The Impact of Financial Crisis on Cash Dividend Policy in China: Ownership Structure Perspective

(1)

The Impact of Financial Crisis on Cash Dividend

Policy in China: Ownership Structure Perspective

Master Thesis Peiqi Xu University of Groningen: S2264250 Uppsala University: 890917-P236 Email: piercexu@gmail.com Supervisor: Dr. H. Gonenc Assessor: Dr. J. H. von Eije

January 9th, 2014

Abstract

This paper aims to investigate the impact of financial crisis and state ownership on cash dividend policy in China. Logit and tobit regressions are implemented, using firms listed on Shenzhen and Shanghai A-share Stock Exchange from 2006 to 2012. The results indicate that the cash dividend policy of firms in China is negatively influenced by financial crisis while positively affected by state ownership. There is no apparent evidence to show that the difference of cash dividend payout between state owned firms and non-state owned firms gets bigger during the financial crisis period.

JEL Classification: G35, G32, E32

(2)

1 1. Introduction

Dividend policy is one of the most important issues in corporate finance and it has been a great challenge for experts to explain why companies tend to pay out dividends. Regarding financial issues in most companies, managers are commonly faced with two main operating decisions, which are how to invest and how to finance capitals. When the company has gained profits, a dilemma has been rose whether the firm should distribute the earnings to the shareholders or use them to reinvest in the business activities. No matter which decision the managers will choose, they should focus on how to maximize the shareholders’ wealth. Black (1976) argues that the reasons why company pay dividends and why investors pay attention to dividend are still ambiguous. With pieces that do not fit together, the dividend figure seems like a puzzle if scholars dig deep into the questions (Black, 1976). Miller and Modigliani (1961) suggests that it is extremely important for corporate officials who set up regulations, investors who manage portfolios and economists who try to understand the capital markets to study the dividend policy.

(3)

2

the payout of cash dividends of the current year (Chen et al., 2009).

(4)

3

shares and attitude change of managers. Thus, the research question is formed: Do financial crisis and state ownership have impacts on cash dividend policy in China?

The paper acquires data mainly from the China Stock Market Accounting Research (CSMAR). Panel data have been selected from firms listed as A-share in Shenzhen and Shanghai Stock Exchange from 2006 to 2012. 2008 is defined as the year of financial crisis and percentage of the largest shareholders ownership is used to measure the extent of state control. Logit regression is run to see the propensity of cash dividend payouts and tobit regressions are used to find how exactly these two factors affect the cash dividend policy from payout ratio and total amount perspectives. Results show that financial crisis has a significant negative impact on cash dividend policy while state ownership has a significant positive one. However, no apparent evidence shows that difference of cash dividend policy between state owned firms and non-state owned firms gets bigger during the crisis period. We also find that firms with higher profitability, lower growth opportunity, larger size, less debt, more cash holding, more retained earnings and lower income risk tend to pay more cash dividends, which are consistent with the studies by Fama and French (2001), DeAnglo et al. (2006), Von Eije and Megginson (2008) and Bradford et al. (2013). It is also proved that the lag year cash dividend policy and current year stock dividend policy have positive impacts on current year cash dividend policy. The paper adds contribution to the existing literature by suggesting that financial crisis can be a separate impact factor on cash dividend policy from a quantitative perspective and it is also the first study using data from Chinese market. In addition, the paper also provides a holistic view of state ownership in China and what specific role it plays on cash dividend decisions.

(5)

4

models and provides some analysis. Section five concludes the paper with some implications and limitations.

2. Theoretical Background and Hypotheses Development

Based on the perfect market assumption, firm’s value is determined by its earning ability and decision of investment, not by how much retained earnings it distributes to the shareholders and the choice of dividend policy should not affect the share price (Miller and Modigliani, 1961). Thus, the financial decisions of managers shall be irrelevant to the dividend policy. However, in reality, the perfect market does not exist because of the tax difference, transaction and floating cost, information asymmetry, agency problems and whether participants are price takers. By challenging and releasing some of the assumptions, several dividend theories have been developed, such as the bird-in-hand theory (Diamond, 1967; Gordon and Shapiro, 1956), dividend tax theory (Brennan, 1970; Litzenberger and Ramaswamy, 1982), tax clientele theory (Allen et al., 2000), agency theory (Jensen, 1986), lifecycle theory (Fama and French, 2001; DeAnglo et al. 2006) and catering theory (Baker and Wurgler, 2004). It is recommended to read the review of dividend policy by Bhattacharyya (2007), Al-Malkawi et al. (2010), Frankfurter and Wood (2002) and Baker (2009) since this paper will not summarize all the dividend theories. Due to the unique institutional environment of China which is extremely different from U.S and other developed countries, it is interesting to see what factors affect the dividend policy in China. Lee and Xiao (2004) empirically test the dividend tunneling theory suggested by La Porta et al. (1999a) and Johnson et al. (2000) using the data of Chinese firms. This paper will mainly focus on discussing the impact of financial crisis on cash dividend policy in China from an ownership structure perspective.

2.1.Dividend Policy in U.S, E.U and China

(6)

5

(Chen et al., 2002). Deng et al. (2013) argue that the previous existing dividend theories based on the U.S data have limited explanatory power for the dividend decisions in China. Thus, there are some differences existing among the dividend policy of U.S, E.U and China.

The general dividend payout ratio varies differently in these countries. Compared to Von Eije and Megginson (2008) and Westin and Siu (2003), who studies dividend policy in E.U and U.S respectively, the payout ratio and propensity to pay cash dividend is lower in China (Huang et al., 2011). The dividend payout ratio of China is higher in the later years but still about 20% (Von Eije and Megginson, 2008) lower than that of E.U and 40% lower than that of U.S. (Westin and Siu, 2003). DeAngelo et al. (2004) explain that the concentration of earnings and dividend among limited large corporations results in the high aggregate payout ratio. The payout ratio in East Asia is lower than that of E.U since corporate insiders tend to invest in negative NPV projects and try to expropriate private benefits resulting from comparatively low investor protection in East Asia (Faccio et al., 2001). Wei and Xiao (2009) show that the five largest shareholders account for average 56% of the total shares in China while average 25% in the U.S. Using a sample of Chinese firms from 1996 to 1999, Lee and Xiao (2004) find that 47% of the firms pay cash dividend and the average payout ratio is 20%. However, Fama and French (2001) argue that the propensity of U.S firms paying cash dividends has declined sharply to 20.8% in 1999 and at this period, the payout ratio and frequency of payments in China are quite similar to that of U.S (Lee and Xiao, 2004).

(7)

6

payout ratio with the speed of earnings change. Based on the U.S data, Skinner (2008) finds that more and more firms tend to repurchase shares instead of paying cash dividends and finds that earnings has little impact on cash dividends. However, Von Eije and Megginson (2008) document that the relation between earnings and cash dividend is stronger than that of earnings and share repurchase, indicating that still earnings have an impact on cash dividend in E.U. In China, Wang et al. (2011) show that the cash dividend payout ratio of firms responds quickly to the change of earnings and this finding contradict the result of Brav et al. (2005) and is consistent with that of Von Eije and Megginson (2008). They also find that the average payout ratio of cash dividend in China falls between emerging economics and developed economics.

