Managerial overconfidence and dividend policy in Vietnamese enterprises

Abstract The dividend policy in an enterprise depends not only on the company’s strategy but also on the characteristics of the managers. In particular, the element of overconfidence of the CEO contributes to the decision-making process for the dividend. Therefore, this study is conducted to determine the effect of overconfident CEO on dividend policy through dividend yield and dividend payout. The study carried out an evaluation based on 576 companies listed on Vietnam’s stock market from 2014 to 2018. Panel data model with GLS (error correction model) is used in regression analysis. The analysis results show that CEO overconfidence has a positive impact on Dividend Payout and Dividend Yield. At the same time, the study also shows that the influence of CEO overconfidence on Dividend Payout is the same between state-owned and non-state enterprises and companies listed on different exchanges (HNX and HOSE). In addition, the results also indicate that there is no influence of CEO overconfidence on Dividend Yield in state-owned enterprises as well as companies listed on the HOSE. From the research results, the author also made some conclusions and recommendations for managers and investors.


ABOUT THE AUTHORS
The authors are from universities in Vietnam. The research team implements topics related to finance and economics. Dat Dinh Nguyen is a Doctor at Faculty of Banking and Finance, Foreign Trade University, Hanoi, Vietnam. His areas of specialization and research are financial management, and securities investors Tha Hien To is a Doctor at Le Quy Don Technical University, Hanoi, Vietnam. His areas of specialization and research are economics, business and management, financial management Duy Nguyen is an Independent Researcher at Quantitative Analysis Center, QA Global, Vietnam. His recent research has been published in Journal of Sustainable Tourism, International journal of business and globalisation, Data in Brief, Journal of Risk and Financial Management, etc. Huyen Phuong Do is a Doctor at International School, Vietnam National University, Hanoi, Vietnam. His areas of specialization and research are financial management, and securities investors

PUBLIC INTEREST STATEMENT
This study is one of the first studies focusing on the impact of overconfident CEO on dividend policy in Vietnam.The analysis results show that CEO overconfidence has a positive impact on dividend payout and dividend yield (Overconfident CEO tends to pay dividend policy more than CEO non-OverconfidenceThere is no influence of CEO overconfidence on dividend policy in state-owned enterprises.There is an influence of CEO overconfidence on dividend policy in others-owned enterprises.

Introduction
Dividend policy is the payment of financial benefits in a joint stock company to its shareholders (Masum, 2014;. This is the profit that shareholders receive through their investments and the company. In other words, the shareholders always deserve the benefits from their contribution to the joint stock company . However, whether or not the dividend is paid or paid is affected by the decisions of the company management when dividends are taken from the company's income (Allen & Saunders, 2002;Deshmukh et al., 2013;Masum, 2014). Shortterm or long-term investment strategies of enterprises are also reflected in the dividend policy of shareholders in the company (in the case of long-term investment, enterprises often limit dividend policy and use the retained earnings to continue investing) (Deshmukh et al., 2013).
The policy of paying dividends depends not only on the company's policies and development strategy (Deshmukh et al., 2013) but also the CEO's confidence (Malmendier & Tate, 2005b). CEOs are overconfident when their behaviors consistently perform above their actual capacity in a certain area (Gervais et al., 2003). Overconfident CEO often thinks that his competence is above average for other CEOs (Malmendier & Tate, 2005b;Wrońska-Bukalska, 2018). When the CEO's decisions are overconfident to bring good results, it makes them feel more confident, but when the CEO's decision is overconfident to bring bad results, they often consider this to be unlucky (Nguyen, Dang et al., 2020a). At the same time, the consequences brought when overconfident CEOs are more severe than non-confident CEOs (Wrońska-Bukalska, 2018).
There have been many studies showing the relationship between overconfident CEO and dividend payment policy. Some studies show that CEOs who are overconfident about their enterprise's ability to grow in the near future will increase dividend payment when the CEO expects that the cash flow will be higher in the short term (Wu & Liu, 2011). Conversely, CEOs who are not overconfident about the short-term future will tend to pay less dividends when they want to use more retained earnings to invest in enterprises for long-term strategies (Ben-David et al., 2007). Overconfident CEOs will take more advantage of retained earnings for their subsequent investments when they have high expectations of the ability to succeed in their decisions leading to tighten the dividend payment policy (Ben-David et al., 2007;Deshmukh et al., 2013) Researches on CEOs overconfidence in Vietnamese enterprises in recent years are limited. Although there has been researched by Nguyen (2020) on the impact of Overconfidence CEO on dividend policy, this study was only done with a group of companies listed on the HOSE (one of the many on the stock exchange of Vietnam) and only considered the dividend payout index without the dividend yield. Therefore, the analysis of the impact of the CEO overconfidence on the dividend payment policy will find out how this relationship exists in the environment of Vietnamese enterprises. The study will focus on analysis for enterprises operating on the Vietnam Stock Exchange from 2011 to 2018. The research results will help managers to strategically use CEOs suitable for their companies. At the same time, the results will recommend to investors who have decided to choose in accordance with their investment strategy.

