

are concerned beyond the unexpected cash dividend, however, the controller of the

firm may lend financial support to the underlying listed firm.

ii. The empirical results indicate that although level of control rights have little

influence on the stock abnormal return, the types of controlling shareholders, along

with unexpected cash dividends, show divergent impacts on the stock abnormal

return. More specifically, investors respond more positively to the unexpected cash

dividend from firms with SAMB or Private controllers. This finding is in line with the

findings in Chapter 4, that firms with SAMB controllers are less likely to distribute

cash dividends. The positive stock abnormal return indicates that the investors

welcome the cash dividend shock.

iii. Some firm specific features contribute to the stock abnormal return, of which the

idiosyncratic risk has a remarkable negative influence on the CAR. Meanwhile, firm

size has a positive influence on the stock abnormal return in certain event windows.

176

7.2.3. Dividend policy and future earnings growth

i. Agency cost and the consequent free cash flow problem lead to inferior investment

return and abuse of the listed firms’ cash balance. Dividend policy is regarded as a

substitute tool for the traditional shareholdermonitoringmechanisms. Although the

traditional pecking order theory focuses on the cost of capital, the empirical results

largely support the agency cost theory.

ii. The results based on the analysis of the Chinese listed firms between 2001 and 2010

are consistent with Arnott and Asness (2003), Zhou and Ruland (2006) and

Vermeulen and Smit (2011). Earnings growth is measured by the growth rate in EPS

and EBIT per share.

iii. The research supplies support for a positive linear relationship between the current

cash dividend, as well as changes, and the future earnings growth, which is

consistent with the notion that the cash dividend works as a solution to reduce the

free cash flow hoarded by the management. In addition, this relationship is more

significant when the underlying firms are exposed to more over investment

problems, which is proxied by higher Tobin’s Q.

iv. There is some evidence that lends support to a potential non linear relationship

between current cash payout and future earnings growth. The justification of this

conjecture is that the cost of capital should be a consideration when the

management design the payout policy (Myers, 1984, Gordon, 1962). The result of

177

the quadratic model suggests managers are aware of the cost of capital issue, and

take it into account.

v. Besides cash dividends, stock dividends are another popular practice in the Chinese

market. The relationship between current stock dividend decisions and future

earnings growth is also investigated. The empirical results show some differences,

depending on the measurement of earnings growth. When it is measured by growth

in EPS, stock dividends, from either capital reserves or retained earnings, are

negatively correlated with future earnings growth. When earnings growth is

measured by growth in EBIT, future earnings growth is positively related with stock

dividends from capital surplus, but negatively correlated with stock dividends from

retained earnings.

vi. Firm level factors, such as the estimated profitability, dividend yield and firm size,

also have a significant impact on future earnings growth.

7.3. Summary

The dividend payout policy is widely debated in the corporate finance literature, in

isolation and also in connection with other modern finance topics, such as firm valuation

and corporate governance mechanisms. In recent years, academic researchers have

concentrated on the function of cash dividends as a mechanism to mitigate the agency cost

stemming from the separation of ownership and management. The agency problem is

exacerbated in emerging markets, as the ownership structure in these markets is not as

dispersed as that in more developed markets such as in the US. The key objective of this

thesis is to provide empirical research into the potential relationship that exists between

178

dividend policy and other corporate finance issues, based on a data set from a prominent

emerging market, China. This thesis enhances the understanding of dividend practices in

the Chinese market by illustrating the existence of: (i) the relationship between corporate

control rights, dividend payout policy and subsequent equity financing; (ii) the relationship

between dividend payout policy and future earnings growth; and (iii) the relationship

between stock abnormal return, dividend policy and corporate control rights.

This thesis contributes to the corporate finance literature on the Chinese equity

market, especially in the areas of dividend policy and its influence on corporate governance

and asset pricing. With the existence of ultimate controlling shareholders, the dividend

policy is designed to mitigate the agency problem between shareholders and management,

because the ultimate controlling shareholders need to exercise more effort to monitor the

firm’s performance. Apparently, this extra monitoring work has to be rewarded with

economic benefit, which explains a positive relationship between the control rights and

cash dividend payouts, as well as the types of ultimate controllers. It is reported that

Chinese listed firms aremore likely tomake a cash dividend announcement and to paymore

retained earnings as the level of control rights increases. Cash dividend payouts are also

sensitive to the types of ultimate controllers. Apart from this issue, this research sheds some

light on the effectiveness of the CSRC regulationwhich links the subsequent equity financing

with cash distribution. The policy does have some influence on the likelihood and

magnitude of corporate dividend payouts, but the changes are more likely to be the result

of listed firms adapting to the CSRC policy in order to pursue equity financing in the

following financial year.

179

A higher propensity and magnitude of cash dividends also achieves the outcome of

decreasing the free cash flow controlled by managers and, consequently, alleviates the

agency costs. Therefore, this research extends to the relationship between dividend

payouts and future earnings growth. The empirical results support the viewpoint that a

higher cash dividend payout is associated with better future earnings growth. This

relationship is more significant in firms with greater over investment problems. On the

contrary, stock dividend announcements have a negative relationship with future earnings

growth, which further highlights the influence of the free cash flow problems.

Finally, this thesis also provides some evidence onwhether investors react to dividend

shock and earnings shock, as well as whether the corporate control rights influence the

stock abnormal return surrounding dividend and earnings announcements. Although the

stock dividend announcement is not accompanied by the expectation of superior future

earnings growth, it is welcomed by investors. Unexpected cash dividends have a marginal

positive contribution to the stock abnormal return, while the unexpected earnings have an

inverse relationship with stock abnormal returns. It suggests that minority shareholders are

less concerned about the cash dividend decision and more focused on the capital gain,

which is in line with the findings in Cheng et al. (2009). Thus, the CSRC regulations to

improve the cash dividend are less appealing to minority shareholders. Further, the

relationship between corporate control rights and stock abnormal returns is better

reflected by the types of ultimate controller, rather than the level of control rights, through

the unexpected dividend announced by the underlying firms.

The empirical results of this research propose some interesting implications to the

market participants and regulators. First, CSRC has imposed several regulations to

180

encourage the cash distribution, but their actual influence on the listed firms’ payout

patterns is questionable. For example, to bond the issuance of new securities with cash

dividend distribution effectively increases the financial cost of listed firms. Instead,

investors will determine whether they prefer future growth or current cash return by using

their voting rights in the annual general meeting. As the market evolves, excessive

regulations slow down or even impede the normal development of listed firms and the

capital market. Second, the different tax treatment on cash dividend and capital gain

undermine the attractiveness of making cash distribution. It might be more constructive

that the regulator contemplate a thorough reform with regards to the taxation of financial

transactions. Finally, this study reflects some inherent weakness of a government

sponsored capital market, especially the imbalanced ownership structure and the tendency

to protect large shareholders. It may require more resolutions to sort out these weakness

and make the capital market more functional as an important part of economic

infrastructure.

Similar to any major thesis, some difficult choices have been made in order to

maintain this research within a controllable size. Therefore, below are some suggestions for

future research:

i. Although the management of Chinese listed firms is likely to have been appointed by

the ultimate controlling shareholders, the agency cost could be further explored

because the personal interest of management is not always in line with that of the

controller or the listed firm. The potential quantitative measures include the

efficiency of asset utilization and discretionary expenses, considering the impact of

the board size (Hermalin and Weisbach, 2003), and executive compensation

181

(Aggarwal and Samwick, 2006). These measures of agency costs can be used to

replace the earnings growth, and in empirical tests of whether the higher cash

dividend payout reduces the agency costs.

ii. The ultimate controllers’ potential tunnelling activities are also in the radar of further

research. Ultimate controlling shareholders may utilize their influence to extract the

financial resources from the listed firms (Friedman et al., 2003). It would be

interesting to investigate whether the controlling shareholders are able to provide

sufficient support to the firm in cases of financial distress.

iii. The information content of changes in cash dividends could be further investigated,

along with the changes in future earnings of Chinese firms. Of particular interest is

whether current cash dividend announcements reveal information on past operating

performance or on expected future earnings.

iv. Last but not least, due to the short term history of the Chinese equity market, the

predictive power of the proposedmodels in this thesis need further testing withmore

current data.

7.4. Concluding remarks

Corporate finance research in the Chinese capital market is far from comprehensive

as the market is growing rapidly. Although this thesis only focuses on the cash dividend

payout policy, it has demonstrated the continuing evolution of the Chinese equity market.

Cash dividend payouts used to be neglected by Chinese investors, but the situation has

changed dramatically as market participants are exhibiting more interest in utilizing cash

dividend payouts to achieve better investment returns, and enhancement of corporate

182

governance. Especially, listed firms with private investors exhibit stronger enthusiasm in

deliberately designing their cash dividend policy to mitigate the agency cost of free cash

flow. Consistent with economic fundamentals, competition is the best catalyst to encourage

innovation and improve efficiency. In the long run, the Chinese government’s financial

deregulation and decreasing involvement in economics are expected to enhance

competition and set the Chinese equity market on a path to economic prosperity.

183

BibliographyADITHIPYANGKUL, P., ALON, I. & ZHANG, T. 2011. Executive perks: Compensation and corporate

performance in China. Asia Pacific Journal of Management, 28, 401 425.

ADJAOUD, F. & BEN AMAR, W. 2010. Corporate governance and dividend policy: Shareholders’protection or expropriation? Journal of Business Finance & Accounting, 37, 648 667.

AGGARWAL, R. K. & SAMWICK, A. A. 2006. Empire builders and shirkers: Investment, firmperformance, and managerial incentives. Journal of Corporate Finance, 12, 489 515.

AGRAWAL, A. & KNOEBER, C. R. 1996. Firm performance and mechanisms to control agencyproblems between managers and shareholders. The Journal of Financial and QuantitativeAnalysis, 31, 377 397.

AHARONY, J., FALK, H. & SWARY, I. 1988. Information content of dividend increases: The case ofregulated utilities. Journal of Business Finance & Accounting, 15, 401 414.

AHARONY, J. & SWARY, I. 1980. Quarterly dividend and earnings announcements andstockholders' returns: An empirical analysis. Journal of Finance, 35, 1 12.

ALI, M. B. & CHOWDHURY, T. A. 2010. Effect of dividend on stock price in emerging stock market:A study on the listed private commercial banks in DSE. International Journal of Economics& Finance, 2, 52 64.

ALLEN, F., BERNARDO, A. E. & WELCH, I. 2000. A theory of dividends based on tax clienteles.Journal of Finance, 55, 2499 2536.

ALLEN, F., JUN, Q. & MEIJUN, Q. 2005. Law, finance, and economic growth in China. Journal ofFinancial Economics, 77, 57 116.

ALTIOK YILMAZ, A. & AKBEN SELCUK, E. 2010. Information content of dividends: Evidence fromIstanbul stock exchange. International Business Research, 3, 126 132.

AMIHUD, Y. & LI, K. 2006. The declining information content of dividend announcements and theeffects of institutional holdings. The Journal of Financial and Quantitative Analysis, 41,637 660.

ANDERSON, H. D., CHI, J., ING ARAM, C. & LIANG, L. 2011. Stock dividend puzzles in China. Journalof the Asia Pacific Economy, 16, 422 447.

ANG, J. S. 1975. Dividend policy: Informational content or partial adjustment? Review ofEconomics & Statistics, 57, 65 70.

ARNOTT, R. D. & ASNESS, C. S. 2003. Surprise! Higher dividends = Higher earnings growth. FinancialAnalysts Journal, 59, 70 87.

ASQUITH, P. & MULLINS, D. W., JR. 1983. The impact of initiating dividend payments onshareholders' wealth. The Journal of Business, 56, 77 96.