(8)

7

wealth and benefits from the minority shareholders by using the dividend policy. Tradable and non-tradable shares are a typical symbol of Chinese firms. Compared to the U.S markets, China is a fast growing economy and the ownership structure is unusual with tradable and non-tradable shares (Nguyen and Wang, 2012). Though part of the non-tradable shares become liquid after the ownership structure reform, the intervention of the government on limiting the tradable amount and the inherent illiquidity of controlling shares still play a significant role to maintain the concentrated ownership (Liu et al., 2014). Faccio et al. (2001) suggest that there is a decrease in dividends in Asia while an increase in Europe with the presence of large shareholders. However, in China, large shareholders always hold non-tradable shares and the fraction of non-tradable shares to the total shares is positively related to the cash dividend payouts (Huang et al., 2011). Huang et al. (2011) also find that the percentage of shares held by controlling shareholders and proportion of non-tradable shares are positive correlated to the likelihood and payout ratio of cash dividend in China and there is significant evidence to show that controlling shareholders expropriate wealth from the minority shareholders.

(9)

8

transactions (Su et al., 2014). Such expropriation could have a large impact on the cash dividend policy. Chen et al. (2002) show that, in China, stock dividend plays a better signaling role than cash dividend which has little significant impact on the abnormal stock return or even negative impact indicating that investors respond pessimistically to cash dividends (Chen et al., 2009). The negative market reaction to cash dividend announcement provides evidence to support the tunneling hypothesis.

2.2.Financial Crisis and Cash Dividend

The financial crisis in 2008 has been regarded as the worst financial shock after the Great Depression in 1930 and has an international impact, affecting most of the countries worldwide (Erkens et al., 2012). Taylor (2009) points out that there is a bust and boom in the real estate industry in U.S and later it turns out to be a financial storm over the world. They also suggest that the main reason causing the financial crisis is the frequently monetary excess. The investment bank Lehman Brothers bankrupted without bearing the financial shock and the rate of large borrowers with new loans had fallen sharply during this crisis period (Ivashina and Scharfstein, 2010). According to the decoupling hypothesis, the business cycle in Asian emerging economics is less correlated and less simultaneous with that of the OECD countries. However, the economic in Asian developing countries, especially China, are significantly affected by the 2008 financial crisis according to Fidrmuc and Korhonen (2010) and they document that the business frequency and dynamic correlations, though weak, influenced mostly by the financial shock.

(10)

9

decreased in a huge extent in the 2008 global financial crisis. Consistent with the former argument, Lardy and Subramanian (2011) also show that in China, due to the financial shock in 2008, both state ownership firms and private firms have experienced a significant fall in earnings and profitability and the GDP growth in 2009 has declined to 9.2%. The earnings and profitability of firms have a huge impact on the cash dividend payout policy. Fama and French (2001) find that firms pay less or no cash dividends always share the characteristics of less profitability. Additionally, they also suggest that the average earnings to asset ratio, a proxy of profitability, of dividend payers is 2.5% higher than that of non-payers from 1963 to 1998, indicating that more profitable firms tend to pay dividends. DeAnglo et al. (2004) show that the losses of firms are correlated with less dividend payouts. They conclude that earnings, to some extent, do have impact on dividend decisions. In emerging markets, the profitability, debt and market to book value could explain the cash dividend payout while to how much it would affect varies over country (Aivazian et al., 2003). Wang et al. (2011) find in China, the cash dividend respond quickly to earnings change. Thus, financial crisis will affect the cash dividend policy resulting from the profitability and earnings change.

(11)

10

out that the release of dividend may contain information about firm’s future sustainable earnings and if managers have predicted a bright future with sufficient cash flows, they tend to increase the payouts of cash dividend. Darling (1957) argues that the cash dividend policy will be adjusted by the change of expectation of management on future earnings. Based on the study of U.K firms, Benito and Young (2003) find that the change of cash dividend policy is a reflection of “Great Expectation” rather than “Hard Times” of firms. Firms do not pay or omit dividends tend to be less profitable. In the U.S, the expected dividend growth is significantly positive correlated to the expected stock return (Lettau and Ludvigson, 2005). They argue that the dividend forecasts are also affected by business cycle variation of expected stock return. Therefore, financial crisis will change the future expectation of firms and managers to be more pessimistic, leading to decline in cash dividends.

(12)

11

Chay and Suh (2009) use volatility of stock returns as a proxy of cash flow uncertainty to show that firms with uncertainty cash flows tend to pay less or omit cash dividends and this impact plays a stronger role than other factors such as investment opportunities and agency conflicts. However, Deng et al. (2013) find that Chinese firms do not cut investment and dividend when facing cash flow uncertainty. Thus, cash flow uncertainty caused by the financial crisis will partly have a negative impact on cash dividend payouts.

As Campello et al. (2010) suggest, firms will face capital constraint during the financial crisis and they will likely to cut cash dividend to ensure cash flow supporting their operating activities. Abreu and Gulamhussen (2013) also study the dividend policy of U.S bank holding companies during the financial crisis but they do not regard financial crisis as an individual impact factor. China is also influenced by the global financial crisis. Thus, the first hypothesis is developed:

H1: Financial crisis has a negative impact on cash dividend policy in China.

2.3.State Ownership and Cash Dividend

(13)

12

SOEs are less likely to face capital constraint. The capital constraint hypothesis argues that SOEs pay more cash dividends than NSOEs due to less capital constraint (Bradford et al., 2013). Most of the firms in China issue equities instead of issue bonds as the bond market is small (Huang and Song, 2006). Four big banks in China are state owned, indicating that most of the banking assets are under the control of state (Firth et al., 2008). It is more difficult for the NSOEs to finance long term debt from the bank (Brandt and Li, 2003). Because of serious information asymmetry and low property right protection in China, banks tend to prefer issuing debt to the SOEs and large firms since these firms may have lower debt risks and higher investment abilities (Deng et al., 2013). Banks will have a softer lending policy to the SOEs than that of NSOEs (Fan et al., 2008). Therefore, compared to SOEs, NSOEs are unlikely to get bigger loans and may face higher standards to get them. Additionally, listed firms are required to meet the performance and accounting standards regulated by the China Securities Regulatory Commission (CSRC) if they want to issuing new shares. However, sometimes the requirements can be negotiated if they have some reasonable explanations (Bradford et al., 2013). The SOEs are always able to get the approval of the CSRS to issue equities as exceptions without meeting the requirements since the CSRS is also an agency of the state. Green (2003) argues that NSOEs are at a weak position compared to SOEs to get right to finance through equities. With difficulties in external financing through debt and equity, NSOEs are more relying on internal financing. Chiou et al. (2010) suggest that SOEs are less financial constraint by showing that state ownership is negatively related to investment cash flow sensitivities. Fazzari et al. (1988) argue that firms facing capital constraint are likely to cut or omit dividends. Thus, with less capital constraint, SOEs tend to pay more cash dividend compared to NSOEs.

