Impact of Board Characteristics and State Ownership on Dividend Policy in Vietnam*

The relationship between board characteristics and state ownership on dividend policy has been carefully examined in the literature in developed and emerging markets. This paper would seem to be the first to analyse the impacts of board gender diversity and state ownership in the Vietnam market for the following: (1) the number of female directors on the boards of directors is about 25%, which is the highest among South-East Asian countries; and (2) Vietnam has not fully transformed into a market economy, so the state ownership in the listed firm is apparent. Therefore, it is worth trying to examine the impacts of board gender diversity and state ownership on dividend policy in Vietnam. This paper illustrates that the female CEO and female CEO duality decrease the dividend ratio, while chairwomen tend to use dividend ratio as a tool to manage the company. Moreover, Boards of Directors (BODs) with more than three female members usually decrease the dividend payout ratio. The empirical findings are consistent with agency theory, as managers increase payouts to reduce free cash flow and agency costs in firms. The results also indicate that there is an inverse relationship between state ownership and the dividend payout ratio. It is due to higher state ownership allows firms to access external debts easily, so these firms pay high dividends to reduce agency costs. Moreover, governments perceive dividends from holding capital in listed firms as a source of revenue for the state budget. Therefore, firms with higher state ownership are more likely to increase dividend payouts.


Introduction
Recently, various gender diversity studies have focused on the effects of having a female CEO or female board member on firm performance (Ahern and Dittmar, 2012;Dezsö and Ross, 2012;Matsa and Miller, 2013), as well as risk-taking behaviour (Faccio et al., 2016). Furthermore, recent research has suggested that female executives can have a strong influence on specific governance policies. Shaukat et al. (2016) argue that female directors concentrate on corporate social responsibility, Miller and Triana (2009) find that female executives also invest more in research and development programs, while Faccio et al. (2016) note that female executives take fewer debts and make less risky financial decisions.
There is also evidence that female leaders tend to exert a more positive impact on the workplace environment than do their male colleagues (Carter et al., 2003;Adams and Ferreira, 2009), and that the quality of conference discussion on complex situations is also enhanced as a result of the presence of female directors.
The relationship between having a female director and the stock dividend yield has been intensively examined in the finance industry. Chen et al. (2017)  In striking contrast, Saeed and Sameer (2017) point out that having an additional female board member would lead to lower dividend yields in three developing markets, namely India, China, and Russia, based on data from 552 countries between 2007-2014. In addition to gender diversity, state ownership also has an impact on dividend policy. Berle and Means (1991) suggest that state ownership encourages managers to make decisions that may not maximize the shareholder value. Shleifer and Vishny (1986), McConnell and Servaes (1990), Zingales (1994) suggest that there is a strong relationship between the ownership structure and firm performance in the US and other financial markets. Duygun et al. (2018) analyzed the dividend policy of 369 companies from Indonesia, and the results showed that government-owned companies pay higher dividends than do family-owned companies. Yung (2011), Wang et al. (2011) also suggests that firms with higher state ownership have a higher likelihood of paying more dividends. Le and Le (2017) examine 180 listed companies on the Vietnam stock exchange between 2009-2013, and their empirical results show that companies paid higher dividends when the government was the largest shareholder.
On the other hand, Ben-Nasr (2015) observed a negative relationship between dividend policy and state ownership by analyzing data from 43 multinational corporations between 1985-2007. The author also indicates that state-owned enterprises tend to over-invest instead of paying dividends or buy back shares.
The extant literature examines the empirical relationship between either focus on gender diversity or state ownership on dividend policy. Moreover, they provide mixed findings on the impact of gender diversity and state ownership on stock dividend policy. A report from the Boston Consulting Group in 2017 indicates the number of female directors on the board of directors is about 25%, which is the highest among South-East Asian countries. Therefore, it is worth examining the impact of female directors on dividend policy in Vietnam.
As Vietnam has not yet fully transformed into a market economy, state ownership in listed firms is apparent. Thi paper uses data from 514 companies listed on the Vietnam stock exchange between 2007-2017. The empirical findings from the paper emphasizes how the growing gender diversity on the BOD and state ownership could affect the dividend policy of listed companies in Vietnam.

