1. Stock Dividend Puzzles in China Hamish D. Anderson* Massey University Jing Chi Massey University Chayot Ing-aram Massey University Lu Liang Massey University 報告人 : 佘濰任, 黎氏翠恆

2. Abstract To examine the stock dividend puzzle in China during 1992-2008 by analyzing the market reaction, who pays more or solely stock dividends compared to cash dividends and the reason. Stock dividends send a positive market signal. In China, state-owned shareholders prefer cash dividends, while legal-person shareholders prefer stock dividends.

3. Introduction Stock dividends are especially popular in the Chinese stock market. This paper objections are: - What is the scope of the stock dividend payments by the Chinese listed companies? - Why are so many companies engaged in such financial activities? - What kinds of companies are likely to pay more or solely stock dividends? - What are the market reactions to the stock dividend announcements? - The stock dividend payment behaviour and the relationship between the stock dividends and cash dividends.

4. Literature Review The Signaling Hypothesis The Cash Substitution Hypothesis The Trading Range Hypothesis Ownership Structure and Stock Dividends

5. The Signaling Hypothesis The private information about a company’s future perspectives needs to be conveyed to the market through a credible channel, and stock dividends can be a signaling device. Baker and Phillips (1993) show that US managers strongly believe that stock dividends have a positive investor psychological impact, and investors perceive stock dividend announcements as positive signals of future performances and investment opportunities. In China Chen et al. (2002) suggests that stock dividends are used to signal future earnings information to the market.

6. The Cash Substitution Hypothesis Due to the limited financial resources or asset expansion requirements, companies can use stock dividends instead of paying cash dividends. Baker and Phillips (1993) agree that stock dividends are used as temporary substitute for cash dividends. To the companies regularly paying stock dividends, 28.9% of them agree that stock dividends are a substitute for cash dividends. Wei et al. (2004) conclude that in order to satisfy the financial requirements for future growth, companies with good investment opportunities are more likely to pay stock dividends, and stock dividends are used as substitute of cash dividends for conserving cash flows.

7. The Trading Range Hypothesis Stock dividends could lead to a decrease in share prices to a desired range which makes the stocks more attractive to investors. In a study of Korean stock market, Dhatt et al. (1997) find that stock dividend payout ratios are positively related to stock prices and negatively related to firm size before stock dividends announcements. Wei and Xiao (2009) find that in China management does not use stock dividends to reduce share prices, since there is no significant relationship between stock price and stock dividend payout ratios.

8. Ownership Structure and Stock Dividends Controlling shareholders might expropriate minority shareholders through dividend policy to maximize returns. Khan (2006) examines the relationship between dividends and ownership structure of 330 large UK companies from 1985 to 1997 and the results indicate that there is a strong positive relationship between dividend payout ratio and ownership concentration in the UK market. Cheng et al. (2009) and Wei and Xiao (2009) both find that in China the cash dividend level is significantly and positively associated with the proportion of non- tradable shares and the stock dividend level is significantly and positively related to the proportion of tradable shares.

9. Data All of the dividend announcements, financial information and market returns are collected from the China Stock Market and Accounting Research (CSMAR) database. The sample of the stock dividend announcements from 1992 to 2007 is used in this paper with complete trading information during the estimation period and the window period. The regression analysis sample period is from January 2000 to December 2008, since the corporate governance data used as independent variables is only available in CSMAR database during this period.

10. Statistic Summary

11. Table 2: Cash dividend payment

15. Market Reaction of Stock Dividends Announcements Measuring CARs: An event study is used in this paper to investigate the risk-adjusted market reaction of stock dividends announcements from January 1992 to 2007. The sample using in the event study analysis is 1993 stock dividends announcements The market model used to calculate the daily abnormal return Ajt, for firm J, on day t, is shown below: Ajt = Rjt – (αj + βj x Rm) Rjt is the return for share of company j Rm is the index at day t t is working time, estimated 250 working days

16. Event Study Estimation Results

17. Explaining Stock Dividends in China Hypotheses for Stock Dividend Payments Hypotheses for Cash Dividend Payments Hypotheses for CARs around Stock Dividend Announcements Regression Models

18. Hypotheses for Stock Dividend Payments SD H1: There is a negative relationship between the state ownership and the stock dividend per share. State (-) SD H2: There is a positive relationship between the legal person ownership and the stock dividend per share. Legal (+) SD H3: There is a positive relationship between company’s assets growth rates and stock dividend per share. GrwAssets (+) SD H4: There is a positive/negative relationship between company’s cash level and stock dividend per share. Cash-to-assets (+/-) SD H5: There is a positive relationship between EPS and stock dividend per share.EPS (+) SD H6: There is a negative relationship between total number of outstanding shares and stock dividend per share. Shares (-) SD H7: There is a negative relationship between the listing time and stock dividend per share. Listing time (-) SD H8: There is a negative relationship between the independent director dummy and the stock dividend per share. Independent directors (-)

19. Hypotheses for Cash Dividend Payments CD H1: There is a positive relationship between the state ownership and cash dividend per share. State (+) CD H2: There is a negative relationship between stock dividend per share and cash dividend per share. SDPS (-)