(14)

13

also find that state owned shares with a concentrated ownership have a positive impact on cash dividend. La Porta et al. (2000b) argue that due to high concentration of ownership, agency conflicts between majority shareholders and minority shareholders have been aggravated especially in countries with less investor protection. China is a country with less investor protection (La Porta et al., 2000a). The inside shareholders are more likely to expropriate the company assets from the outside minority shareholders and thus boost the value of cash dividend payouts (Opler et al., 1999). Firms tend to pay less cash dividend as an outcome of acquiring personal interests by cash holdings (La Porta et al., 2000b). In contrast, paying more cash dividend can also be viewed as a substitute of firm to achieve reputation in low investor protection country (La Porta et al., 2000b; Gomes, 2000). Controlling shareholders in China can pay more dividends to mitigate the ability of expropriation by managers. Johnson et al. (2000), La Porta et al. (1999b) and Lee and Xiao (2004) point out the dividend tunneling effect indicating that controlling shareholders could tunnel minority shareholders through cash or assets for their own benefits. Chen et al. (2009) argue that ownership concentration can let the firms to extract cash early financed from the IPOs, especially for SOEs which can own shares at a huge price discount. They further suggest that the higher ownership percentage of SOEs, the more bargaining power they have to develop dividend tunneling plans. Bradford et al., (2013) also show that the minority shareholders of the SOEs can have less influence on the decision than that of NSOEs. Liu et al. (2014) indicate that concentrated firms are likely to distribute more cash dividends to meet the cash needs of controlling shareholders, especially SOEs which always lack cash, since most of the cash dividends are released to them. Thus, due to high ownership concentration and incentive of tunneling, SOEs are likely to pay more cash dividends.

(15)

14

(16)

15

have some non-profit units such as primary schools and dining rooms and they both need cash flow to support their daily routines. Most of the shares held by NSOEs are tradable (Bradford et al., 2013). Thus, SOEs are likely to pay more cash dividends.

However, some contradict views also exist. First, Jensen and Meckling (1976) and Jensen (1986) argue that the payout of cash dividend can mitigate the agency conflicts between managers and shareholders. Dividend can be used as a corporate governance mechanism to reduce the free cash flows in managers’ hand in order to prevent them from abusing the cash earnings or investing in value decreased projects for compensations (Lang and Litzenberger, 1989). However, in China, managers of SOEs are appointed by the state and their compensation is not based on their investment performance and therefore the controlling shareholders can effectively monitor managers leading to less agency conflicts (Huang et al., 2011). The state controlling shareholders are powerful and can force mangers to pay cash dividend (Wei and Xiao, 2009). Thus, less cash dividend should be released in SOEs since there are less agency conflicts between managers and controlling shareholders. Second, the free cash flow hypothesis suggests that minority shareholders would prefer more cash dividends to mitigate agency conflicts (Lang and Litzenberg, 1989). Bradford et al., (2013) argued that minority shareholders of NSOEs can have a greater influence than that of SOEs, which are controlled by the government, on firm’s policies and dividend decisions, and will force corporate insiders to pay more dividends. Third, La Porta et al. (2000b) argue that paying less cash dividends is an outcome of controlling shareholders to acquire benefits from minority shareholders. However, the SOEs would like to expropriate benefits through dividend tunneling without keeping retained earnings (Lee and Xiao, 2004). Paying more dividends can be viewed as a substitute of firms to enhance reputation in markets with less investor protection (La Porta et al., 2000b). In contrast, SOEs are state controlled and do not need to enhance reputation for financing or increasing share price due to illiquidity (Liu et al., 2014).

(17)

16

contradict views on discussing the impact of state ownership on cash dividend policy. The tunneling hypothesis explains that SOEs are trying to expropriate private benefits from the minority shareholders through cash dividends. We expect that state ownership can have a positive impact on cash dividend policy because of less capital constraint, concentrated ownership and inherent illiquidity shares. Thus, the second hypothesis is developed:

H2: State ownership has a positive impact on cash dividend policy in China.

2.4.State Ownership and Cash Dividend During the Financial Crisis

The 2008 financial crisis has slowed down the China’s GDP growth which is regarded as a factor of social stability (Morrison, 2009). Several industries, such as the bank and export sectors, are seriously hit by the financial shock and many employees lose their jobs. SOEs are distinctive from the NSOEs, especially in their ownership structure, which will lead to different corporate governance and decision makings (Chen et al., 2009). Thus, the influence of financial crisis on cash dividend policy in China could be different between SOEs and NSOEs.

(18)

17

al., 2004). With huge impact of the crisis, the probability of NSOEs is low to get large loans from the banks as SOEs have the priority since state owned banks have to first fulfill the cash flow needs of SOEs. Therefore, with weak financing position during the crisis period, NSOEs will have more capital constraint than SOEs. Following the capital constraint hypothesis (Bradford et al., 2013), the difference of cash dividend payouts between SOEs and NSOEs will get bigger during the crisis period.

(19)

18

During the financial crisis, the attitude of managers to corporate policy decisions will be different between SOEs and NSOEs. Campello et al. (2010) suggest that managers will be more pessimistic with the future performance and expectations of firms due to the recession of macro-economic environment and tend to retain more earnings instead of cash dividend releasing. This will be true for NSOEs managers but will aggravate the agency conflicts between managers and shareholders. However, the managers in SOEs are appointed by the state with no performance requirements related to their compensations (Huang et al., 2011). They will follow the instructions of the state regarding decision makings and less monitored by minority shareholders. In addition, SOEs are strongly supported by the state with priority resource distributions (Su et al., 2014). Thus, managers of SOES will less be worried about the future performance compared to NSOEs. Still, the financial crisis has brought pressure to both SOEs and NSOEs. Gonenc and Aybar (2006) find evidence that firm in emerging markets with concentrated ownership have weak performance during the crisis period. Chiou et al. (2010) also argue that the percentage of state ownership has negative impact on firms’ performance. Both SOEs and NSOEs have incentives to expropriate private benefits from the minority shareholders. NSOEs pay less cash dividends to retain earnings and utilize free cash flows to meet their own needs (Lang and Litzenberger, 1989). Instead, the SOEs tend to expropriate benefits through more cash dividend payout since they inherently want to keep concentration control without acquiring capital gains from selling shares and tunneling incentives can be efficiently implemented by their managers (Liu et al., 2014). Thus, difference of cash dividend payout tends to be bigger between SOEs and NSOEs during the crisis period.