ATTIG, N. 2007. Excess control and the risk of corporate expropriation: Canadian evidence.Canadian Journal of Administrative Sciences, 24, 94 106.

184

BABA, N. 2009. Increased presence of foreign investors and dividend policy of Japanese firms.Pacific Basin Finance Journal, 17, 163 174.

BAE, K. H., KANG, J. K. & KIM, J. M. 2002. Tunneling or value added? Evidence from mergers byKorean business groups. Journal of Finance, 57, 2695 2740.

BAEK, J. S., KANG, J. K. & LEE, I. 2006. Business groups and tunneling: Evidence from privatesecurities offerings by Korean Chaebols. Journal of Finance, 61, 2415 2449.

BAJAJ, M. & VIJH, A. M. 1990. Dividend clienteles and the information content of dividendchanges. Journal of Financial Economics, 26, 193 219.

BALACHANDRAN, B. 2003. UK interim and final dividend reductions: A note on price reaction.European Journal of Finance, 9, 379 390.

BAMBER, L. S. 1987. Unexpected earnings, firm size, and trading volume around quarterly earningsannouncements. Accounting Review, 62, 510 532.

BANERJEE, S., GATCHEV, V. A. & SPINDT, P. A. 2007. Stock market liquidity and firm dividendpolicy. Journal of Financial & Quantitative Analysis, 42, 369 397.

BAR YOSEF, S. & HUFFMAN, L. 1986. The information content of dividends: A signalling approach.The Journal of Financial and Quantitative Analysis, 21, 47 58.

BARCLAY, M. J., HOLDERNESS, C. G. & SHEEHAN, D. P. 2009. Dividends and corporateshareholders. Review of Financial Studies, 22, 2423 2455.

BARNEA, A., HAUGEN, R. A. & SENBET, L. W. 1981. An equilibrium analysis of debt financing undercostly tax arbitrage and agency problems. Journal of Finance, 36, 569 581.

BEAVER, W. H. 1968. The information content of annual earnings announcements. Journal ofAccounting Research, 6, 67 92.

BECHMANN, K. L. & RAABALLE, J. 2007. The differences between stock splits and stock dividends:Evidence on the retained earnings hypothesis. Journal of Business Finance & Accounting,34, 574 604.

BECHMANN, K. L. & RAABALLE, J. 2010. Taxable cash dividends A money burning signal.European Journal of Finance, 16, 1 26.

BENARTZI, S., MICHAELY, R. & THALER, R. 1997. Do changes in dividends signal the future or thepast? The jOurnal of Finance, 52, 1007 1034.

BENESH, G. A., KEOWN, A. J. & PINKERTON, J. M. 1984. An examination of market reaction tosubstantial shifts in dividend policy. Journal of Financial Research, 7, 131 142.

BENJAMIN, M. 2006. Family ownership and firm performance: Empirical evidence from WesternEuropean corporations. Journal of Corporate Finance, 12, 321 341.

BERKMAN, H., COLE, R. A. & FU, L. J. 2009. Expropriation through loan guarantees to relatedparties: Evidence from China. Journal of Banking & Finance, 33, 141 156.

185

BERNHARDT, D., DOUGLAS, A. & ROBERTSON, F. 2005. Testing dividend signaling models. Journalof Empirical Finance, 12, 77 98.

BHATTACHARYA, S. 1979. Imperfect information, dividend policy, and "the bird in the hand"fallacy. Bell Journal of Economics, 10, 259 270.

BHAUMIK, S. K. & GREGORIOU, A. 2010. ‘Family’ ownership, tunnelling and earnings management:A review of the literature. Journal of Economic Surveys, 24, 705 730.

BLAU, B. M. & FULLER, K. P. 2008. Flexibility and dividends. Journal of Corporate Finance, 14, 133152.

BORN, J. A. & RIMBEY, J. N. 1993. A test of the Easternbrook hypothesis regarding dividendpayments and agency costs. Journal of Financial Research, 16, 251 260.

BRAV, A., GRAHAM, J. R., HARVEY, C. R. & MICHAELY, R. 2005. Payout policy in the 21st century.Journal of Financial Economics, 77, 483 527.

BROCKMAN, P. & UNLU, E. 2009. Dividend policy, creditor rights, and the agency costs of debt.Journal of Financial Economics, 92, 276 299.

BROCKMAN, P. & UNLU, E. 2011. Earned/contributed capital, dividend policy, and disclosurequality: An international study. Journal of Banking and Finance, 35, 1610 1625.

BURKART, M., GROMB, D. & PANUNZI, F. 1997. Large shareholders, monitoring, and the value ofthe Firm. Quarterly Journal of Economics, 112, 693 728.

CHAE, J., KIM, S. & LEE, E. J. 2009. How corporate governance affects payout policy under agencyproblems and external financing constraints. Journal of Banking & Finance, 33, 2093 2101.

CHANG, X., TAN, T. J., WONG, G. & ZHANG, H. 2007. Effects of financial constraints on corporatepolicies in Australia. Accounting & Finance, 47, 85 108.

CHAREST, G. 1978a. Dividend information, stock returns and market efficiency II. Journal ofFinancial Economics, 6, 297 330.

CHAREST, G. 1978b. Split information, stock returns and market efficiency I. Journal of FinancialEconomics, 6, 265 296.

CHAY, J. B. & SUH, J. 2009. Payout policy and cash flow uncertainty. Journal of FinancialEconomics, 93, 88 107.

CHEN, D. H., LIU, H. H. & HUANG, C. T. 2009a. The announcement effect of cash dividend changeson share prices. Chinese Economy, 42, 62 85.

CHEN, D., JIAN, M. & XU, M. 2009b. Dividends for tunneling in a regulated economy: The case ofChina. Pacific Basin Finance Journal, 17, 209 223.

CHEN, G., CHENG, L. T. W. & GAO, N. 2005a. Information content and timing of earningsannouncements. Journal of Business Finance and Accounting, 32, 65 95.

186

CHEN, G., FIRTH, M. & GAO, N. 2002. The information content of concurrently announcedearnings, cash dividends, and stock dividends: An investigation of the Chinese stockmarket. Journal of International Financial Management & Accounting, 13, 101 124.

CHEN, G., FIRTH, M. & XU, L. 2009c. Does the type of ownership control matter? Evidence fromChina’s listed companies. Journal of Banking & Finance, 33, 171 181.

CHEN, J. J., CHENG, P. & XIAO, X. 2011. Related party transactions as a source of earningsmanagement. Applied Financial Economics, 21, 165 181.

CHEN, S., WANG, Y. & ZHAO, Z. 2009d. Regulatory incentives for earnings management throughasset impairment reversals in China. Journal of Accounting, Auditing & Finance, 24, 589620.

CHEN, Z., CHEUNG, Y. L., STOURAITIS, A. & WONG, A. W. S. 2005b. Ownership concentration, firmperformance, and dividend policy in Hong Kong. Pacific Basin Finance Journal, 13, 431449.

CHENG, L., FUNG, H. G. & LEUNG, T. 2007. Information effects of dividends: Evidence from theHong Kong market. Review of Quantitative Finance and Accounting, 28, 23 54.

CHENG, L. T. W., FUNG, H. G. & LEUNG, T. Y. 2009. Dividend preference of tradable share and nontradable share holders in mainland China. Accounting & Finance, 49, 291 316.

CHENG, L. T. W. & LEUNG, T. Y. 2008. Is there information content from insider trading activitiespreceding earnings and dividend announcements in Hong Kong? Accounting & Finance,48, 417 437.

CHEUNG, Y. L., JING, L., LU, T., RAU, P. R. & STOURAITIS, A. 2009a. Tunneling and propping up: Ananalysis of related party transactions by Chinese listed companies. Pacific Basin FinanceJournal, 17, 372 393.

CHEUNG, Y. L., QI, Y., RAGHAVENDRA RAU, P. & STOURAITIS, A. 2009b. Buy high, sell low: Howlisted firms price asset transfers in related party transactions. Journal of Banking &Finance, 33, 914 924.

CHEUNG, Y. L., RAU, P. R. & STOURAITIS, A. 2006. Tunneling, propping, and expropriation:Evidence from connected party transactions in Hong Kong. Journal of Financial Economics,82, 343 386.

CHOI, D. & CHEN, S. S. 1997. The differential information conveyed by share repurchase tenderoffers and dividend increases. Journal of Financial Research, 20, 529 543.

CLAESSENS, S., DJANKOV, S., FAN, J. P. H. & LANG, L. H. P. 2002. Disentangling the incentive andentrenchment effects of large shareholdings. Journal of Finance, 57, 2741 2771.

CLAESSENS, S., DJANKOV, S. & LANG, L. H. P. 2000. The separation of ownership and control in eastasian corporations. Journal of Financial Economics, 58, 81 112.

CONROY, R. M., EADES, K. M. & HARRIS, R. S. 2000. A test of the relative pricing effects ofdividends and earnings: evidence from simultaneous announcements in Japan. Journal ofFinance, 55, 1199 1227.

187

CRAWFORD, D., FRANZ, D. R. & LOBO, G. J. 2005. Signaling managerial optimism through stockdividends and stock Splits: A reexamination of the retained earnings hypothesis. Journal ofFinancial & Quantitative Analysis, 40, 531 561.

CRUTCHLEY, C. E. & HANSEN, R. S. 1989. A test of the agency theory of managerial ownership,corporate leverage, and corporate dividends. FM: The Journal of the FinancialManagement Association, 18, 36 46.

CRUTCHLEY, C. E. & JENSEN, M. R. H. 1999. Agency problems and the simultaneity of financialdecision making. International Review of Financial Analysis, 8, 177 197.

DEANGELO, H. & DEANGELO, L. 2006. The irrelevance of the MM dividend irrelevance theorem.Journal of Financial Economics, 79, 293 315.

DEANGELO, H. & DEANGELO, L. 2007. Payout policy pedagogy: What matters and why. EuropeanFinancial Management, 13, 11 27.

DEANGELO, H., DEANGELO, L. & SKINNER, D. J. 1996. Reversal of fortune: Dividend signaling andthe disappearance of sustained earnings growth. Journal of Financial Economics, 40, 341371.

DEANGELO, H., DEANGELO, L. & SKINNER, D. J. 2004. Are dividends disappearing? Dividendconcentration and the consolidation of earnings. Journal of Financial Economics, 72, 425456.

DEL BRIO, E. B. & DE MIGUEL, A. 2010. Dividends and market signalling: An analysis of corporateinsider trading. European Financial Management, 16, 480 497.

DENIS, D. J., DENIS, D. K. & SARIN, A. 1994. The information content of dividend changes: Cashflow signaling, overinvestment, and dividend clienteles. The Journal of Financial andQuantitative Analysis, 29, 567 587.

DENIS, D. J., DENIS, D. K. & SARIN, A. 1997. Agency problems, equity ownership, and corporatediversification. Journal of Finance, 52, 135 160.

DHILLON, U. S. & JOHNSON, H. 1994. The effect of dividend changes on stock and bond Prices. TheJournal of Finance, 49, 281 289.

DIVECHA, A. & MORSE, D. 1983. Market responses to dividend increases and changes in payoutratios. Journal of Financial & Quantitative Analysis, 18, 163 173.

DOCKING, D. S. & KOCH, P. D. 2005. Sensitivity of investor reaction to market direction andvolatility: Dividend change announcements. Journal of Financial Research, 28, 21 40.

DYCK, A. & ZINGALES, L. 2004. Private benefits of control: An international comparison. Journal ofFinance, 59, 537 600.

EASTERBROOK, F. H. 1984. Two agency cost explanations of dividends. American EconomicReview, 74, 650 659.