Dividence policy
The dividend payment policy to shareholders is decided by the corporate manager based on the company's strategy. Dividend payment is based on business results or cash retained by the company (Deshmukh et al., 2013;Okafor et al., 2011). Dividend payment comes from net profit earned on each stock in cash or other assets . Measurement of dividend payment (dividend yield and payout ratio): The dividend yield and payout ratio are two valuation ratios which investors and analysts use to evaluate companies as investments for dividend income.
Dividend yield is the expected annual dividend of the stock divided by its current price. The dividend yield is the percentage return the investor expects from the dividend paid by the stock. The dividend yield provides a good basic measurement for an investor to use by comparing the dividend income from their current holdings to potential dividend income available through investing in other equities or mutual funds . In regard to overall investment returns, it is important to note that increases in share price reduce the dividend yield ratio even though the overall investment return from owning the stock may have improved substantially. Conversely, a drop in share price shows a higher dividend yield, but may indicate the company is experiencing problems and lead to a lower total investment return.
The dividend payout ratio is considered more useful for evaluating a company's financial condition and the prospects for maintaining or improving its dividend payouts in the future. The dividend payout ratio reveals the percentage of net income, a company is paying out in the form of dividend . If the dividend payout ratio is excessively high, it may indicate less likelihood a company will be able to sustain such dividend payouts in the future, due to the fact that the company is using a smaller percentage of earnings to reinvest in company growth. Therefore, the rate of stable dividend payments often takes precedence over a large proportion of irregularities. A good way to determine whether the incidence of corporate expenses is reasonable or not.

Theory of dividend policy
Theories of dividend policy, such as bird-in-hand theory, signaling theory, agency cost theory, Clientele effect: In theory bird-in-the-hand, an investor is interested in dividend cash flow to cash flow from capital gains. The main reason of the dividend is the cash flow that investors will surely get instantly, while reinvested profits from the success of the company in the future is uncertain (Diamond, 1967;Dong et al., 2005) Signaling theory or hypothesis information content said that the dividend policy carries the information, which is a signal to investors. Due to the existence of information asymmetry, investors always have little information about the company over its management; they think dividends carry information that managers want to convey to the market (Asquith & Mullins, 1983;Bhattacharya, 1979).
Agency cost occurs when there are conflicts of interest between managers and shareholders in the company; one side is the owner of the firm's assets, the other is the user of the firm's assets (D'Souza & Saxena, 1999;Jensen et al., 1992).
Effects customer group assumes that companies with customer groups and different customer groups have different interests, so the change of dividend policy may cause the satisfaction of customer groups large, negative impact to change the interest of customers (Dhaliwal et al., 1999;Lewellen et al., 1978). Effects customers: Differences in tax incentives among the investor group to produce the effect customers, including dividend policy of a company that is optimized for tax incentives of client investors. The individuals in the highest tax bracket have priority for the shares not pay dividends or dividend lower, while investors and companies exempt preference for stocks with high dividends (Booth & Johnston, 1984;Zhan Shu, 2000). According to this argument, investors may be attracted by the stocks that suit consumer preferences/their savings. That is, if the dividend income is taxed at a higher rate of capital gains, investors (or customers) in a high tax bracket may prefer dividend stocks do not pay dividends or lower, and vice versa. In addition, the presence of transaction costs can create a certain number of customers