(20)

19

sticky in U.S (Nguyen and Wang, 2012). Chen et al. (2009) even find a negative announcement effect of increasing cash dividends. In the financial crisis period, firms always experience a difficult situation with capital constraint while releasing cash dividend can deliver a positive signal to the investors by increasing their confidence of the firms. Fuller and Goldstein (2011) find that monthly average stock return of firms paying cash dividends in declining markets will be 1% to 2% higher than that of firms paying cash dividends in advancing markets. NSOEs can sell the stocks when they witness an appreciation in share price through cash dividend payouts to gain cash flow during hard times (Lee and Xiao, 2004). Regarding the SOEs, they prefer to maintain ownership concentration with more controlling power and are not willing to sell shares (Wei and Xiao, 2009). Thus, there is less incentive for them to distribute cash dividends as a signaling tool to increase investors’ confidence during the crisis. In addition, regarded cash dividend payout as a tunneling inducement, SOEs are prevented from releasing cash dividends which may even have a negative impact on share price. Therefore, the difference of cash dividend policy between SOEs and NSOEs may be smaller during the crisis period.

Based on the above arguments, we try to test whether the difference of cash dividend policy between SOEs and NSOEs gets bigger during the crisis period. Therefore, the third hypothesis is developed:

H3: The interaction variable of financial crisis and state ownership has a positive

impact on cash dividend policy in China.

3. Data and Methodology 3.1.Data Sources

(21)

20

the dividend income tax rate of individuals was changed from 20% to 10% on June 13th, 2005 (Wang and Guo, 2011) and therefore the paper would like to exclude the effect of change in tax rate. Second, the split-share structure reform began in 2005 where non-tradable shares can be gradually converted into tradable shares (Wang et al., 2011) and thus the paper would like to exclude the effect of this policy change. Regarding the data directly got from the database, the paper sets up several benchmarks to select the efficient data points. First, firms that have unavailable data for a certain variable are all excluded. Second, 41 companies in the financial industries are excluded because of their special nature and different operations from others firms. Third, in order to exclude the outliers and abnormal data points, the paper wisorizes the data at 1% from the top and bottom and also manually exclude the abnormal data, for example, negative leverage ratio. After following these rules, the final sample includes 2402 firms from 5 different industries and 9053 firm-year observations are used for regression analysis.

3.2.Variables

This paper follows the approaches of Fama and French (2001), DeAngelo et al. (2006), Von Eije and Megginson (2008) and Bradford et al. (2013). The former three papers provide guideline for the general determinants of cash dividend payout policy based on the data collected from U.S and E.U markets. The last one has taken country-specific institution background into consideration to see what other factors may have impacts on cash dividend policy in China.

3.2.1. Dependent Variables

(22)

21

total dividend payout amount as proxies for cash dividend payout policy suggested by Bradford et al. (2013) and Wei et al. (2011). In addition, this paper uses total cash dividend amount divided by net sales as a proxy of cash dividend payout policy suggested by Liu et al. (2014) and Faccio et al. (2001) for robustness check.

3.2.2. Explanatory Variables

Figure 1

Note: ROA and ROE are collected from the CSMAR and data have been trimmed at 1% level to

exclude abnormal data points. The real GDP growths are collected from the World Bank website1.

Financial crisis is defined as a dummy variable, 1 stands for the crisis period and 0 for the non-crisis period. Campello et al. (2010) suggest that firms suffer from the credit constraint most severely in year 2008 and they have cut spending and dividends. Similarly, Huang et al. (2011) find that Chinese economic and firms are most influenced by the financial crisis in 2008. As shown in figure 1, the ROE and ROA of Chinese firms drop sharply in year 2008 while recover to the average level in 2009 and China real GDP growth also slips dramatically in 2008. Therefore, this paper defines the year 2008 as the crisis period in China. Kahle and Stulz (2013) indicate that the crisis starts from year 2007 to 2009 in the U.S. Fidrmuc and Korhonen (2010) argue that China is influenced by the U.S. economic during the crisis period. In figure 1, U.S. real GDP growth witnesses a declining trend from year 2007 to 2009 and 1 (http://data.worldbank.org/indicator/NY.GDP.MKTP.KD.ZG?page=1) -4.00% -2.00% 0.00% 2.00% 4.00% 6.00% 8.00% 10.00% 12.00% 14.00% 16.00% 2006 2007 2008 2009 2010 2011 2012 ROA Mean ROA Median ROE Mean ROE Median China Real GDP Growth

(23)

22

recovers in year 2010. Thus, this paper uses the crisis period from 2007 to 2009 for robustness check.

Many scholars such as Bradford et al. (2013) and Huang et al. (2011) find that firms with a state ownership pay more cash dividends than that of non-state ownership. Wei and Xiao (2009) and Chen et al. (2006) define the state ownership as firms directly controlled by government departments such as the National State-Owned Assets Administration Bureau (NSAAB), central governments, local governments and other SOEs or state-owned legal persons through state shares or state-owned legal person shares. Lin et al. (2010) argue that SOEs are always with a concentrated ownership. Instead of using dummy variable, this paper uses the percentage of largest shareholder ownership, generally (directly or indirectly) the state, to measure extent of state ownership as suggested by Wang et al. (2011).

To see the difference of cash dividend payout between SOEs and NSOEs during the financial crisis, the paper adds a third explanatory variable as financial crisis times state ownership. If the coefficient of this variable represents significantly positive, then hypothesis three will be confirmed indicating that the difference of cash dividend policy between SOEs and NSOEs gets bigger during the crisis period.

3.2.3. Control Variables

(24)

23

Growth opportunities: Based on E.U and U.S data respectively, Von Eije and Megginson (2008) and DeAngelo et al. (2006) find negative relationship between cash dividend payouts and firms growth, which is measured by market to book value,

sales growth or asset growth rate. This paper uses the market to book value, employed by Chen et al. (2009), to control for firm growth and investment opportunities.

Firm size: Fama and French (2001) find that larger firms are more likely to pay cash dividends. Regarding the Chinese market, Bradford et al. (2013) show that firm size is significantly positive correlated with cash dividend payouts. Different from the study of Von Eije and Megginson (2008) and DeAngelo et al. (2006), this paper defines the firm size as the natural logarithm of total assets at the end of year (Liu et al., 2014).

Cash holding: De Angelo et al. (2006) and Von Eije and Megginson (2008) use cash to total assets for controlling cash holdings and find that the more cash firms hold, the less likely they will pay cash dividends. However, Bradford et al. (2013) and Wei et al. (2011) find evidence that cash holding have a significant positive impact on cash dividend policy based on the study in Chinese markets. Thus, the impact of cash holding is ambiguous and this paper uses cash to total assets as a proxy of liquidity and cash holdings.

Leverage: Most of the research on cash dividend policy includes leverage ratio as a control for capital structure. DeAngelo et al. (2006) use total equity to total assets while Von Eije and Megginson (2008) employ total debt to total assets by arguing that leverage can help control for agency cost as more debt will reduce the need of cash releasing to shareholders. Further, Wei and Xiao (2009) argue that the debt of firm restrict the dividend payouts. Following the study of Chen et al. (2009) and Huang et al. (2009), this paper uses total debt to total asset to control for capital structure.