ECKBO, E. B. & VERMA, S. 1994. Managerial shareownership, voting power, and cash dividendpolicy. Journal of Corporate Finance, 1, 33 62.

188

EIJE, H. V. & MEGGINSON, W. L. 2008. Dividends and share repurchases in the European Union.Journal of Financial Economics, 89, 347 374.

ELGERS, P. T. & MURRAY, D. 1985. Financial characteristics related to managements' stock splitand stock dividend decisions. Journal of Business Finance & Accounting, 12, 543 551.

ELTON, E. J. & GRUBER, M. J. 1970. Marginal stockholder tax rates and the clientele effect. Reviewof Economics & Statistics, 52, 68 74.

EUN, C. S. & HUANG, W. 2007. Asset pricing in China's domestic stock markets: Is there a logic?Pacific Basin Finance Journal, 15, 452 480.

FACCIO, M. & LANG, L. H. P. 2002. The ultimate ownership of Western European corporations.Journal of Financial Economics, 65, 365 395.

FACCIO, M., LANG, L. H. P. & YOUNG, L. 2001. Dividends and expropriation. American EconomicReview, 91, 54 78.

FAMA, E. F. & BABIAK, H. 1968. Dividend policy: An empirical analysis. Journal of the AmericanStatistical Association, 63, 1132 1161.

FAMA, E. F. & FRENCH, K. R. 1993. Common risk factors in the returns on stock and bonds. Journalof Financial Economics, 33, 3 56.

FAMA, E. F. & FRENCH, K. R. 2001. Disappearing dividends: Changing firm characteristics or lowerpropensity to pay? Journal of Financial Economics, 60, 3 43.

FAMA, E. F. & FRENCH, K. R. 2002. Testing trade off and pecking order predictions about dividendsand debt. Review of Financial Studies, 15, 1 33.

FAMA, E. F. & JENSEN, M. C. 1983. Agency problem and residual claims. Journal of Law &Economics, 26, 327 350.

FAMA, E. F. & MACBETH, J. D. 1973. Risk, return, and equilibrium: Empirical tests. Journal ofPolitical Economy, 81, 607 636.

FARINHA, J. & LOPEZ DE FORONDA, O. 2009. The relation between dividends and insiderwwnership in different legal systems: International evidence. European Journal of Finance,15, 169 189.

FLUCK, Z. 1998. Optimal financial contracting: Debt versus outside equity. Review of FinancialStudies, 11, 383 418.

FRIEDMAN, E., JOHNSON, S. & MITTON, T. 2003. Propping and tunneling. Journal of ComparativeEconomics, 31, 732 750.

FULLER, K. & BLAU, B. M. 2010. Signaling, free cash flow and “nonmonotonic” dividends. FinancialReview, 45, 21 56.

GAO, L. & KLING, G. 2008. Corporate governance and tunneling: Empirical evidence from China.Pacific Basin Finance Journal, 16, 591 605.

189

GHOSH, C. & WOOLRIDGE, J. R. 1988. An analysis of shareholder reaction to dividend cuts andomissions. Journal of Financial Research, 11, 281 294.

GHOSH, C. & WOOLRIDGE, J. R. 1991. Dividend omissions and stock market rationality. Journal ofBusiness Finance & Accounting, 18, 315 330.

GILLAN, S. L. 2006. Recent developments in corporate governance: An overview. Journal ofCorporate Finance, 12, 381 402.

GORDON, L. A. & MYERS, M. D. 1998. Tobin's Q and overinvestment. Applied Economics Letters, 5,1 4.

GORDON, M. J. 1962. The savings investment and valuation of a corporation. Review of Economicsand Statistics, 37 51.

GRAHAM, J. R., KOSKI, JENNIFER L. & LOEWENSTEIN, U. 2006. Information flow and liquidityaround anticipated and unanticipated dividend announcements. Journal of Business, 79,2301 2336.

GRAHAM, J. R. & KUMAR, A. 2006. Do dividend clienteles exist? Evidence on dividend preferencesof retail investors. Journal of Finance, 1305 1336.

GRINBLATT, M. S., MASULIS, R. W. & TITMAN, S. 1984. The valuation effects of stock splits andstock dividends. Journal of Financial Economics, 13, 461 490.

GRINSTEIN, Y. & MICHAELY, R. 2005. Institutional holdings and payout policy. Journal of Finance,60, 1389 1426.

GROSSMAN, S. J. & HART, O. D. 1980. Takeover bids, the free rider problem, and the theory of thecorporation. Bell Journal of Economics, 11, 42 64.

GRULLON, G., MICHAELY, R. & SWAMINATHAN, B. 2002. Are dividend changes a sign of firmmaturity? Journal of Business, 75, 387 424.

GUGLER, K. 2003. Corporate governance, dividend payout policy, and the interrelation betweendividends, R&D, and capital investment. Journal of Banking & Finance, 27, 1297 1321.

GUGLER, K. & YURTOGLU, B. B. 2003. Corporate governance and dividend pay out policy inGermany. European Economic Review, 47, 731 758.

GUL, F. A. 1999. Government share ownership, investment opportunity set and corporate policychoices in China. Pacific Basin Finance Journal, 7, 157 172.

GWILYM, O. A., SEATON, J., SUDDASON, K. & THOMAS, S. 2006. International evidence on thepayout ratio, earnings, dividends, and returns. Financial Analysts Journal, 62, 36 53.

HAMADI, M. 2010. Ownership concentration, family control and performance of firms. EuropeanManagement Review, 7, 116 131.

HANSEN, R. S., KUMAR, R. & SHOME, D. K. 1994. Dividend policy and corporate monitoring:Evidence from the regulated electric utility industry. Financial Management, 23, 16 22.

190

HARADA, K. & NGUYEN, P. 2005. Dividend change context and signaling efficiency in Japan. PacificBasin Finance Journal, 13, 504 522.

HARRIS, M. & RAVIV, A. 1988. Corporate governance: Voting rights and majority rules. Journal ofFinancial Economics, 20, 203 235.

HAW, I. M., QI, D., WU, D. & WU, W. 2005. Market consequences of earnings management inresponse to security regulations in China. Contemporary Accounting Research, 22, 95 140.

HEALY, P. M. & PALEPU, K. G. 1988. Earnings information conveyed by dividend initiations andomissions. Journal of Financial Economics, 21, 149 175.

HERMALIN, B. E. & WEISBACH, M. S. 1991. The effects of board composition and direct incentiveson firm performance. Financial Management, v20, 101 112.

HERMALIN, B. E. & WEISBACH, M. S. 2003. Boards of directors as an endogenously determinedinstitution: A survey of the economic literature. Economic Policy Review, 9, 7 26.

HOLDERNESS, C. G. 2003. A survey of blockholders and corporate control. Economic Policy Review,9, 51 63.

HOW, J. C. Y., VERHOEVEN, P. & WU, C. L. 2008. Dividends and expropriation in Hong Kong. AsianAcademy of Management Journal of Accounting & Finance, 4, 71 85.

HUANG, C. S., YOU, C. F. & LIN, S. H. 2009. Cash dividends, stock dividends and subsequentearnings growth. Pacific Basin Finance Journal, 17, 594 610.

HUANG, J. J., SHEN, Y. & SUN, Q. 2011. Nonnegotiable shares, controlling shareholders, anddividend payments in China. Journal of Corporate Finance, 17, 122 133.

HUSSIN, B. M., AHMED, A. D. & YING, T. C. 2010. Semi strong form efficiency: Market reaction todividend and earnings announcements in Malaysian stock exchange. IUP Journal ofApplied Finance, 16, 36 60.

IBBOTSON, R. G. & CHEN, P. 2003. Long run stock returns: Participating in the real economy.Financial Analysts Journal, 59, 88 98.

JENSEN, G. R., SOLBERG, D. P. & ZORN, T. S. 1992. Simultaneous determination of insiderownership, debt, and dividend policies. Journal of Financial & Quantitative Analysis, 27,247 263.

JENSEN, M. C. 1986. Agency costs of free cash flow, corporate finance, and takeovers. AmericanEconomic Review, 76, 323 329.

JENSEN, M. C. & MECKLING, W. H. 1976. Theory of the firm: Managerial behavior, agency costsand ownership structure. Journal of Financial Economics, 3, 305 360.

JIANG, G., LEE, C. M. C. & YUE, H. 2010. Tunneling through intercorporate loans: The Chinaexperience. Journal of Financial Economics, 98, 1 20.

JIANG, G., YUE, H. & ZHAO, L. 2009. A re examination of China’s share issue privatization. Journalof Banking & Finance, 33, 2322 2332.

191

JIRAPORN, P. & NING, Y. 2006. Dividend policy, shareholder rights, and corporate governance.Journal of Applied Finance, 16, 24 36.

JOHN, K. & LANG, L. H. P. 1991. Insider trading around dividend announcements: Theory andevidence. Journal of Finance, v46, 1361 1389.

JOHN, K. & WILLIAMS, J. 1985. Dividends, dilution, and taxes: A signalling equilibrium. Journal ofFinance, 40, 1053 1070.

KALAY, A. 1981. Earnings uncertainty and the payouts ration: Some empirical evidence. Review ofEconomics & Statistics, 63, 439 443.

KALAY, A. & LOEWENSTEIN, U. 1985. Predictable events and excess returns: The case of dividendannouncements. Journal of Financial Economics, 14, 423 449.

KALAY, A. & LOEWENSTEIN, U. 1986. The informational content of the timing of dividendannouncements. Journal of Financial Economics, 16, 373 388.

KHAN, T. 2006. Company dividends and ownership structure: Evidence from UK panel data.Economic Journal, 116, 172 189.

KIM, C. S., MAUER, D. C. & SHERMAN, A. E. 1998. The determinants of corporate liquidity: Theoryand evidence. Journal of Financial and Quantitative Analysis, v33, 335 359.

KWAN, C. C. Y. 1981. Efficient market tests of the informational content of dividendannouncements: Critique and extension. Journal of Financial & Quantitative Analysis, 16,193 206.

LAKONISHOK, J. & LEV, B. 1987. Stock splits and stock dividends: Why, who, and when. Journal ofFinance, 42, 913 932.

LANG, L. H. P. & LITZENBERGER, R. H. 1989. Dividend announcements: Cash flow signalling vs. freecash flow hypothesis? Journal of Financial Economics, 24, 181 191.

LA PORTA, R., LOPEZ DE SILANES, F. & SHLEIFER, A. 1999. Corporate ownership around the world.Journal of Finance, 54, 471 517.

LA PORTA, R., LOPEZ DE SILANES, F., SHLEIFER, A. & VISHNY, R. 1998. Law and finance. Journal ofPolitical Economy, 106, 1113 1155.

LA PORTA, R., LOPEZ DE SILANES, F., SHLEIFER, A. & VISHNY, R. 2000a. Agency problems anddividend policies around the world. Journal of Finance, 55, 1 33.

LA PORTA, R., LOPEZ DE SILANES, F., SHLEIFER, A. & VISHNY, R. 2000b. Investor protection andcorporate governance. Journal of Financial Economics, 58, 3 27.

LEE, B. S. 1995. The response of stock prices to permanent and temporary shocks to dividends.Journal of Financial and Quantitative Analysis, 30, 1 22.

LEE, Y. T., LIU, Y. J., ROLL, R. & SUBRAHMANYAM, A. 2006. Taxes and dividend clientele: Evidencefrom trading and ownership structure. Journal of Banking & Finance, 30, 229 246.

192

LIE, E. 2000. Excess funds and agency problems: An empirical study of incremental cashdisbursements. Review of Financial Studies, 13, 219 248.

LIN, Y. H., CHIOU, J. R. & CHEN, Y. R. 2010. Ownership structure and dividend preference:Evidence from China's privatized state owned enterprises. Emerging Markets Finance &Trade, 46, 56 74.