20. Hypotheses for CARs around Stock Dividend Announcements CARs H1: There is a positive relationship between stock dividend payout ratio (SDPS) and CARs around the announcements. SDPS (+) CARs H2: There is a positive relationship between cash dividend payout ratio (CDPS) and CARs around the announcements. CDPS (+)

21. Regression Models The regression models using stock dividend per share as dependent variable to examine the determinants of stock dividend payout ratio are shown in Equations 1 and 2. Equations (1-2) SDPS it = α0 + β1 Stateit /Legalit + β2 Cash − to − assetsit + β3 GrwAssets it +β4 EPSit + β5 Sharesit + β6 Listing Timeit +β7 Independent Directorsit + β8 YRD00it + β9 YRD01it +β10 YRD02it + β11 YRD03it + β12 YRD04it + β13 YRD06it +β14 YRD07it +β15 YRD08it + µ it

22. Regression Models Using cash dividend per share as the dependent variable to allow comparisons with stock dividends and to examine whether there is a substitute relationship between stock dividends and cash dividends is shown in Equations 3 and 4 Equations (3-4) CDPS it = α0 + β1 Stateit /Legalit + β2 SDPSit + β3 Cash − to – assets it + β4 GrwAssetsit + β5 EPSit + β6 Listing Time it + β7 Independent Directorsit + β8 YRD00it + β9 YRD01it + β10 YRD02it + β11 YRD03it + β12 YRD04it + β13 YRD06it + β14 YRD07it +β15 YRD08it + µ it

23. Regression Models Examining the signaling hypothesis whether stock dividend announcements have impacts on CARs around the announcements is shown in Equations 5 and 6. Equations (5-6) CARs it = α0 + β1 SDPSit + β2 CDPSit + β3 Cash − to − assetsit + β4 GrwAssetsit + β5 EPSit /EPS Currentit +β6 Listing Timeit + β7Independent Directorsit + β8 YRD00it + β9 YRD01it + β10 YRD02it + β11 YRD03it + β12 YRD04it + β13 YRD06it + β14 YRD07it + µ it

26. Empirical Results for Regression Analyses Determinants of Stock Dividend Payments Stock Dividends can be a Substitute for Cash Dividends Determinants of CARs around Stock Dividend Announcements Logit Model for Robustness Test

27. Empirical Results for Regression Analyses

28. Determinants of Stock Dividend Payments The relationship between the percentage of the state-owned shares and the stock dividends payment is significantly negative and the relationship between the percentage of the legal-person shares and the stock dividends payment is significantly positive. The coefficient of assets growth rates is significantly positive at the 5% level, indicating that companies which have better investment opportunities tend to pay more stock dividends. This result is consistent with the cash substitution hypothesis. The coefficient of EPS is significantly positive at the 1% level, showing that profitable companies tend to pay out more stock dividends. A company’s total number of outstanding shares is negatively related to the stock dividend payout ratio at the 1% significance level, which implies that companies that have high total number of outstanding shares are more cautious in terms of paying stock dividends.

29. Stock Dividends can be a Substitute for Cash Dividends The negative relationship between SDPS and CDPS is statistically significant at the 1% level. It was found that companies with higher cash level, low investment opportunity, and higher profitability pay more cash dividends. The coefficient of the state-ownership variable is significantly positive at the 1% level, the state-owned shareholders would prefer to have cash dividends rather than stock dividends

30. Determinants of CARs around Stock Dividend Announcements The stock dividend payout ratio has a significantly positive impact on CARs. It was found investors believe that stock dividends are positive signals of companies’ performance, and their belief is stronger when companies pay cash dividends at the same time. The relationship between the listing time and CARs at the 5% level is significant positive, which implies that if a company is listed in the stock market for longer time and still pays stock dividends, it is a sign of strong growth and good performance of the company.

31. Logit Model for Robustness Test The binary logit model is used to investigate the characteristics of the firms which only pay stock dividends. “SD Only” is used as the dependent variable. “CD Only” and “SD and CD” are used as dependent variables for comparison purposes. Equations (7-8) SD Only it = α0 + β1 stateit /Legalit + β2 Cash − to − assetsit + β3 GrwAssetsit + β4 EPSit + β5 Sharesit + β6 Listing Timeit + β7 Independent Directorsit + β8 YRD00it + β9 YRD01it + β10 YRD02it + β11 YRD03it + β12 YRD04it + β13 YRD06it +β14 YRD07it + β15 YRD08it + µ it The results of logit models are shown in Table 9. This difference shows that when Chinese listed companies do not perform well, they would pay stock dividends only.

33. Conclusions Dividing the whole sample period into two parts (from 1990 to 1999 and from 2000 to 2008), the percentage of firms paying stock dividends is shown to be much less in the second half, but with much larger payout ratios, while the percentage of companies paying cash dividends is much higher in the second half, but with lower payout ratios. The stock dividends in China have other use to control shareholders. The state-owned shareholders much more prefer cash dividends to maximize their returns. The listed companies use stock dividends to substitute cash dividends. Listed companies like to use stock dividends to attract investors.