(25)

24

benefits or overinvest instead of distributing through dividends to shareholders. Lang and Litzenberger (1989) state that cash dividend increase of overinvesting firms will have a positive impact on firm’s market value and can deliver information about their investment policies. Following the study of Bradford et al. (2013) and Lee and Xiao (2004), this paper uses free cash flow, calculated as net cash flow from operating activities less cash expenditures on the acquisition of fixed assets, intangible assets and other long-term assets, divided by total assets as a proxy to control for agency problem and signaling incentive.

Lifecycle: DeAngelo et al. (2006) test the lifecycle theory by using retained earnings to total equity or retained earnings to total assets as proxy indicating, that these two ratios are positively related to the propensity of cash dividend. Von Eije and Megginson (2008) suggest that younger firms are less likely to raise fund and always face many investment opportunities thus they include the age of the firm in addition to fraction of retained earnings to total equity. Only a few studies in China such as Wei et al. (2011) apply the natural logarithm of firm age to measure the maturity of firms. This paper uses the retained earnings to total assets to control for lifecycle effect.

Income risk: Von Eije and Megginson (2008) suggest that firms are less likely to pay cash dividends when they have higher income risk and thus they include the standard deviation of net income during the past 5 years related to actual annual sales to control for risk. Similarly, Wei and Xiao (2009) and Lie (2005) document that with higher business risks, firm will pay less cash dividends and use the standard deviation of past three years’ EBIT as a proxy. This paper follows the measurement by Von Eije and Megginson (2008) to control for cash flow uncertainty and risk but only use 3 years for standard deviation.

(26)

25

find significant relation between current year cash dividend payout and lag year cash dividend payout, which can also be found in studies using Chinese market (Lam et al., 2012; Lee and Xiao, 2004). This paper uses the lag year dependent variable to control for the impact of lag cash dividend policy.

Current stock dividend payout: Stock dividends are rarely paid in the U.S (Skinner and Soltes, 2011). Firms in U.S are more likely to use stock splits or share repurchase as a substitute of stock dividends. However, stock splits are forbidden in China (Anderson et al., 2011). Chen et al. (2002) find that there is a positive market reaction of stock dividend releasing while little market reaction of cash dividend distribution. Stock dividend can be regarded as a substitute of cash dividend and need to be controlled (Chen et al., 2009; Wei and Xiao, 2009). This paper employs a dummy variable which 1 stands for firms pay stock dividend in the current year while 0 for firms do not.

This paper also uses dummy variables to control for years and industries. After excluding the financial industry, four dummy variables are used to control for five different industries. Five dummy variables are used to control for seven years since one year has already been defined as the year of financial crisis. The summary of variables can be found in Table 1.

3.3.Research Design

The paper uses three regression models, which are commonly implemented in much research such as Wei et al. (2011) and Bradford et al. (2013), to test the three hypotheses.

(27)

26

Table 1 Variable definitions

Variables Symbol Definition

Cash dividend propensity CDP 1 for cash dividend payers and 0 for non-payers

Cash dividend payout ratio CDR Cash dividend per share to earnings per share

Total cash dividend amount DIV Natural logarithm of total cash dividend amount

Financial crisis FC 1 for crisis period, and 0 for non-crisis period

State ownership SO The percentage of largest shareholder ownership

Difference of cash dividend policy between SOEs and NSOEs during financial crisis

FC.SO The result of financial crisis times state ownership

Profitability PROF Net income to total asset

Growth opportunity GROW Market to book value

Firm size SIZE Natural logarithm of total asset

Cash holding CASH Cash holding to total asset

Leverage LEV Total debt to total asset

Free cash flow FCF Free cash flow to total asset

Lifecycle LIFE Retained earnings to total asset

Income risk RISK Standard deviation of EBIT relative to total asset

from t-2 to t

Lag cash dividend LCD Lag year dependent variable

Current stock dividend CSD 1 for firms paying stock dividends in the current

year, and otherwise 0

Industry INDUSTRY Four dummies controlling for five industries

Year YEAR Five dummies controlling for seven years

fiscal year while 0 for non-payers.

Logit (CDP) = 𝛼0+ 𝛼1FCij+ 𝛼2SOij+ 𝛼3FCij. SOij+ 𝛼4PROFij+ 𝛼5GROWij

+ 𝛼6SIZEij+ 𝛼7CASHij+ 𝛼8LEVij+ 𝛼9FCFij+ 𝛼10LIFEij+ 𝛼11RISKij

+ 𝛼12LCDij+ 𝛼13CSDij+ 𝛼14∑ INDUSTRYij+ 𝛼15∑ YEARij+ 𝑢𝑖𝑡

(28)

27

Tobit (CD) = 𝛼0+ 𝛼1FCij+ 𝛼2SOij+ 𝛼3FCij. SOij+ 𝛼4PROFij+ 𝛼5GROWij

+ 𝛼₆SIZE_ij+ 𝛼₇CASH_ij+ 𝛼₈LEV_ij+ 𝛼₉FCF_ij+ 𝛼₁₀LIFE_ij+ 𝛼₁₁RISK_ij

+ 𝛼₁₂LCD_ij+ 𝛼₁₃CSD_ij+ 𝛼₁₄∑ INDUSTRY_ij+ 𝛼₁₅∑ YEAR_ij+ 𝑢_𝑖𝑡

4. Empirical Results and Analysis 4.1.Descriptive Statistics

Table 2 Descriptive statistics

Descriptive statistics of the variables including A-share listed firms from 2006 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount and CSD refers to current stock dividend.

Variables Overall Sample Cash dividends

=0 >0

N=9053 N=3962 N=5091

Mean Median Maximum Minimum Std. Mean Mean

CDP 0.56 1.00 1.00 0.00 0.50 0.00 1.00 CDR 0.15 0.00 1.00 0.00 0.22 0.00 0.26 DIV 9.92 16.10 24.24 0.00 8.83 0.00 17.64 FC 0.13 0.00 1.00 0.00 0.33 0.15 0.11 SO 0.36 0.35 0.89 0.02 0.15 0.33 0.39 FC.SO 0.04 0.00 0.84 0.00 0.13 0.05 0.04 PROF 0.04 0.04 0.53 -0.82 0.06 0.01 0.06 GROW 1.82 1.46 11.32 0.37 1.14 1.93 1.74 SIZE 21.75 21.62 28.41 17.88 1.23 21.33 22.08 CASH 0.16 0.12 0.92 0.00 0.12 0.13 0.18 LEV 0.50 0.51 1.00 0.01 0.20 0.54 0.46 FCF 0.04 0.06 0.98 -2.52 0.16 0.03 0.05 LIFE 0.09 0.12 0.80 -5.12 0.26 -0.02 0.18 RISK 0.07 0.02 2.00 0.00 0.16 0.11 0.03 LCDP 0.54 1.00 1.00 0.00 0.50 0.21 0.79 LCDR 0.18 0.03 1.00 0.00 0.23 0.07 0.26 LDIV 9.48 15.84 24.24 0.00 8.83 3.61 14.04 CSD 0.06 0.00 1.00 0.00 0.23 0.01 0.10