Board Gender Diversity and Stock Dividend Policy
Al-Amarneh et al. (2017) examine the effect of gender diversity on the dividend policy of 13 listed companies in the Amman Stock Exchange between 2005-2014. The research showed that BOD with gender diversity pays higher cash dividends because female directors could settle investor requirements better. However, the BOD tnds to become more conservative during a financial crisis, so they will retain capital and reduce dividends because female executives have a strategy of taking less risk. Adams and Ferreira (2009) suggest that female CEOs attend conferences more frequently and were more diligent than male CEOs. Eckel and Füllbrunn (2015) recommend that women are more cautious than their male counterparts, so their decisions are less radical and do not show overconfidence. Companies with more female CEOs have lower financial leverage ratios because female executives tend to engage in lower debts. and also select less risky investment decisions and portfolios (Faccio et al. 2016). Arshad et al. (2013) and Obradovich and Gill (2013) suggest that CEO duality had a significant impact on dividend policy, but they do not show how CEO gender might play a role in dividend decisions. Román-Martínez et al. (2012) claim that female CEOs could use their managing ability to cooperate with managers for personal benefits,, in which case stock dividends are reduced. Chen et al. (2017) study the role of gender in BOD on the stock dividend policy of companies listed on the US stock exchange between 1997-2001. The empirical findings show that companies with female CEOs had a higher dividend payouts, as they tended to stock dividends as a monitoring tool in companies with weak corporate governance. The authors emphasized the influence of individual female CEOs on dividend yields. Ye et al. (2019) evaluated the correlation between gender diversity in BOD and stock dividends.
The empirical results suggest that BOD with gender diversity use stock dividends to reduce agency costs. Specifically, both the number and ratio of female members in BOD affect dividend policies.

Capital Structure and stock Dividend
Chen and Dhiensiri (2009), Wang and Yung (2011), Bradford et al. (2013) examine the impact of state ownership on the dividend policy of listed firms in China. Chen and Dhiensiri (2009) and Wang and Yung (2011) find a positive relationship between state ownership and dividend payouts, and argue that firms with higher state ownership could raise external capital quickly, so they tend to increase payouts. Similarly, Bradford et al. (2013) shows that state-owned enterprises paid higher dividends than private companies in China by explaining that the private sector had financing constraints, so they would pay fewer dividends.
Similarly, Duygun et al. (2018) show that companies with higher state ownership pay greater dividends because the Indonesian Government considers dividends from state ownership as an essential source of revenue, apart from the unavoidable corporate tax. On the other hand, Ben-Nasr (2015) indicates the negative relationship between State ownership and dividend yield, and shows that the negative impacts are more robust, especially in countries with weak protection mechanisms for investors.

Hypotheses
According to Chen et al. (2017), and Ye et al. As the relationship between board gender diversity and dividend payout is mixed, we propose the following testable hypotheses: HA0: Gender diversity in the BOD does not influence dividend payouts; HA1: HA0 is not correct.
Gender diversity may affect the efficiency of the board of management at both the personal and group levels. At the individual level, it is understood that women and men are different in how they can improve the efficiency of the BOD. For example, Adams and Ferreira (2009) observe that female CEOs are more diligent and joined conferences more frequently than did their male counterparts. According to Bernardi and Arnold (1997), male executives tend to pay greater attention to their incomes and career promotion opportunities, and were more likely to bypass regulations to achieve success, while female leaders preferred to follow laws and principles. Gul et al. (2011) recognizes that female directors are more sensitive to moral and ethical issues.
Moreover, women have a lower risk preference than do males (Charness and Gneezy, 2012).
From these observations, we propose a second hypothesis: HB0: There are no differences between males and females in dividend decisions; HB1: HB0 is not correct.

Researchers have mixed findings on the relationship between State ownership and dividends. Nizar
Al-Malkawi (2007) and Ben-Nasr (2015) argue that State ownership has a negative relationship with dividend yields.
Therefore, we suggest a third hypothesis:

HC0: State ownership does not affect company dividend yield;
HC1: HC0 is not correct.

Data
We collected the required data from the library of Ton Duc Thang University, Vietnam, and followed the suggestions in Fama and French (1992) to exclude financial and public utilities firms, and applied Winsorizing to mitigate extreme value and outliers in the sample as they can lead to biased estimates and inferences. The final data sample includes 4635 observations from 514 companies that are listed on the Vietnam Stock Exchange from 2009-2017.
We used the Panel Data method in the empirial analysis for the following reasons. Hsiao (1985), Klevmarken (1989), and Solon (1989) suggest that the Panel Data model presents data with detailed information and figures with specific complications that are sufficient for empirical research. According to Baltagi (1997), the Panel Data model provides more data than do other statistical models, has higher fluctuations, and reduces the existence of severe collinearity among the explanatory variables.