LINTNER, J. 1956. Distribution of incomes of corporations among dividends, retained earnings, andtaxes. American Economic Review, 46, 97 113.

LITZENBERGER, R. H. & RAMASWAMY, K. 1979. The effect of personal taxes and dividends oncapital asset prices: Theory and empirical evidence. Journal of Financial Economics, 7, 163195.

LONIE, A. A., ABEYRATNA, G., POWER, D. M. & SINCLAIR, C. D. 1996. The stock market reaction todividend announcements: A UK study of complex market signals. Journal of EconomicStudies, 23, 32 52.

LU, D., THANGAVELU, S. & HU, Q. 2005. Biased lending and non performing loans in China'sbanking sector. Journal of Development Studies, 41, 1071 1091.

MCCONNELL, J. J. & SERVAES, H. 1995. Equity ownership and the two faces of debt. Journal ofFinancial Economics, 39, 131 157.

MCNICHOLS, M. & DRAVID, A. 1990. Stock dividends, stock splits, and signaling. Journal of Finance,45, 857 879.

MERCADO MENDEZ, J. & WILLEY, T. 1995. Agency costs in the banking industry : An examinationof ownership behavior, leverage and dividend. Journal of Economics & Finance, 19, 105117.

MICHAELY, R., THALER, R. H. & WOMACK, K. L. 1995. Price reactions to dividend initiations andomissions: Overreaction or drift? Journal of Finance, 50, 573 608.

MILLER, M. H. & MODIGLIANI, F. 1961. Dividend policy, growth, and the valuation of shares.Journal of Business, 411 433.

MILLER, M. H. & ROCK, K. 1985. Dividend policy under asymmetric information. Journal of Finance,40, 1031 1051.

MITTON, T. 2004. Corporate governance and dividend policy in emerging markets. EmergingMarkets Review, 5, 409 426.

MORCK, R., SHLEIFER, A. & VISHNY, R. W. 1988. Management ownership and market valuation: Anempirical analysis. Journal of Financial Economics, 20, 293 315.

MYERS, S. C. 1984. The capital structure puzzle. Journal of Finance, 39, 575 592.

MYERS, S. C. 2000. Outside equity. Journal of Finance, 55, 1005 1037.

NG, A., YUCE, A. & CHEN, E. 2009. Determinants of state equity ownership, and its effect onvalue/performance: China's privatized firms. Pacific Basin Finance Journal, 17, 413 443.

193

NISSIM, D. & ZIV, A. 2001. Dividend changes and future profitability. Journal of Finance, 56, 21112133.

NOE, T. H. & REBELLO, M. J. 1996. Asymmetric information, managerial opportunism, financing,and payout policies. Journal of Finance, v51, 637 660.

OFER, A. R. & SIEGEL, D. R. 1987. Corporate financial policy, information, and market expectations:An empirical investigation of dividends. Journal of Finance, 42, 889 911.

OFER, A. R. & THAKOR, A. V. 1987. A theory of stock price responses to alternative corporate cashdisbursement methods: stock repurchases and dividends. Journal of Finance, v42, 365394.

PATTENDEN, K. & TWITE, G. 2008. Taxes and dividend policy under alternative tax regimes. Journalof Corporate Finance, 14, 1 16.

PETTIT, R. R. 1972. Dividend announcements, security performance, and capital market efficiency.Journal of Finance, 27, 993 1007.

PROWSE, S. D. 1992. The structure of corporate ownership in Japan. Journal of Finance, 47, 11211140.

RENNEBOOG, L. & TROJANOWSKI, G. 2007. Control structures and payout policy. ManagerialFinance, 33, 43 64.

ROSS, S. A. 1973. The economic theory of agency: The principal's problem. American EconomicReview, 63, 134 139.

ROZEFF, M. S. 1982. Growth, beta and agency costs as determinants of dividend payout ratios.Journal of Financial Research, 5, 249 259.

SHAFER, W., E. & WANG, Z. 2011. Effects of ethical context and Machiavellianism on attitudestoward earnings management in China. Managerial Auditing Journal, 26, 372 392.

SHEN, C. H. & CHIH, H. L. 2007. Earnings management and corporate governance in Asia'semerging markets. Corporate Governance: An International Review, 15, 999 1021.

SHLEIFER, A. & VISHNY, R. W. 1986. Large shareholders and corporate control. Journal of PoliticalEconomy, 94, 461 488.

SHLEIFER, A. & VISHNY, R. W. 1997. A survey of corporate governance. Journal of Finance, 52, 737783.

SHORT, H., ZHANG, H. & KEASEY, K. 2002. The link between dividend policy and institutionalownership. Journal of Corporate Finance, 8, 105 122.

SKINNER, D. J. 2008. The evolving relation between earnings, dividends, and stock repurchases.Journal of Financial Economics, 87, 582 609.

SU, D. 2005. Corporate finance and state enterprise reform in China. China Economic Review(1043951X), 16, 118 148.

194

SUN, Q. & TONG, W. H. S. 2003. China share issue privatization: The extent of its success. Journalof Financial Economics, 70, 183 222.

TIM, O., LEE, P., RENÉ, S. & ROHAN, W. 1999. The determinants and implications of corporate cashholdings. Journal of Financial Economics, 52, 3 46.

TRUONG, C. 2011. Post earnings announcement abnormal return in the Chinese equity market.Journal of International Financial Markets, Institutions & Money, 21, 637 661.

TRUONG, T. & HEANEY, R. 2007. Largest shareholder and dividend policy around the world.Quarterly Review of Economics & Finance, 47, 667 687.

TWU, M. 2010. Does the order between dividend payment and new stock issuance matter to stockprice? Evidence from Taiwan. Review of Pacific Basin Financial Markets & Policies, 13, 363380.

VENKATESH, P. C. 1989. The impact of dividend initiation on the information content of earningsannouncements and returns volatility. Journal of Business, 62, 175 197.

VERMEULEN, M. & SMIT, E. V. D. M. 2011. Dividend payout and future earnings growth: A SouthAfrican study. South African Journal of Business Management, 42, 33 46.

VLADIMIR, A. 2005. How much value can blockholders tunnel? Evidence from the Bulgarian massprivatization auctions. Journal of Financial Economics, 76, 191 234.

VON MISES, L. 1951. Socialism: An economic and sociological analysis / by L. von Mises. Trans.from the German by J.Kahane, Lond. : Cape, 1951. 2nd ed.

WANG, Y., CHEN, S. K., LIN, B. X. & WU, L. 2008. The frequency and magnitude of earningsmanagement in China. Applied Economics, 40, 3213 3225.

WATTS, R. 1973. The information content of dividends. Journal of Business, 46, 191 211.

WATTS, R. L. & LEFTWICH, R. W. 1977. The time series of annual accounting earnings. Journal ofAccounting Research, 15, 253 271.

WEI, G. & XIAO, J. Z. 2009. Equity ownership segregation, shareholder preferences, and dividendpolicy in China. British Accounting Review, 41, 169 183.

WEI, K. C. J. & ZHANG, Y. 2008. Ownership structure, cash flow, and capital investment: Evidencefrom East Asian economies before the financial crisis. Journal of Corporate Finance, 14,118 132.

WHITE, H. 1980. A heteroskedasticity consistent covariance matrix estimator and a direct test forheteroskedasticity. Econometrica, 48, 817 838.

WILLIAMS, J. 1988. Efficient signalling with dividends, investment, and stock repurchases. Journalof Finance, 43, 737 747.

WOOLRIDGE, J. R. 1982. The information content of dividend changes. Journal of FinancialResearch, 5, 237 247.

195

XU, L. C., ZHU, T. & LIN, Y. M. 2005. Politician control, agency problems and ownership reform.Economics of Transition, 13, 1 24.

YEH, C. T., LIOU, Y. H. & LIN, H. K. 2011. The information content of dividend change at the annualshareholders meeting: Evidence in Taiwan's stock market. International Research Journalof Finance & Economics, 68 80.

YEH, Y. H. & WOIDTKE, T. 2005. Commitment or entrenchment?: Controlling shareholders andboard composition. Journal of Banking and Finance, 29, 1857 1885.

YI, Y., KE, D. & ZHANG, X. 2007. Shareholder wealth effect of dividend policy: Empirical evidencefrom the Chinese securities market. Frontiers of Business Research in China, 1, 437 455.

YOON, P. & STARKS, L. 1995. Signaling, investment opportunities, and dividend announcements.Review of Financial Studies, 8, 995 1018.

ZHANG, H. 2008. Corporate governance and dividend policy: A comparison of Chinese firms listedin Hong Kong and in the Mainland. China Economic Review (1043951X), 19, 437 459.

ZHOU, P. & RULAND, W. 2006. Dividend payout and future earnings growth. Financial AnalystsJournal, 62, 58 69.

ZHOU, Z. G. & ZHOU, J. 2010. Chinese IPO activity, pricing, and market cycles. Review ofQuantitative Finance & Accounting, 34, 483 503.

ZOU, H., WONG, S., SHUM, C., XIONG, J. & YAN, J. 2008. Controlling minority shareholder incentiveconflicts and directors’ and officers’ liability insurance: Evidence from China. Journal ofBanking & Finance, 32, 2636 2645.

196

AppendicesAppendix 4.1 Robustness check: Comparison of various econometric models

Panel A Dummy_CASHDIV Panel B CASH_PAYOUT(1) OLS p value (2) FMB p value (1) OLS p value (2) FMB p value (3) OLS p value (4) FMB p value (5) FE p value (6) RE p value

Intercept 0.135** (0.047) 0.037 (0.885) 0.056 (0.499) 0.108 (0.341) 0.408*** (0.000) 0.472*** (0.001) 1.142*** (0.000) 0.691*** (0.000)CONTROL_RIGHTS 0.185*** (0.000) 0.183*** (0.002) 0.156*** (0.000) 0.157*** (0.000) 0.108*** (0.000) 0.107*** (0.000) 0.134** (0.015) 0.129*** (0.000)

Dummy_LowControl 0.066*** (0.002) 0.045 (0.205) 0.024** (0.034) 0.012 (0.515) 0.003 (0.842) 0.006 (0.674) 0.031 (0.130) 0.020 (0.224)

Size 0.050*** (0.000) 0.041*** (0.006) 0.012*** (0.000) 0.009 (0.106) 0.007** (0.036) 0.006 (0.107) 0.036*** (0.007) 0.018*** (0.000)

PRICE_TO_BOOK 0.029*** (0.000) 0.042*** (0.002) 0.003* (0.085) 0.008* (0.065) 0.016*** (0.000) 0.020*** (0.002) 0.005* (0.048) 0.009*** (0.000)

FREE_CF 0.011 (0.369) 0.016 (0.196) 0.013* (0.072) 0.016* (0.093) 0.009 (0.181) 0.007 (0.413) 0.000 (0.957) 0.003 (0.694)

ROE 2.359*** (0.000) 2.609*** (0.000) 0.234*** (0.000) 0.328** (0.018) 1.168*** (0.000) 1.281*** (0.000) 1.294*** (0.000) 1.261*** (0.000)

RPTTIMING 0.081*** (0.000) 0.072*** (0.004) 0.023** (0.019) 0.020 (0.276) 0.025*** (0.010) 0.022 (0.212) 0.017* (0.088) 0.021** (0.020)

LEVERAGE 0.126*** (0.000) 0.126*** (0.000) 0.069*** (0.000) 0.073*** (0.001) 0.036*** (0.000) 0.038*** (0.001) 0.025** (0.039) 0.036*** (0.000)