(29)

28

the mean of cash dividend propensity (CDP) is 0.56, indicating that over half of the total observation firms pay cash dividends. However, the payout ratio (CDR) is still low where firms averagely use only 15% of their total earnings to pay cash dividends. The mean of financial crisis (FC) is approximately one-seventh of 1 as the paper only includes 2008 as the crisis year among the seven-year analysis period. The overall state ownership (SO) shows that ownership is highly concentrated where largest shareholders own nearly 36% of the whole firms and most of the SOEs in China have a concentrated ownership (Wang et al., 2011). Regarding the separate samples, the results show that, on average, firms which pay cash dividends are less suffered from financial crisis (FC) and have higher state ownership (SO), higher profitability (PROF), lower growth opportunity (GROW), larger firm size (SIZE), more cash holding (CASH), lower leverage ratio (LEV), higher retained earnings ratio (LIFE) and lower income risk (RISK) than firms which do not pay cash dividends. The mean of variables referring to interaction between SOEs and NSOEs during crisis (FC.SO) and free cash flow (FCF) are less significantly different between cash dividend payers and non-cash dividend payers. The descriptive statistics only provide a common illustration of the sample and further analysis will be done through several regression models to briefly see how each factor affects cash dividend payout.

4.2.Correlation Analysis

(30)

29

Table 3 Correlation matrix

Correlation matrix of the variables including A-share listed firms from 2006 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount and CSD refers to current stock dividend.

CDP CDR DIV FC SO FC.SO PROF GROW SIZE

CDP 1.00 CDR 0.60 1.00 DIV 0.99 0.61 1.00 FC -0.05 -0.03 -0.06 1.00 SO 0.17 0.16 0.19 -0.01 1.00 FC.SO -0.02 0.00 -0.02 0.91 0.13 1.00 PROF 0.39 0.23 0.41 -0.06 0.11 -0.03 1.00 GROW -0.08 -0.06 -0.08 -0.17 -0.15 -0.17 0.20 1.00 SIZE 0.30 0.17 0.36 -0.05 0.28 -0.01 0.11 -0.38 1.00 CASH 0.18 0.14 0.18 -0.05 0.00 -0.05 0.28 0.18 -0.13 LEV -0.19 -0.19 -0.18 0.02 0.02 0.02 -0.34 -0.26 0.34 FCF 0.08 0.14 0.08 0.02 0.07 0.02 0.09 -0.02 0.04 LIFE 0.39 0.24 0.39 -0.04 0.12 -0.02 0.37 -0.16 0.27 RISK -0.25 -0.17 -0.25 0.03 -0.08 0.01 -0.14 0.13 -0.21 LCDP 0.58 0.60 0.59 -0.04 0.18 -0.01 0.32 -0.07 0.30 LCDR 0.39 0.47 0.40 -0.02 0.15 0.01 0.19 -0.06 0.15 LDIV 0.59 0.62 0.61 -0.04 0.20 -0.01 0.34 -0.07 0.36 CSD 0.18 0.04 0.17 -0.01 0.00 -0.02 0.15 0.11 0.04

CASH LEV FCF LIFE RISK LCDP LCDR LDIV CSD

CASH 1.00 LEV -0.41 1.00 FCF -0.18 0.01 1.00 LIFE 0.16 -0.31 0.18 1.00 RISK -0.08 0.00 -0.12 -0.37 1.00 LCDP 0.18 -0.17 0.12 0.36 -0.25 1.00 LCDR 0.12 -0.16 0.10 0.24 -0.18 0.46 1.00 LDIV 0.17 -0.17 0.13 0.37 -0.24 0.99 0.47 1.00 CSD 0.04 0.01 -0.02 0.10 -0.04 0.07 0.05 0.06 1.00

(31)

30

Table 4 Regression results

Regression results of the models including A-share listed firms from 2006 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount, CSD refers to current stock dividend, CONSTANT refers to constant, INDUSTRY refers to controlling for industries and YEAR refers to controlling for years.

Whole sample Model 1: Logit（CDP） Model 2:Tobit（CDR） Model 3:Tobit（DIV）

Variables Coefficient P value Coefficient P value Coefficient P value

FC -0.64 0.007 -0.06 0.078 -2.40 0.008 SO 0.61 0.005 0.10 0.001 2.23 0.004 FC.SO 0.07 0.903 -0.03 0.715 0.34 0.873 PROF 15.59 0.000 1.12 0.000 49.52 0.000 GROW -0.35 0.000 -0.04 0.000 -1.62 0.000 SIZE 0.35 0.000 0.06 0.000 1.40 0.000 CASH 0.80 0.007 0.17 0.000 2.32 0.021 LEV -0.46 0.052 -0.18 0.000 -1.44 0.097 FCF -0.28 0.190 0.18 0.000 -2.52 0.001 LIFE 5.60 0.000 0.58 0.000 22.22 0.000 RISK -4.71 0.000 -1.15 0.000 -22.11 0.000 LCDP 1.77 0.000 LCDR 0.60 0.000 LDIV 0.55 0.000 CSD 2.18 0.000 0.05 0.001 5.51 0.000 CONSTANT -8.09 0.000 -1.28 0.000 -28.26 0.000

INDUSTRY YES YES YES

YEAR YES YES YES

Pseudo R2 0.43 0.38 0.13

N 9053 9053 9053

4.3.Regression Results

The results of logit and tobit regressions of whole sample are presented in Table 4. The paper uses logit regression to see the likelihood of firms paying cash dividends affected by different factors and uses tobit regressions to see how these factors influence the real cash dividend payout policy.

(32)

31

cash dividends and during the crisis period, the propensity of firms to pay cash dividends is lower. The former implication is supported by the significant of coefficients at at least 10% level. Cash dividend policy of the former year and stock dividend policy of the current year also have a positive impact on the decision of current year’s cash dividend policy. However, it is still unclear whether the difference of cash dividend policy get bigger between SOEs and NSOEs during the crisis period and free cash flow could not fully explain why firms distribute cash dividends.

The regression models in the second and third column document how each factor affects the real cash dividend payout policy. Firms which have higher state ownership, higher profitability, lower growth opportunity, larger firm size, more cash holding, less debt, more retained earnings, lower income risk and operate during the non-crisis period have higher cash dividend payout ratio and release more total amount of cash dividends. Current stock dividend policy and lag year cash dividend policy both have a positive impact on current year cash dividend policy. The coefficient of variable (FC.SO) is also not statistically significant from 0, indicating that this variable has less explanatory power for the cash dividend policy. The free cash flow variables (FCF) are both significant in the two models while with opposite sign and thus the impact of free cash flow cannot be completely deduced.