Variable Definitions
For the dependent variable, we follow Chen et al. (2017) and choose the Dividend Payout Ratio, which is the ratio of the total amount of dividends paid to shareholders relative to the net income of the company (DIVNI). Dividend relative to revenue (DIVSL) is another dependent variable that is used to examine a company's stability.  Table 2.
Consequently, excessive financial leverage has negative impacts on company profits, so that dividends will also be affected.
In Model 3 below, we follow Leary and Michaely (2011) and Harford et al. (2008) to add the ROS variable. ROS reflects business efficiency and company financial strengths. An increase in ROS leads to improvements in business performance in order to affect the payout decision.

Research Methodology
Hwang et al. (2013)   LEV has a maximum value of 0.914, a minimum value of 0.044, and a mean value of 0.502, which means that, on average, one dollar in assets is used to invest 0.502 dollars in debt.
ROS has a maximum value of 0.605, a minimum value of -0.707, and a mean value of 0.222, which means that one dollar in revenue leads to 0.222 dollars in profit.
EPSGROWTH has a maximum value of 19.615, a minimum value of -23.758, and a mean value of 3.785, which indicates that the average growth rate of earnings per share is 378.5%.
CATA has a maximum value of 0.258, a minimum value of 0.001, and a mean value of 0.045, which suggests that for each dollar in asset, the company has 0.045 dollars in cash.
CUR has a maximum value of 16.011, a minimum value of 0.383, and a mean value of 2.203, which shows that a company has 2.203 dollars in short-term asset funds to cover short-term debt.
When this variable is more significant than 1, it shows that company ability to pay short-term debt is secured.

Collinearity
High collinearity arises in the data set when there are close linear dependencies among the independent explanatory variables. In short, it is a potential problem with the data, and not with the model specifiction. In Table 4, we present the Pearson correlation matrix to check the collinearity among the independent variables. Kim (2019) suggests arbitrarily that, when the correlation coefficient between two variables is greater than 0.7, there is a potential collinearity issue. Table 4 shows that there are no correlation coefficients greater than 0.7. Therefore, it seems that the variable selection is appropriate and should not affect the empirical results, namely the estimates and theirstatistical significance, that are presented below.

Consider the Variance Inflation Factor equation (VIF):
VIFj= 1 1− 2 in which 2 is the R 2 value obtained by regressing the Xj on the remaining variables. According to Hair et al. (1998), if the VIF factor is higher than 10, that means the model has severe collinearity.
However, the evaluation criteria of 10 are more suitable in engineering and physics topics that do not use the Likert scale. On the other hand, in social science research, it seems that when VIF is greater than 2, the model is deemed to have high collinearity.
As there is a large number of quantitative explanatory variables in the models, the quantitative variables will use a VIF value of 2, and qualitative variables will use a VIF value of 10. Table 3 shows that the VIF of all the qualitative variables is less than ten, and the VIF of all the quantitative variables are less than 2, which means there is no severe collinearity among the variables in the alternative model specifications.

Hausman Specification Test
In order to choose between the FEM (Fixed Effects) and REM (Random Effects) models, we perform the Hausman specification test to choose between them, namely:

H0: Error and independent variables are not related;
H1: H0 is not correct. Table 5 indicates that the p-value is 0.2597, which is greater than 0.05, so we do not reject the null hypothesis, H0, so that the REM model will be used in all cases in the empirical analysis.

Heteroskedasticity
After performing the Breusch-Pagan test for heteroskedasticity in the residuals of the estimated model, the result in Table 6 show that heteroskedasticity is present when Prob. F(12,3182) is less than α = 5%. Heteroskedasticity is a correctable flaw in the model which can affect the validity of the empirical results.