SDCR 0.090*** (0.000) 0.087*** (0.000) 0.010 (0.260) 0.010 (0.325) 0.040*** (0.000) 0.039*** (0.005) 0.050*** (0.000) 0.045*** (0.000)

Year Dummy Yes No Yes No Yes No Yes No

Industry Dummy Yes Yes Yes Yes Yes Yes Yes Yes

N 6,386 6,386 6,386 6,386 4,243 4,243 4,243 4,243

R2 0.191 0.211 0.080 0.081 0.189 0.153 0.272

Adj. R2 0.188 0.077 0.185 0.27

F test 95.5 1148.1 31.1 161.9 53.3 34.3 38.4

X2

This table reports the robustness checks of Eq.4.1 and Eq.4.2 with various econometric models. The error term is adjusted for heteroscedasticity (White, 1980). Panel Areports the estimations of Eq.4.1 with ordinary least squares regression and the Fama Macbeth two stage procedure (Fama and Macbeth, 1973). Panel B reports theestimations of Eq.4.2 with ordinary least squares regression and the Fama Macbeth two stage procedure (Fama and Macbeth, 1973), fixed effect model and randomeffect model. Fixed effect models and random effect models are applied to the cash payer sample. Refer to Table 6.2 for variable definitions. p values are reported inparentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

197

Appendix 4.2 Robustness check: MulticollinearityPanel A Likelihood of cash dividend

Eq.4.1Eq.4.3 (UCS_Dummy) Eq.4.3 (Cross effect)

DPrivate DSOECG DSOELG DSAMB DPrivate DSOECG DSOELG DSAMB

Variables VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF

CONTROL_RIGHTS 1.49 0.6711 1.53 0.6542 1.49 0.6708 1.49 0.6692 1.50 0.6651 1.50 0.6686 1.51 0.6622 1.54 0.6480 1.50 0.6651

UCS_Dummy ×CONTROL_RIGHTS 1.16 0.8612 1.01 0.9795 1.07 0.9361 1.13 0.8849

UCS_Dummy 1.20 0.8350 1.00 0.9951 1.03 0.9741 1.13 0.8849

Dummy_LowControl 1.40 0.7151 1.40 0.7150 1.40 0.7139 1.40 0.7150 1.40 0.7147 1.40 0.7137 1.40 0.7145 1.40 0.7149 0.40 0.7147

SIZE 1.41 0.7117 1.47 0.6802 1.41 0.7111 1.41 0.7085 1.48 0.6742 1.47 0.6783 1.41 0.7111 1.41 0.7075 1.48 0.6742

PRICE_TO_BOOK 1.26 0.7954 1.26 0.7918 1.26 0.7953 1.26 0.7916 1.26 0.7952 1.27 0.7895 1.26 0.7952 1.26 0.7913 1.26 0.7952

FREE_CF 1.50 0.6657 1.51 0.6623 1.50 0.6657 1.50 0.6651 1.50 0.6646 1.51 0.6622 1.50 0.6657 1.50 0.6647 1.50 0.6646

ROE 1.40 0.7142 1.41 0.7078 1.40 0.7139 1.40 0.7128 1.40 0.7129 1.41 0.7085 1.40 0.7139 1.40 0.7127 1.40 0.7129

RPTTIMING 1.05 0.9557 1.05 0.9557 1.05 0.9554 1.05 0.9551 1.05 0.9551 1.05 0.9556 1.05 0.9556 1.05 0.9544 1.05 0.9551

LEVERAGE 1.66 0.6018 1.66 0.6011 1.66 0.6012 1.66 0.6010 1.66 0.6015 1.66 0.6014 1.66 0.6010 1.66 0.6009 1.66 0.6015

SDCR 1.04 0.9640 1.07 0.9327 1.04 0.9631 1.04 0.9603 1.05 0.9533 1.08 0.9259 1.04 0.9632 1.04 0.9604 1.05 0.9533

Mean VIF 1.36 1.36 1.32 1.33 1.34 1.35 1.32 1.33 1.34

Panel B Magnitude of cash dividend

Eq.4.2Eq.4.4 (UCS_Dummy) Eq.4.4 (Cross effect)

DPrivate DSOECG DSOELG DSAMB DPrivate DSOECG DSOELG DSAMB

Variables VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF VIF 1/VIF

CONTROL_RIGHTS 1.49 0.6711 1.53 0.6542 1.49 0.6708 1.49 0.6692 1.50 0.6651 1.50 0.6686 1.51 0.6622 1.54 0.6480 1.69 0.5927

UCS_Dummy×CONTROL_RIGHTS 1.16 0.8612 1.02 0.9795 1.07 0.9361 1.34 0.7477

UCS_Dummy 1.20 0.8350 1.00 0.9951 1.03 0.9741 1.13 0.8849

Dummy_LowControl 1.40 0.7151 1.40 0.7150 1.40 0.7139 1.40 0.7150 1.40 0.7147 1.40 0.7137 1.40 0.7145 1.40 0.7149 1.40 1.7150

SIZE 1.41 0.7117 1.47 0.6802 1.41 0.7111 1.41 0.7085 1.48 0.6742 1.47 0.6783 1.41 0.7111 1.41 0.7075 1.48 0.6751

PRICE_TO_BOOK 1.26 0.7954 1.26 0.7918 1.26 0.7953 1.26 0.7916 1.26 0.7952 1.27 0.7895 1.26 0.7952 1.26 0.7913 1.26 0.7949

FREE_CF 1.50 0.6657 1.51 0.6623 1.50 0.6657 1.50 0.6651 1.50 0.6646 1.51 0.6622 1.50 0.6657 1.50 0.6647 1.50 0.6649

ROE 1.40 0.7142 1.41 0.7078 1.40 0.7139 1.40 0.7128 1.40 0.7129 1.41 0.7085 1.40 0.7139 1.40 0.7127 1.40 0.7136

RPTTIMING 1.05 0.9557 1.05 0.9557 1.05 0.9554 1.05 0.9551 1.05 0.9551 1.05 0.9556 1.05 0.9556 1.05 0.9544 1.05 0.9550

LEVERAGE 1.66 0.6018 1.66 0.6011 1.66 0.6012 1.66 0.6010 1.66 0.6015 1.66 0.6014 1.66 0.6010 1.66 0.6009 1.66 0.6017

SDCR 1.04 0.9640 1.07 0.9327 1.04 0.9631 1.04 0.9603 1.05 0.9533 1.08 0.9259 1.04 0.9632 1.04 0.9604 1.05 0.9528

Mean VIF 1.36 1.36 1.362 1.33 1.34 1.35 1.32 1.33 1.38

This table reports the variance inflation factor (VIF) and tolerance (1/VIF) of all regression results documented in Chapter 4. Refer to Table 4.2 for variable definitions.

198

Appendix 4.3 Robustness check: Selection biasPanel A: selection bias between cash dividend decision and control rights

coefficient p value

Step 2 CONTROL_RIGHTSIntercept 0.468*** (0.000)Lagged SIZE 0.041*** (0.000)

Lagged LEVERAGE 0.057*** (0.000)

Lagged PRICE_TO_BOOK 0.007*** (0.000)

Step 1 DUMMY_CASHDIVIntercept 5.274*** (0.000)SIZE 0.297*** (0.000)

PRICE_TO_BOOK 0.087*** (0.000)

FREE_CF 0.004 (0.930)

ROE 8.218*** (0.000)

RPTTIMING 0.241*** (0.000)

LEVERAGE 0.474*** (0.000)

SDCR 0.170*** (0.003)

Mill’s lambda 0.017 (0.136)

N 5,1262 180.0

rho 0.115

sigma 0.148

Panel B: selection bias between cash dividend payout and types of UCSs

coefficient p value coefficient p value coefficient p value coefficient p value

Step 1 DPrivate DSOECG DSOELG DSAMB

Intercept 3.135*** (0.000) 0.525*** (0.000) 0.316** (0.048) 1.925*** (0.000)Lagged SIZE 0.132*** (0.000) 0.026*** (0.000) 0.009 (0.213) 0.115*** (0.000)

Lagged LEVERAGE 0.053*** (0.006) 0.027** (0.011) 0.003 (0.818) 0.022 (0.278)

Lagged PRICE_TO_BOOK 0.020*** (0.000) 0.002 (0.267) 0.007** (0.016) 0.016*** (0.000)

Step 2 DUMMY_CASHDIV DUMMY_CASHDIV DUMMY_CASHDIV DUMMY_CASHDIV

Intercept 5.200 (0.119) 5.200 (0.119) 5.200 (0.119) 5.200 (0.119)CONTROL_RIGHTS 0.661*** (0.000) 0.661*** (0.000) 0.661*** (0.000) 0.661*** (0.000)

SIZE 0.277 (0.000) 0.277 (0.000) 0.277 (0.000) 0.277 (0.000)

PRICE_TO_BOOK 0.088 (0.000) 0.088 (0.000) 0.088 (0.000) 0.088 (0.000)

FREE_CF 0.016 (0.703) 0.016 (0.703) 0.016 (0.703) 0.016 (0.703)

ROE 8.233*** (0.000) 8.233*** (0.000) 8.233*** (0.000) 8.233*** (0.000)

RPTTIMING 0.220*** (0.000) 0.220*** (0.000) 0.220*** (0.000) 0.220*** (0.000)

LEVERAGE 0.463*** (0.000) 0.463*** (0.000) 0.463*** (0.000) 0.463*** (0.000)

SDCR 0.174*** (0.003) 0.174*** (0.003) 0.174*** (0.003) 0.174*** (0.003)

Mill’s lambda 0.053 (0.119) 0.072 (0.000) 0.027 (0.272) 0.559* (0.095)

N 5,162 5,162 5,162 5,1622 289.7 23.59 6.873 193.7

rho 0.119 0.290 0.084 0.097

sigma 0.442 0.248 0.324 0.479

This table reports the results of Heckman selection models. Panel A reports the causality relation betweencontrol rights and cash dividend decision, while Panel B and C reports the causality relation between cashdividend decision and types ofUCSs. In Panel A, step 1 selects observations with positive cash dividend payout,which is determined by all control variables. Step 2 regress CONTROL_RIGHTS on SIZE, LEVERAGE andPRICE_TO_BOOK. In Panel, step 1 selects observations with positive cash dividend payout, which isdetermined by CONTROL_RIGHTS and other control variables. Step 2 regress UCS_Dummy on lagged SIZE,lagged LEVERAGE and lagged PRICE_TO_BOOK. Refer to Table 4.2 for other variable definitions. p values arereported in parentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

199

Appendix 5.1 Robustness check: Samples before winsorizationPanel A Cash payers before winsorization

CAR_PRE CAR_SHORT CAR_MEDIUM CAR_LONG

Intercept 0.1491*** (0.000) 0.0516** 0.0026 0.0003 (0.990) 0.0895** (0.012)

UNEXPECTED_DPS 0.1379 (0.075) 0.1843** 0.0038 0.2214*** (0.005) 0.2817** (0.016)

UNEXPECTED_EPS 0.0000 (0.897) 0.0001 0.074 0.0001 (0.285) 0.0004** (0.038)

SDCR 0.0216*** (0.000) 0.0225*** (0.000) 0.0328*** (0.000) 0.0405*** (0.000)

SDRE 0.0050 (0.162) 0.0096*** (0.000) 0.0097*** (0.005) 0.0094* (0.070)

RPTTIMING 0.0224*** (0.000) 0.0021 (0.335) 0.0009 (0.750) 0.0108** (0.010)

IDIO_RISK 1.4415*** (0.000) 0.5478*** (0.000) 1.2211*** (0.000) 2.5043*** (0.000)

Size 0.0011 (0.198) 0.0023*** (0.000) 0.0012 (0.165) 0.0000 (0.991)