(33)

32

ownership (SO) are positive and highly significant at 1% level in all three models. SOEs in China are less likely to face capital constraint than NSOEs (Bradford et al., 2013). They usually have a concentrated ownership and have incentive to tunnel through cash dividends (Lee and Xiao, 2004). The inherent illiquidity of shares owned by SOEs, leading to lack of operating cash flow, results in the distribution of cash dividends (Liu et al., 2014). The result shows that firms with stronger state ownership tend to pay more cash dividends, which is consistent with the study of Bradford et al. (2013), Huang et al. (2011) and Lin et al. (2010). Thus, hypothesis two is supported. The coefficients of the variable (FC.SO), which measure the difference of cash dividend policy between SOEs and NSOEs during the crisis period, are totally not significant and also with different signs in the following three models. There is no evidence to show that the difference of cash dividend policy between SOEs and NSOEs become bigger during the crisis period and thus hypothesis three is not supported. The capital constraint hypothesis, inherent illiquidity of shares and managers’ attitude change arguments may explain why the difference of cash dividend policy between SOEs and NSOEs will get bigger during the crisis period. However, difference can also be mitigated supported by the signaling argument. This could provide explanation why the interaction variable is not statistically significant.

(34)

33

three models, indicating that firm with sufficient cash will pay more cash dividends and study by Wei et al. (2011), Lam et al. (2012) and Wei and Xiao (2009) find similar results. With more debt holding, the agency costs will be controlled and firms need to pay cash to the debt holders first thus leading to reduce of cash dividend release (Von Eije and Megginson, 2008). The leverage ratio (LEV) has a significantly negative impact on cash dividend distribution, which is also confirmed by Su et al. (2014) and Wei et al. (2011). Consistent with the lifecycle theory, DeAnglo et al. (2006) and Von Eije and Megginson (2008) finds that well-established firms with high retained earnings ratio are likely to distribute more cash dividends. This view has been proved since the lifecycle variable (LIFE) is highly significant at 1% level with positive signs. The larger the income risk is, the less cash dividends will be distributed (Von Eije and Megginson, 2008; Wei and Xiao, 2009). First pointed out by Lintner (1956), the lag year dividend policy will have an impact on current year’s dividend decision. The coefficients of lag year cash dividends release are all significant with positive signs, which are consistent with DeAnglo et al. (2006) results. However, the impact of free cash flow (FCF) is ambiguous since the coefficients are with different signs and insignificant in each model. As suggested by Jensen (1986), more free cash flow will increase the expropriation of managers, leading to less cash dividends distribution. The results are not sufficient to support this view, where Huang et al. (2011) also do not find clear evidence. Contrast to the view of Chen et al. (2009) and Wei and Xiao (2009) who suggest that stock dividends can be a substitute of cash dividends, the results show that currents stock dividend has a positive impact on cash dividend release, indicating that when firm decide to pay stock dividends, they also tend to pay cash dividends.

4.4.Robustness Test

(35)

34

Table 5 Robustness check with different definition of crisis period

Robustness check for different period including A-share listed firms from 2006 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount, CSD refers to current stock dividend, CONSTANT refers to constant, INDUSTRY refers to controlling for industries and YEAR refers to controlling for years.

Logit（CDP） Tobit（CDR） Tobit（DIV）

2007-2009 as financial crisis period

YEAR YES YES YES

Pseudo R2 0.43 0.38 0.13

N 9053 9053 9053

(36)

35

Table 6 Robustness check with sub-period sample

Robustness check for sub-period including A-share listed firms from 2008 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount, CSD refers to current stock dividend, CONSTANT refers to constant, INDUSTRY refers to controlling for industries and YEAR refers to controlling for years.

Logit（CDP） Tobit（CDR） Tobit（DIV）

Using sample from 2008 to 2012

YEAR YES YES YES

Pseudo R2 0.43 0.37 0.12

N 6902 6902 6902

(37)

36

Table 7 Robustness check with firms only listed in Chinese A-share market

Robustness check for firms only A-share listed from 2006 to 2012. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount, CSD refers to current stock dividend, CONSTANT refers to constant, INDUSTRY refers to controlling for industries and YEAR refers to controlling for years.

A-share listed firms Logit（CDP） Tobit（CDR） Tobit（DIV）

YEAR YES YES YES

Pseudo R2 0.43 0.38 0.13

N 8567 8567 8567

two are still supported.

(38)

37

Table 8 Robustness check with different proxies

Robustness check with different proxies including A-share listed firms from 2006 to 2012. DIV/TS refers to cash dividend to total sales. CDP refers to cash dividend propensity, CDR refers to cash dividend payout ratio, DIV refers to total cash dividend amount, FC refers to financial crisis, SO refers to state ownership, FC.SO refers to financial crisis times state ownership, PROF refers to profitability, ROE refers to roe, GROW refers to growth opportunity, SIZE refers to firms size, CASH refers to cash holding, LEV refers to leverage, FCF refers to free cash flow, LIFE refers to lifecycle, RISK refers to income risk, LCDP refers to lag cash dividend propensity, LCDR refers to lag cash dividend payout ratio, LDIV refers to lag total cash dividend amount, LDIV/TS refers to lag cash dividend to total sales, CSD refers to current stock dividend, CONSTANT refers to constant, INDUSTRY refers to controlling for industries and YEAR refers to controlling for years.

Whole sample Tobit（DIV/TS） Logit（CDP） Tobit（CDR） Tobit (DIV)

Variables C P value C P value C P value C P value

FC -0.01 0.035 -0.64 0.008 -0.06 0.093 -2.26 0.011 SO 0.01 0.025 0.59 0.006 0.10 0.001 2.11 0.006 FC.SO 0.00 0.825 0.13 0.827 -0.04 0.672 0.35 0.868 PROF 0.19 0.000 ROE 7.00 0.000 0.68 0.000 27.80 0.000 GROW 0.00 0.000 -0.29 0.000 -0.03 0.000 -1.49 0.000 SIZE 0.01 0.000 0.37 0.000 0.05 0.000 1.40 0.000 CASH 0.02 0.000 0.83 0.005 0.17 0.000 2.30 0.021 LEV -0.02 0.000 -1.41 0.000 -0.28 0.000 -5.47 0.000 FCF 0.00 0.815 -0.28 0.189 0.18 0.000 -2.37 0.002 LIFE 0.06 0.000 5.97 0.000 0.57 0.000 22.75 0.000 RISK -0.12 0.000 -4.56 0.000 -1.14 0.000 -21.83 0.000 LCDP 1.79 0.000 LCDR 0.04 0.000 LDIV 0.54 0.000 LDIV/TS 0.67 0.000 CSD -0.01 0.971 2.16 0.000 0.04 0.006 5.10 0.000 CONSTANT -0.15 0.000 -8.11 0.000 -1.22 0.000 -26.81 0.000

INDUSTRY YES YES YES YES

YEAR YES YES YES YES

Pseudo R2 0.35 0.43 0.39 0.13

N 9661 8958 8958 8958

(FC.SO) are still not statistically different from 0.