Main Empirical Findings
After resolving the presence of heteroskedasticity in the model, we have the following results: Table 7 indicates that 79.3% of the range of the dependent variable DIVNI (dividend payout ratio) is accounted for by the independent explanatory variables in the model The FCEO and FCEODUAL variables are critical findings in the empirical outcomes. Table 7 shows that the two variables are inversely related to the dividend payout ratio. If the CEO is female, the dividend payout ratio decreases by 13.3%; if a female CEO takes the duality position, the dividend payout ratio decreases by a smaller amount of 3.6%.
Women tend to take less risk than men, so they are less likely to raise external capital.
Consequently, they are more likely to retain profits for reinvestment and development, so the payout ratio is reduced accordingly. The empirical findings are consistent with the results in Lyness and Thompson (1997), and Faccio et al. (2016) Table 7 also shows the positive relationship between dividend payout and the FCHAIR variable.
Specifically, companies with chairwomen increase the dividend payout ratio by 15.1%. This finding is consistent with those in Arshad et al. (2013), Obradovich and Gill (2013), who suggest that chairwomen tend to take less risk than their male counterparts, retain a higher proportion of profits, and use dividend payments as an administrative method.
FD3 has a negative relationship with dividend payments. When the Board of Directors has at least three female members, the dividend payout ratio is reduced by 20.4%. Román-Martínez et al.
(2012) suggest that female board members tend to use their authority for personal benefits, cooperate more closely with managers, and reduce dividend payments.
STATEOWNERSHIP also has an inverse relationship with the dividend payout ratio. When State ownership in a company increases by 1%, the dividend ratio decreases by 8.3%. Setia-Atmaja et al. (2009) and Ben-Nasr (2015) suggest that firms with higher state ownership may access external debts more easily, so that these firms do not pay higher dividends to attract capital from investors.
Moreover, these companies use retained earnings to over-invest on impractical projects. In this case, our empirical findings are not consistent with those in Chen and Dhiensiri (2009), Yung (2011), andBradford et al. (2013).
EPSGROWTH is inversely proportional to DIVNI. When EPS increases by 1%, the dividend payout ratio decreases by 1.5%. This empirical outcome is not consistent with those in Nizar Al-Malkawi (2007), and Farooq et al. (2009), who suggest that EPS could have a positive effect on company dividend ratios. However, an alternative explanation is that the EPS growth rate and dividend ratio do not have a stable relationship with each other, as the EPS growth rate is affected by multiple factors, such as the change in the number of shares outstanding and the earnings growth rate. When a company decides to issue additional shares, EPS will decrease accordingly.
There is a positive relationship between financial leverage and the payout ratio. When a company increases leverage by an additional 1%, the dividend payout ratio increases by 16.5%. The empirical findings in the paper are consistent with those in DeAngelo et al. (2006). The purpose of using leverage is to enhance profits, which can accelerate earnings if a company uses leverage appropriately.
However, Rozeff (1982), Jensen et al. (1992), and Ye et al. (2019) observed that companies with high leverage tended to pay fewer dividends since these companies need to retain their earnings to pay debts on time. Furthermore, indebted companies will be monitored tightly by creditors, so these firms may not pay a high dividend as do less levered firms.
ROS is inversely proportional to the dividend ratio. According to the empirical result, when ROS increases by 1%, the dividend ratio decreases by 84.9%. This result is consistent with that in Ben-Nasr (2015) because a higher ROS shows that the firm is in a productive business phases, so it is pursuing development plans and milestones rather than using profits to pay dividends. However, this empirical result is not consistent with that in Easterbrook (1984) and Jensen and Meckling (1976), who find that a company with high earnings would pay higher dividends in order to reduce agency costs that arise from free cash flows.
The cash ratio is inversely proportional to the dividend payout ratio. When the cash ratio increases by 1%, the dividend ratio decreases by 158%. This empirical outcome is consistent with agency theory in DeAngelo et al. (2006), Amidu and Abor (2006), and Ye et al. (2019). This outcome suggests that a firm will pay higher dividends to prevent managers from over-investing and wasting resources, so that it will lead to reductions in the cash ratio.
The current ratio has positive effects on dividend ratio. The empirical results indicate that when the current ratio increases by 1%, the dividend payout ratio increases by 1.5%. A high value of the current ratio means that short-term asset increases while short-term debt decreases, the debt ratio decreases, thereby reducing interest payments, while increasing profits and dividend yield. This empirical finding is consistent with those in Ye et al. (2019).

Robustness Checks
In order to test the sustainability of the model, we replace the dependent variable DIVNI with DIVSL, and the results are presented in Table 8.
The empirical results presented in Tables 8 and 9 show that all the variables have a statistical significance of 1% when we test sustainability using the alternative dependent variable DIVSL.

Concluding Remarks
The The empirical results showed that the female CEO and CEO chairwoman decrease the dividend ratio, while chairwomen tend to use the dividend ratio as a tool in managing companies effectively.
Moreover, it was found that BODs with more than three female members tended to decrease the dividend payout ratio.
Following Setia-Atmaja et al. (2009) and Ben-Nasr (2015), we support the theory, HC1: State ownership affects the dividend ratio. The empirical results showed that the dividend ratio decreases when State ownership increases, primarily due to higher state ownership allowing firms to access external borrowing plans easily, so these firms are willing to pay high dividends in order to reduce agency costs.
In conclusion, we suggest that gender diversity in BODs and State ownership influence the dividend payout ratio. The empirical findings can be used to enhance the decisions of investors who want to consider profits through cash dividend payments. The results can be used as a basis for to enhancing corporate governance for listed firms in Vietnam through public regulatory policy, and more widely for countries at similar stages of economic and financial development.    Table 4 29 Pearson Correlations Between Independent Variables Table 5 Hausman Test