LEVERAGE 0.0005 (0.730) 0.0007 (0.590) 0.0002 (0.885) 0.0021 (0.411)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.063 0.045 0.063 0.085

F test 16.14 12.65 16.31 20.88

Panel B Cash only payers before winsorization

CAR_PRE CAR_SHORT CAR3_MEDIUM CAR_LONG

Intercept 0.1488*** (0.000) 0.0757*** (0.000) 0.0255 (0.303) 0.0658* (0.086)

UNEXPECTED_DPS 0.0656 (0.408) 0.2017*** (0.002) 0.1997** (0.017) 0.2068* (0.094)

UNEXPECTED_EPS 0.0001 (0.625) 0.0001 (0.243) 0.0001 (0.480) 0.0006** (0.025)

SDCR

SDRE

RPTTIMING 0.0207*** (0.000) 0.0014 (0.561) 0.0018 (0.546) 0.0093** (0.036)

IDIO_RISK 1.3992*** (0.000) 0.4220*** (0.006) 1.0093*** (0.000) 2.2535*** (0.000)

Size 0.0011 (0.228) 0.0034*** (0.000) 0.0024*** (0.007) 0.001 (0.489)

LEVERAGE 0.0008 (0.639) 0.0007 (0.646) 0.0014 (0.437) 0.002 (0.481)

Year Dummy Yes Yes Yes Yes

Adj. R2 0.055 0.023 0.0407 0.0806

F test 12.12 7.20 9.13 16.84

This table reports robustness checks on the samples before winsorization. Equation 5.7 is estimated usingordinary least squares (OLS) regression. Refer to Table 5.2 for variable definitions. The sample period is from1999 to 2010. Figures in parentheses are p values based on White’s heteroscedasticity robust standard error(White, 1980). *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

200

Appendix 5.2 Robustness check: MulticollinearityTable 5.6 (Eq.5.7) Table 5.7 (Eq.5.7) Table 5.9 (Eq.5.9)

Variables VIF 1/VIF VIF 1/VIF VIF 1/VIF

UNEXPECTGED_DPS 1.04 0.9640 1.03 0.9728 1.02 0.9782

UNEXPECTGED_EPS 1.03 0.9673 1.03 0.9723 1.02 0.9779

SDCR 1.16 0.8638

SDRE 1.15 0.8678

RPTTIMING 1.03 0.9750 1.03 0.9693 1.02 0.9790

IDIO_RISK 1.04 0.9645 1.03 0.9677 1.03 0.9755

Size 1.14 0.8752 1.13 0.8844 1.13 0.8814

LEVERAGE 1.14 0.8745 1.13 0.8840 1.11 0.8984

Mean VIF 1.09 1.06 1.06

This table reports the variance inflation factor (VIF) and tolerance (1/VIF) of all regression results documentedin Chapter 5. Refer to Table 5.2 for variable definitions.

Appendix 5.3 Robustness check: Fama Macbeth two step procedurePanel A: Stock dividend and CAR

CAR_PRE p value CAR_SHORT p value CAR_MEDIUM p value CAR_LONG p value

Intercept 0.1271*** (0.010) 0.0536*** (0.005) 0.0088 (0.754) 0.0937 (0.235)

UNEXPECTED_DPS 0.0903 (0.494) 0.2270* (0.082) 0.2506** (0.049) 0.3348** (0.041)

UNEXPECTED_EPS 0.0016** (0.034) 0.0007 (0.465) 0.0005 (0.514) 0.0005 (0.638)

SDCR 0.0193*** (0.003) 0.0221*** (0.000) 0.0311*** (0.000) 0.0347*** (0.000)

SDRE 0.0028 (0.497) 0.0088*** (0.004) 0.0076* (0.074) 0.0076 (0.237)

RPTTIMING 0.0212*** (0.002) 0.0001 (0.956) 0.0026 (0.445) 0.0104 (0.209)

IDIO_RISK 1.2862*** (0.000) 0.4955*** (0.001) 0.9996*** (0.000) 2.3320*** (0.002)

Size 0.0001 (0.939) 0.0026*** (0.001) 0.0016 (0.211) 0.0003 (0.940)

LEVERAGE 0.0025 (0.459) 0.0001 (0.928) 0.0002 (0.946) 0.0049 (0.453)

N 6,071 6,071 6,071 6,071

R2 0.078 0.065 0.064 0.082

Panel B: Unexpected cash dividend and CAR

CAR_PRE p value CAR_SHORT p value CAR_MEDIUM p value CAR_LONG p value

Intercept 0.1162** (0.013) 0.0736*** (0.003) 0.0236 (0.426) 0.079 (0.291)

UNEXPECTED_DPS 0.0647 (0.553) 0.2289* (0.085) 0.2398* (0.086) 0.3052** (0.032)

UNEXPECTED_EPS 0.0015*** (0.010) 0.0003 (0.711) 0.0011 (0.261) 0.0007 (0.593)

RPTTIMING 0.0203*** (0.000) 0.0011 (0.716) 0.0048 (0.322) 0.0119 (0.187)

IDIO_RISK 1.1697*** (0.000) 0.3676** (0.038) 0.7624*** (0.002) 1.9622*** (0.002)

Size 0.0001 (0.967) 0.0036*** (0.000) 0.0025** (0.046) 0.0003 (0.930)

LEVERAGE 0.0008 (0.821) 0.0001 (0.972) 0.0019 (0.528) 0.0047 (0.502)

N 4,806 4,806 4,806 4,806

R2 0.0508 0.0433 0.0399 0.0607

This table reports robustness checks on Sections 5.4.2.1 and 5.4.2.2. Eq.7 is estimated using Fama Macbethtwo stage procedure (Fama andMacbeth, 1973). Refer to Table 5.2 for variable definitions. The sample periodis from 1999 to 2010. Figures in parentheses are p values. *, ** and *** denote significance at 10%, 5% and 1%levels, respectively.

201

Appendix 6.1 Robustness check: Omitting EXPECTED_ROA and/or DIVYIELDPanel A: cash only payer sample

(1) p value (2) p value (3) p value (4) p value (5) p value (6) p value

Intercept 2.872*** (0.005) 2.168*** (0.027) 3.462*** (0.001) 0.742*** (0.007) 0.023 (0.937) 0.931*** (0.001)CASH_PAYOUT 0.612*** (0.000) 0.699*** (0.000) 1.062*** (0.000)

CASH_PAYOUT 0.134** (0.012) 0.168*** (0.001) 0.153*** (0.004)

SIZE 0.155*** (0.000) 0.126*** (0.003) 0.175*** (0.000) 0.005 (0.681) 0.028** (0.023) 0.016 (0.194)

EXPECTED_ROA 16.430*** (0.000) 12.324*** (0.000)

DIVYIELD 21.943*** (0.000) 5.175*** (0.000)

TOBIN’S_Q 0.014 (0.480) 0.074*** (0.000) 0.048** (0.020) 0.008 (0.462) 0.059*** (0.000) 0.004 (0.687)

CURRENT_EPSG 0.007 (0.616) 0.022 (0.143) 0.008 (0.553) 0.015 (0.284) 0.004 (0.793) 0.016 (0.267)

FUTURE_AG 0.403*** (0.000) 0.362*** (0.000) 0.411*** (0.000) 0.532*** (0.000) 0.499*** (0.000) 0.533*** (0.000)

Year Dummies Yes Yes Yes Yes Yes Yes

Adj. R2 0.099 0.123 0.137 0.092 0.118 0.096

F test 20.269 22.749 22.813 27.948 31.95 27.48

Panel B: cash payer sample

(1) p value (2) p value (3) p value (4) p value (5) p value (6) p value

Intercept 3.170*** (0.000) 2.594*** (0.001) 3.570*** (0.000) 0.746*** (0.001) 0.074 (0.738) 0.895*** (0.000)CASH_PAYOUT 0.544*** (0.000) 0.602*** (0.000) 0.999*** (0.000)

CASH_PAYOUT 0.109*** (0.000) 0.136*** (0.000) 0.126*** (0.000)

SDRE 0.250*** (0.000) 0.258*** (0.000) 0.258*** (0.000) 0.268*** (0.000) 0.268*** (0.000) 0.272*** (0.000)

SDCR 0.140*** (0.000) 0.163*** (0.000) 0.164*** (0.000) 0.172*** (0.000) 0.197*** (0.000) 0.196*** (0.000)

SIZE 0.167*** (0.000) 0.143*** (0.000) 0.179*** (0.000) 0.007 (0.424) 0.021** (0.025) 0.016* (0.087)

EXPECTED_ROA 13.492*** (0.000) 10.783*** (0.000)

DIVYIELD 20.472*** (0.000) 4.471*** (0.000)

TOBIN’S_Q 0.021* (0.082) 0.067*** (0.000) 0.043*** (0.000) 0.006 (0.419) 0.051*** (0.000) 0.003 (0.621)

CURRENT_EPSG 0.001 (0.905) 0.011 (0.300) 0.015 (0.126) 0.007 (0.472) 0.002 (0.839) 0.007 (0.447)

FUTURE_AG 0.399*** (0.000) 0.363*** (0.000) 0.399*** (0.000) 0.492*** (0.000) 0.465*** (0.000) 0.490*** (0.000)

Year Dummies Yes Yes Yes Yes Yes Yes

Adj. R2 0.143 0.165 0.18 0.123 0.15 0.126

F test 40.915 44.896 42.639 51.724 58.49 49.395

This table reports the estimation results of Eq.6.1 and Eq.6.2 for FEPSG and FEBITG using OLS, Fama Macbeth (1973) two step procedure and a random effect model.Panel A and Panel B contain the results for cash only payer sample while Panel C and Panel D present the results for cash payer sample. In each panel, the first rowpresents estimation results using OLS; the second row presents estimation results using Fama Macbeth (1973) two step procedure (FM) and the third row presentsestimation results using random effect model (RE). Refer to Table 6.2 for variable definitions. p values are reported in parentheses. *, ** and *** denote significance at10%, 5% and 1% levels, respectively.