5. Conclusion

(39)

38

influence of financial crisis and state ownership on cash dividend policy in China.

We find that financial crisis has a negative impact on cash dividend policy. The 2008 global financial crisis has severely affected the world economy leading to depression. Firms experience low profitability and even negative earnings. Thus, they have to focus on sustaining their daily operating activities and do not have sufficient income to distribute cash dividends. Such situation will also affect the expectations of management on firms’ future performance and they tend to retain more earnings instead of paying them out. Uncertainty of income cash flow and difficulty for financing during this certain period will also largely decrease the likelihood of managers to release cash dividends. The Chinese market is influenced by the global economy and thus firms in China tend to reduce cash dividends distribution during the crisis period.

(40)

39 2013; Su et al., 2014; Wei and Xiao, 2011).

We expect that the difference of cash dividend policy between SOEs and NSOEs will get bigger during the crisis period. SOEs are not facing as much capital constraint as NSOEs during the crisis period. They need working capitals to support their non-profit units, especially during crisis, leading to the increase of tunneling incentives and managers of SOEs afford less pressure than NSOEs. However, we do not find much evidence to support this argument as the result of this variable is not statistically significant.

Our paper makes important contribution to the existing literature. First, Campello et al. (2010) study the impact of financial crisis on cash dividend by collecting questionnaires from CEOs and from a qualitative perspective. For this paper, we use several regression models to prove that financial crisis negatively influence the cash dividend policy from a quantitative perspective and build theoretical structure to support this argument. In addition, this is the first study to separately investigate the influence of financial crisis on cash dividend by using data from China, an emerging market. Second, the influence of state ownership on cash dividend policy has been largely discussed by several scholars (Su et al., 2014; Bradford et al., 2013). Our study integrates different arguments and provides a holistic view to explain why SOEs tend to pay cash dividends. We also try to measure to what extent state ownership impact the cash dividend by using the percentage of largest shareholders as a proxy. Additionally, we also try to analyze interaction of the impact of financial crisis and state ownership though we do not find sufficient evidence to support hypothesis three.

(41)

40

take more consideration on choosing to buy proper shares during the crisis periods. Second, as can be observed, a lot of firms which used to be privatized are now owned by the state after the financial crisis in the western countries such as General Motor and Merrill Lynch. The investigation of state ownership can help managers better understand the role of government and develop suitable cash dividend policy. Investors, who prefer cash dividends, could first consider to buy shares of state owned firms, others things being equal.

(42)

41 Reference

Abreu, J. F., & Gulamhussen, M. A. (2013). Dividend payouts: Evidence from US bank holding companies in the context of the financial crisis. Journal of Corporate Finance, 22, 54-65.

Aivazian, V., Booth, L., & Cleary, S. (2003). Do emerging market firms follow different dividend policies from US firms?. Journal of Financial Research, 26(3), 371-387.

Allen, F., Bernardo, A. E., & Welch, I. (2000). A theory of dividends based on tax clienteles. The Journal of Finance, 55(6), 2499-2536.

Al-Malkawi, H. A. N., Rafferty, M., & Pillai, R. (2010). Dividend policy: a review of theories and empirical evidence. International Bulletin of Business Administration, 9, 171-200.

Anderson, H. D., Chi, J., Ing-aram, C., & Liang, L. (2011). Stock dividend puzzles in China. Journal of the Asia Pacific economy, 16(3), 422-447.

Baker, H. K. (Ed.). (2009). Dividends and dividend policy (Vol. 1). Wiley. com.

Baker, M., & Wurgler, J. (2004). Appearing and disappearing dividends: The link to catering incentives. Journal of Financial Economics, 73(2), 271-288.

Barth, J. R., Caprio Jr, G., & Levine, R. (2004). Bank regulation and supervision: what works best?. Journal of Financial intermediation, 13(2), 205-248.

Benito, A., & Young, G. (2003). Hard Times or Great Expectations? Dividend Omissions and Dividend Cuts by UK Firms*. Oxford bulletin of economics and statistics, 65(5), 531-555.

Berger, A. N., & Bouwman, C. H. (2013). How does capital affect bank performance during financial crises?. Journal of Financial Economics, 109(1), 146-176.

Bernanke, B. S. (1983). Non-monetary effects of the financial crisis in the propagation of the Great Depression.

Bhattacharyya, N. (2007). Dividend policy: a review. Managerial Finance, 33(1), 4-13.

Black, Fischer, 1976, The Dividend Puzzle, Journal of Portfolio Management 2, 5-8. Bradford, W., Chen, C., & Zhu, S. (2013). Cash dividend policy, corporate pyramids, and ownership structure: Evidence from China. International Review of Economics & Finance, 27, 445-464.

(43)

42

Brennan, M. J. (1970). Taxes, market valuation and corporate financial policy.National Tax Journal, 23(4), 417-427.

Campbell, J. Y., Giglio, S., & Polk, C. (2013). Hard times. Review of Asset Pricing Studies, 3(1), 95-132.

Campello, M., Graham, J. R., & Harvey, C. R. (2010). The real effects of financial constraints: Evidence from a financial crisis. Journal of Financial Economics, 97(3), 470-487.

Chay, J. B., & Suh, J. (2009). Payout policy and cash-flow uncertainty. Journal of Financial Economics, 93(1), 88-107.

Chen, D. H., Liu, H. H., & Huang, C. T. (2009). The Announcement Effect of Cash Dividend Changes on Share Prices: An Empirical Analysis of China. chinese economy, 42(1), 62-85.

Chen, D., Jian, M., & Xu, M. (2009). Dividends for tunneling in a regulated economy: The case of China. Pacific-Basin Finance Journal, 17(2), 209-223.

Chen, G., Firth, M., & Gao, N. (2002). The information content of concurrently announced earnings, cash dividends, and stock dividends: an investigation of the Chinese stock market. Journal of International Financial Management & Accounting, 13(2), 101-124.

Chen, G., Firth, M., Gao, D. N., & Rui, O. M. (2006). Ownership structure, corporate governance, and fraud: Evidence from China. Journal of Corporate Finance, 12(3), 424-448.

Chen, G., Firth, M., & Xu, L. (2009). Does the type of ownership control matter? Evidence from China’s listed companies. Journal of Banking & Finance, 33(1), 171-181.

Cheng, L. T., Fung, H. G., & Leung, T. Y. (2009). Dividend preference of tradable‐ share and non-tradable shareholders in Mainland China. Accounting & Finance, 49(2), 291-316.

Chiou, J. R., Chena, Y. R., & Huangb, T. C. (2010). Assets Expropriation via Cash Dividends? Free Cash Flow or Tunneling. China Journal of Accounting Research, 3(1), 71-93.

Claessens, S., Kose, M. A., & Terrones, M. (2011). Financial Cycles: What? How? When?. International Monetary Fund.

Darling, P. G. (1957). The influence of expectations and liquidity on dividend policy. The Journal of Political Economy, 65(3), 209-224.