202

Appendix 6.2 Robustness check: Comparison of various econometric modelsPanel A Cash only payer sample FUTURE_EPSG FUTURE_EBITG

(1) OLS p value (2) FM p value (3) RE p value (4) OLS p value (5) FM p value (6) RE p value

Intercept 0.837*** (0.004) 0.264 (0.487) 0.809*** (0.008) 0.832*** (0.000) 0.169 (0.411) 0.834*** (0.000)

CASH_PAYOUT 0.785*** (0.000) 0.851*** (0.001) 0.880*** (0.000) 0.658*** (0.000) 0.733*** (0.000) 0.688*** (0.000)

SIZE 0.009 (0.477) 0.013 (0.510) 0.008 (0.551) 0.005 (0.577) 0.01 (0.340) 0.005 (0.532)

EXPECTED_ROA 18.361*** (0.000) 21.162*** (0.000) 19.659*** (0.000) 13.699*** (0.000) 16.493*** (0.000) 14.130*** (0.000)

DIVYIELD 19.209*** (0.000) 23.247*** (0.000) 21.035*** (0.000) 16.050*** (0.000) 20.122*** (0.000) 16.585*** (0.000)

TOBIN’S_Q 0.118*** (0.000) 0.177*** (0.002) 0.123*** (0.000) 0.102*** (0.000) 0.151*** (0.000) 0.104*** (0.000)

CURRENT_EPSG 0.012 (0.375) 0.033 (0.129) 0.008 (0.566)

CURRENT_EBITG 0.011 (0.302) 0.009 (0.539) 0.013 (0.254)

FUTURE_AG 0.543*** (0.000) 0.537*** (0.000) 0.523*** (0.000) 0.744*** (0.000) 0.720*** (0.000) 0.734*** (0.000)

Year Dummies Yes No Yes Yes No Yes

R2 0.159 0.15 0.22 0.19

Adj. R2 0.156 0.217

F test 36.68 38.05 48.6 54.48

Panel B Cash payer sample FUTURE_EPSG FUTURE_EBITG

OLS p value FM p value RE p value OLS p value FM p value RE p value

Intercept 0.870*** (0.000) 0.417 (0.162) 0.829*** (0.000) 1.101*** (0.000) 0.385** (0.047) 1.101*** (0.000)

CASH_PAYOUT 0.699*** (0.000) 0.786*** (0.000) 0.760*** (0.000) 0.659*** (0.000) 0.755*** (0.000) 0.659*** (0.000)

SDRE 0.179*** (0.000) 0.166*** (0.000) 0.181*** (0.000) 0.052** (0.047) 0.047 (0.107) 0.052** (0.045)

SDCR 0.255*** (0.000) 0.256*** (0.000) 0.259*** (0.000) 0.072*** (0.000) 0.070** (0.022) 0.072*** (0.000)

SIZE 0.012 (0.218) 0.019 (0.211) 0.010 (0.310) 0.014* (0.067) 0.018** (0.032) 0.014* (0.051)

EXPECTED_ROA 15.241*** (0.000) 17.648*** (0.000) 15.885*** (0.000) 12.734*** (0.000) 15.052*** (0.000) 12.734*** (0.000)

DIVYIELD 16.811*** (0.000) 20.736*** (0.000) 17.907*** (0.000) 15.735*** (0.000) 19.819*** (0.000) 15.735*** (0.000)

TOBIN’S_Q 0.091*** (0.000) 0.149*** (0.001) 0.094*** (0.000) 0.083*** (0.000) 0.140*** (0.000) 0.083*** (0.000)

CURRENT_EPSG 0.004 (0.628) 0.026 (0.137) 0.003 (0.722)

CURRENT_EBITG 0.009 (0.318) 0.000 (0.973) 0.009 (0.310)

FUTURE_AG 0.496*** (0.000) 0.489*** (0.000) 0.488*** (0.000) 0.760*** (0.000) 0.735*** (0.000) 0.760*** (0.000)

Year Dummies Yes No Yes Yes No Yes

R2 0.188 0.173 0.233 0.198

Adj. R2 0.185 0.23

F test 61.06 377.66 60.99 5235.17

203

Appendix 6.2 Robustness check: Comparison of various econometric models (Cont.)(1)

FUTRUE_EPSG p value (2)FUTURE_EBITG p value (3)

FUTRUE_EPSG p value (4)FUTURE_EBITG p value

Panel C: Two Stage Least Squares

Intercept 0.057 (0.406) 0.002 (0.976) 0.024 (0.662) 0.002 (0.961)

CASH_PAYOUT 0.886*** (0.005) 1.304*** (0.000) 0.809*** (0.001) 1.083*** (0.000)

SDRE 0.140*** (0.004) 0.038 (0.356)

SDCR 0.245*** (0.000) 0.098*** (0.000)

EXPECTED_ROA 21.267*** (0.000) 18.760*** (0.000) 17.737*** (0.000) 16.471*** (0.000)

DIVYIELD 17.965*** (0.000) 20.652*** (0.000) 15.685*** (0.000) 17.677*** (0.000)

FUTURE_AG 0.167*** (0.000) 0.300*** (0.000) 0.147*** (0.000) 0.271*** (0.000)

TOBIN’S_Q 0.164*** (0.000) 0.172 (0.000) 0.128*** (0.000) 0.134*** (0.000)

R2 0.070 0.052 0.100 0.080

F test 49.74 64.95 71.45 63.88

# Observations 4,277 4,277 5,380 5,380

Panel D: First Stage for CASH_PAYOUT

Intercept 0.914*** (0.000) 0.904*** (0.000) 0.935*** (0.000) 0.935*** (0.000)

SDRE 0.123*** (0.000) 0.121*** (0.000)

SDCR 0.042*** (0.000) 0.040*** (0.000)

EXPECTED_ROA 4.404*** (0.000) 4.351*** (0.000) 3.738*** (0.000) 3.693*** (0.000)

DIVYIELD 11.344*** (0.000) 11.354*** (0.000) 11.423*** (0.000) 11.427*** (0.000)

FUTURE_AG 0.055*** (0.000) 0.055*** (0.000) 0.044*** (0.000) 0.044*** (0.000)

TOBING’S_Q 0.054*** (0.000) 0.054*** (0.000) 0.042*** (0.000) 0.043*** (0.000)

SIZE 0.033*** (0.000) 0.032*** (0.000) 0.033*** (0.000) 0.033*** (0.000)

CURRENT_EPSG 0.012*** (0.000) 0.010*** (0.000)

CURRENT_EBITG 0.019*** (0.000) 0.017*** (0.000)

R2 0.358 0.361 0.417 0.420

Panel E: Diagnostic Tests

F test of excl.inst.

96.18 106.13 128.93 140.66

Sargan overidp value

0.37 0.53 0.93 0.24

This table reports the estimation results of equation (6.1) and (6.2) for FUTURE_EPSG and FUTURE_EBITGusing OLS, Fama Macbeth (1973) two step procedure and a random effect model. Panel A and Panel B containthe results for cash only payer sample and cash payer sample, respectively. In each panel, the first rowpresents estimation results using OLS; the second row presents estimation results using Fama Macbeth (1973)two step procedure (FM), and the third row presents estimation results using a random effect model (RE).Panel C, D and E report 2 stage least squares estimates, instrumenting CASH_PAYOUT using current EPS (EBIT)growth and firm size. Refer to Table 6.2 for variable definitions. p values are reported in parentheses. *, ** and*** denote significance at 10%, 5% and 1% levels, respectively.

204

Appendix 6.3 Robustness check: High/Medium/Low CASH_PAYOUT ( CASH_PAYOUT)Panel A: FUTURE_EPSG – cash only –payer sample

(1) High p value (2) Medium p value (3) Low p value (4) High p value (5) Medium p value (6) Low p valueIntercept 0.674 (0.257) 0.194 (0.532) 0.487 (0.642) 0.190 (0.778) 0.062 (0.919) 0.712 (0.361)CASH_PAYOUT 0.594** (0.026) 0.864** (0.042) 1.807*** (0.001)CASH_PAYOUT 0.509** (0.013) 0.321 (0.336) 0.105 (0.523)

SIZE 0.043 (0.129) 0.003 (0.867) 0.020 (0.696) 0.021 (0.478) 0.001 (0.975) 0.030 (0.425)EXPECTED_ROA 27.948*** (0.000) 19.377*** (0.000) 20.681*** (0.000) 22.816*** (0.000) 13.745*** (0.003) 18.026*** (0.001)DIVYIELD 24.432*** (0.000) 25.514*** (0.001) 37.968*** (0.000) 20.177*** (0.003) 7.186*** (0.006) 10.756*** (0.003)TOBIN’S_Q 0.228*** (0.009) 0.159*** (0.004) 0.212*** (0.000) 0.191*** (0.000) 0.094** (0.046) 0.120** (0.013)CURRENT_EPSG 0.007 (0.914) 0.047 (0.113) 0.008 (0.686) 0.001 (0.982) 0.037 (0.718) 0.013 (0.649)FUTURE_AG 0.523*** (0.000) 0.464*** (0.002) 0.641*** (0.002) 0.613** (0.004) 0.476*** (0.000) 0.509*** (0.002)N 1,283 1,711 1,283 1,283 1,711 1,283R2 0.186 0.178 0.19 0.163 0.14 0.156F test 18.34 97.526 231.025 36.282 46.669 7.267Panel B: FUTURE_EBITG – cash only payer sample

(1) High p value (2) Medium p value (3) Low p value (4) High p value (5) Medium p value (6) Low p valueIntercept 0.202 (0.648) 0.166 (0.422) 0.304 (0.693) 0.102 (0.860) 0.189 (0.651) 0.780 (0.106)CASH_PAYOUT 0.534** (0.011) 0.806** (0.026) 1.737*** (0.000)CASH_PAYOUT 0.332*** (0.006) 0.488 (0.121) 0.164 (0.309)

SIZE 0.023 (0.267) 0.011 (0.341) 0.012 (0.750) 0.020 (0.750) 0.004 (0.750) 0.034 (0.750)EXPECTED_ROA 22.435*** (0.000) 15.056*** (0.000) 14.793*** (0.000) 17.630*** (0.000) 11.145*** (0.002) 12.540*** (0.003)DIVYIELD 21.485*** (0.000) 21.773*** (0.000) 35.332*** (0.000) 17.983*** (0.000) 6.200*** (0.001) 7.529*** (0.009)TOBIN’S_Q 0.218*** (0.003) 0.133*** (0.000) 0.154*** (0.001) 0.163*** (0.000) 0.095*** (0.006) 0.083*** (0.001)CURRENT_EBITG 0.012 (0.440) 0.03 (0.280) 0.013 (0.486) 0.055** (0.045) 0.066 (0.147) 0.023 (0.384)FUTURE_AG 0.655*** (0.000) 0.771*** (0.000) 0.689*** (0.000) 0.750*** (0.000) 0.703*** (0.000) 0.669*** (0.000)N 1,283 1,711 1,283 1,283 1,711 1,283R2 0.211 0.251 0.225 0.194 0.206 0.203F test 18.646 20.95 423.592 39.952 46.779 93.97

205

Appendix 6.3: Robustness check: High/Medium/Low CASH_PAYOUT ( CASH_PAYOUT) (Cont.)Panel C: FUTURE_EPSG cash payer sample

(13) High p value (14) Medium p value (15) Low p value (16) High p value (17) Medium p value (18) Low p valueIntercept 0.745 (0.138) 0.168 (0.684) 0.615 (0.220) 0.266 (0.599) 0.057 (0.907) 0.333 (0.547)CASH_PAYOUT 0.541** (0.011) 1.174*** (0.001) 2.176*** (0.000)CASH_PAYOUT 0.441*** (0.007) 0.037 (0.871) 0.150 (0.170)

SDRE 0.142 (0.110) 0.128** (0.016) 0.112*** (0.000) 0.328*** (0.004) 0.171*** (0.006) 0.217*** (0.001)SDCR 0.263*** (0.001) 0.243*** (0.000) 0.237*** (0.000) 0.307*** (0.000) 0.286*** (0.000) 0.254*** (0.000)SIZE 0.044* (0.071) 0.007 (0.746) 0.022 (0.380) 0.023 (0.298) 0.005 (0.833) 0.013 (0.619)EXPECTED_ROA 24.201*** (0.000) 17.624*** (0.000) 14.581*** (0.000) 19.139*** (0.000) 12.361*** (0.001) 15.276*** (0.000)DIVYIELD 21.835*** (0.000) 26.964*** (0.000) 32.653*** (0.002) 18.572*** (0.001) 5.409** (0.022) 8.521*** (0.002)TOBIN’S_Q 0.189*** (0.004) 0.154*** (0.000) 0.159*** (0.004) 0.144*** (0.000) 0.088** (0.028) 0.116*** (0.002)CURRENT_EPSG 0.017 (0.629) 0.047 (0.055) 0.026 (0.393) 0.003 (0.888) 0.020 (0.482) 0.002 (0.897)FUTURE_AG 0.462*** (0.000) 0.491*** (0.000) 0.480*** (0.000) 0.468*** (0.003) 0.428*** (0.000) 0.500*** (0.002)N 1,614 2,152 1,614 1,614 2,152 1,614R2 0.201 0.19 0.214 0.162 0.178 0.181F test 79.422 55.672 8.30E+04 650.027 28.312 357.178Panel D: FUTURE_EBITG – cash payer sample

(19) High p value (20) Medium p value (21) Low p value (22) High p value (23) Medium p value (24) Low p valueIntercept 0.410 (0.277) 0.365 (0.143) 0.558 (0.140) 0.314 (0.538) 0.045 (0.906) 0.425 (0.309)CASH_PAYOUT 0.544*** (0.004) 0.999*** (0.000) 2.414*** (0.000)CASH_PAYOUT 0.357*** (0.001) 0.35 (0.131) 0.209* (0.060)

SDRE 0.044 (0.274) 0.008 (0.871) 0.008 (0.819) 0.243** (0.014) 0.055 (0.106) 0.088* (0.072)SDCR 0.095* (0.064) 0.074** (0.035) 0.076 (0.109) 0.069 (0.132) 0.042 (0.158) 0.040 (0.142)SIZE 0.030 (0.112) 0.015 (0.185) 0.02 (0.187) 0.028 (0.240) 0.004 (0.772) 0.020 (0.310)EXPECTED_ROA 20.331*** (0.000) 13.420*** (0.000) 14.416*** (0.000) 15.716*** (0.000) 10.479*** (0.001) 11.823*** (0.000)DIVYIELD 20.268*** (0.000) 22.918*** (0.000) 39.125*** (0.000) 17.617*** (0.000) 5.786*** (0.006) 6.590*** (0.002)TOBIN’S_Q 0.198*** (0.001) 0.128*** (0.000) 0.158*** (0.001) 0.135*** (0.000) 0.084*** (0.008) 0.094*** (0.001)CURRENT_EBITG 0.006 (0.583) 0.039* (0.095) 0.001 (0.953) 0.045** (0.025) 0.046 (0.100) 0.004 (0.840)FUTURE_AG 0.713*** (0.000) 0.801*** (0.000) 0.665*** (0.000) 0.699*** (0.000) 0.672*** (0.000) 0.764*** (0.000)N 1614 2152 1614 1,614 2,152 1,614R2 0.221 0.242 0.256 0.182 0.224 0.219F test 74.393 1414.31 1122.024 1576.655 202.521 7250.117

This table reports the robustness check of Eq.6.1 and Eq.6.2 for FEPSG and FEBITG of High/Medium/Low CASH_PAYOUT ( CASH_PAYOUT) sub samples. Panel A andPanel B use cash only payer sample while Panel C and Panel D use cash payer sample. The sample is sorted by either CASH_PAYOUT or CASH_PAYOUT, respectively,and the top 30% is defined as High, the middle 40% is defined as Medium while the bottom 30% is defined as Low. Both equations are estimated using Fama Macbeth(1973) two step procedure. The first and fourth rows of each panel present estimation results of High CASH_PAYOUT ( CASH_PAYOUT) sub group. The second and fifthrows of each panel present estimation results of Medium CASH_PAYOUT ( CASH_PAYOUT) sub group. The third and sixth rows of each panel present estimation resultsof Low CASH_PAYOUT ( CASH_PAYOUT) sub group. Refer to Table 6.2 for variable definitions. p values are reported in parentheses. *, ** and *** denote significance at10%, 5% and 1% levels, respectively.

206

Appendix 6.4 Robustness check: Cash payers with simultaneous stock dividend(1) p value (2) p value (3) p value (4) p value

Intercept 0.757* (0.066) 0.777** (0.052) 0.981* (0.078) 1.067** (0.047)

CASH_PAYOUT 0.642*** (0.008) 0.667*** (0.005) 0.905*** (0.001) 0.948*** (0.001)

SDRE 0.060 (0.160) 0.001 (0.971)

SDCR 0.119** (0.040) 0.113** (0.043) 0.104 (0.104) 0.129* (0.053)

STOCK_PAYOUT 0.123 (0.114) 0.081 (0.416)

SIZE 0.026 (0.123) 0.027 (0.107) 0.039* (0.098) 0.041* (0.068)

EXPECTED_ROA 11.227*** (0.000) 11.175*** (0.000) 12.800*** (0.000) 13.056*** (0.000)

DIVYIELD 16.099*** (0.003) 16.643*** (0.003) 20.290*** (0.003) 20.597*** (0.002)

TOBIN’S_Q 0.094 (0.104) 0.094 (0.101) 0.124* (0.063) 0.126* (0.056)

CURRENT_EPSG 0.012 (0.403) 0.011 (0.456)

CURRENT_EBITG 0.047 (0.168) 0.044 (0.206)

FUTURE_AG 0.370*** (0.003) 0.363*** (0.004) 0.779*** (0.000) 0.759*** (0.000)

N 1103 1103 1103 1103

R2 0.248 0.247 0.319 0.322

F test 85.641 89.39 26.178 183.89

This table reports the estimation results of equation (6.2) for FUTURE_EPSG and FUTURE_EBITG of cashdividend payers with simultaneous stock dividend. Eq.6.2 is estimated using the Fama Macbeth two stepprocedure (Fama and Macbeth, 1973). Refer to Table 6.2 for variable definitions. p values are reported inparentheses. *, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

207

Appendix 6.5 Robustness check: EPS growth over 2 year windowPanel A: cash only payer sample

FUTURE_EPSG2 (1) p value (2) p value (3) p value

Intercept 3.590*** (0.000) 0.403 (0.112) 0.118 (0.460)

CASH_PAYOUT 0.771*** (0.000) 0.634***

CASH_PAYOUT 0.148*** (0.000)

SIZE 0.160*** (0.000) 0.021* (0.096) 0.002 (0.816)

EXPECTED_ROA 2.216* (0.053) 3.195*** (0.006) 0.144 (0.851)

DIVYIELD 14.796*** (0.000) 13.247*** (0.000) 4.507*** (0.000)

TOBIN’S_Q 0.094*** (0.000) 0.092*** (0.002) 0.031*** (0.000)

CURRENT_EPSG2 0.176*** (0.000) 0.273*** (0.000) 0.238*** (0.000)

FUTURE_AG2 0.107*** (0.000) 0.122*** (0.000) 0.115*** (0.000)

Year Dummies Yes No Yes

R2 0.244 0.253 0.175

Adj. R2 0.241 0.172

F test 38.21 150.08 31.77

Panel B: cash payer sample

FUTURE_EPSG2 (1) p value (2) p value (3) p value

Intercept 3.515*** (0.000) 0.491* (0.052) 0.003 (0.982)

CASH_PAYOUT 0.724*** (0.000) 0.592*** (0.000)

CASH_PAYOUT 0.126*** (0.000)

SDRE 0.082*** (0.001) 0.072*** (0.003) 0.142*** (0.000)

SDCR 0.119*** (0.000) 0.127*** (0.000) 0.148*** (0.000)

SIZE 0.158*** (0.000) 0.024* (0.051) 0.001 (0.902)

EXPECTED_ROA 3.493*** (0.000) 2.692*** (0.007) 0.107 (0.857)

DIVYIELD 13.615*** (0.000) 12.258*** (0.000) 4.010*** (0.000)

TOBIN’S_Q 0.071*** (0.000) 0.076*** (0.002) 0.025*** (0.000)

CURRENT_EPSG2 0.184*** (0.000) 0.266*** (0.000) 0.233*** (0.000)

FUTURE_AG2 0.105*** (0.000) 0.122*** (0.000) 0.112*** (0.000)

Year Dummies Yes No Yes

R2 0.286 0.272 0.208

Adj. R2 0.283 0.205

F test 58.32 967.33 54.45

This table reports the estimation results of Eq.6.1 and Eq.6.2. The first column of each panel reports estimationresults using a fixed effect model; the second column of each panel reports estimation results, using the FamaMacbeth (1973) two step procedure; the third column of each panel reports estimation results using OLS.Dependent variable, FUTURE_EPSG2, is compounded EPS growth between year t and year t+2.CURRENT_EPSG2 refers lagged FUTURE_EPSG. FUTURE_AG2 is the growth in total assets between year t andyear t+2. Refer to Table 6.2 for other variable definitions. p values are reported in parentheses. *, ** and ***

denote significance at 10%, 5% and 1% levels, respectively.

208

Appendix 6.6 Robustness check: selection bias – Heckman selection modelCoefficient p value coefficient p value Coefficient p value coefficient p value

Step 2: FUTURE_EPSG FUTURE_EBITG FUTURE_EPSG FUTURE_EBITG

Intercept 4.121* (0.097) 3.750 (0.202) 2.860 (0.140) 1.287 (0.552)

CASH_PAYOUT 0.536*** (0.000) 0.742*** (0.000) 0.507*** (0.000) 0.717*** (0.000)

SDRE 0.191** (0.013) 0.015 (0.829)

SDCR 0.262*** (0.000) 0.072 (0.130)

SIZE 0.157 (0.101) 0.143 (0.206) 0.108 (0.146) 0.048 (0.559)

EXPECTED_ROA 20.901*** (0.000) 14.370*** (0.000) 17.126*** (0.000) 12.893*** (0.000)

DIVYIELD 12.576*** (0.000) 13.211*** (0.000) 10.990*** (0.000) 12.165*** (0.000)

TOBIN’S_Q 0.150*** (0.000) 0.148*** (0.000) 0.129*** (0.000) 0.111*** (0.000)

CURRENT_EPSG 0.008 (0.408) 0.006 (0.305)

CURRENT_EBITG 0.003 (0.250) 0.001 (0.452)

FUTURE_AG 0.130*** (0.001) 0.300*** (0.000) 0.128*** (0.000) 0.293*** (0.000)

Step 1: CASH_PAYOUT CASH_PAYOUT CASH_PAYOUT CASH_PAYOUT

Intercept 6.501*** (0.000) 6.645*** (0.000) 6.466*** (0.000) 6.621*** (0.000)

SIZE 0.308*** (0.000) 0.314*** (0.000) 0.307*** (0.000) 0.313*** (0.000)

TOBIN’S_Q 0.060*** (0.000) 0.071*** (0.000) 0.093*** (0.000) 0.101*** (0.000)

CURRENT_EPSG 0.050*** (0.000) 0.046*** (0.000)

CURRENT_EBITG 0.006** (0.019) 0.005** (0.019)

Mill’s lambda 0.998 (0.105) 0.283 (0.603) 0.732 (0.138) 0.273 (0.615)No. ofObservations 7,067 7,067 8,475 8,475

2 116.8 162.1 179.3 211.4

rho 0.583 0.563 0.488 0.218

sigma 1.712 1.466 1.500 1.253

This table reports the results of Heckman selection model on the causality relation between earnings growthand cash dividend payout. The sample size includes all observations with positive or zero cash dividend payout.Step 1 selects observations with positive CASH_PAYOUT, which is determined by SIZE, TOBIN’S_Q andCURRENT_EPSG (CURRENT_EBITG). Step 2 regress FUTURE_EPSG (FUTURE_EBITG) on CASH_PAYOUT andother control variables. Refer to Table 6.2 for other variable definitions. p values are reported in parentheses.*, ** and *** denote significance at 10%, 5% and 1% levels, respectively.

Appendix 6.7 Robustness check: MulticollinearityVariables VIF 1/VIF

CASH_PAYOUT 2.01 0.4974

CASH_PAYOUT 1.01 0.9937

SIZE 1.18 0.8504

EXPECTED_ROA 1.35 0.7415

DIVYIELD 2.30 0.4343

TOBIN’S_Q 1.47 0.6807

CURRENT_EPSG 1.02 0.9835

CURRENT_EBITG 1.01 0.9898

FUTURE_AG 1.06 0.9412

Mean VIF 1.38

This table reports the variance inflation factor (VIF) and tolerance (1/VIF) of all regression resultsdocumented in Chapter 5. Refer to Table 5.2 for variable definitions.