Overconfidence CEO
Overconfidence is considered to be the CEO's overestimation of the cash flow that can be created in the future and underestimating the risks that may be encountered in the process of creating cash flow. These expectations are considered by the CEO overconfidence to be reasonable, so they often do not care about risks (Bharati et al., 2016). In some special cases, when under pressure from risk, people become more confident in their work, even though they perceive the risks that may occur to themselves (Chi et al., 2020;Huynh, 2020;Nguyen, Pham et al., 2020). At the same time, overconfident CEOs who think about themselves are likely to be above average for others, leading to their decisions often resulting in significant deviations leading to unexpectedly high efficiency waiting or serious damage (Malmendier & Tate, 2005b;Nguyen, Dang et al., 2020a).
Regarding the method of measuring Overconfidence, Malmendier & Tate (2005a) conducted an approach to measuring Overconfidence in two ways: (1) When the CEO continues to hold the options later than the original contract proposal will show. The CEO is overconfident in the ability to raise stock prices, and the CEO will profit from holding options; (2) Continue to consider the end of the option period. If a CEO is optimistic enough about the company's future performance for which he holds options until expiry, the CEO is classified as overconfident. Finally, because underdiversified CEOs avoid buying more equity, the authors categorize CEOs who regularly increase the company's shareholding ratio as overconfident (Malmendier & Tate, 2005a) 3. Method

Research model
From the previous studies, the author's research model is presented as follows: The research variables are detailed in Table 1 The dependent variable in the model is measured through two indicators, Dividend Yield and Dividend Payout. Also, the independent variable is Overconfidence-Dummy variable (OV) if the CEO is overconfident, then OV = 1 and OV = 0 if CEO is not overconfident. In addition, the study also conducted the use of control variables such as: Firm size, Revenue growth; Leverage; age of firm.

Data and research method
The research data is used based on the reports of enterprises listed on HOSE and HNX on Vietnam stock market from 2014 to 2018. Data after being collected in financial statements will be coded and included in STATA software for analysis. With the characteristics of table data, the author conducted using basic models such as fixed effect model, random effect model and GLS model to consider the sustainability of the model.

Descriptive
The research variables are collected and encoded into STATA software for analysis. Although the study collected over 277 companies, however, when analyzing there was 1 enterprise lacking complete data, 5 observations were lost. Also, some businesses were missing, with a total missing for 4 observations. Therefore, the total number of actually analyzed enterprises was only 276 enterprises. The description of the variables indicates the mean of Dividend Payout is 0.397, of which the largest is 1,093 and the smallest is 0. Mean of Dividend Yield is 0.063, the largest is 0.233 and the smallest is 0. The number of years becomes set up a mean of 3 years, of which the largest number of years is 12 years and the lowest is 5 years. The mean of leverage ratio is 0.738 of which the largest is 2.604 and the smallest is 0. The mean of revenue growth is 0.344 corresponding to 34.4%, of which the largest is 102%/year and the smallest is −45%/year (see in Table 2) With qualitative variables, the author will include the frequency analysis. The results show that the number of overconfident CEOs in the latest year 2018 was 71 people accounting for 12.31%, CEOs not overconfident accounted for the majority with 506 people accounting for 87.69%. Stateowned enterprises only account for 21.14% by 122 enterprises; The remaining 78.86% are nonstate owned enterprises. Information on the number of enterprises listed on HNX is 296 enterprises, accounting for 51.3%; The number of enterprises on HOSE is 281, accounting for 48.7% (see in Table 3.)

Correlation matrix
The correlation coefficient matrix shows that for control variables AGE, LEV and GROWTH are not correlated with Dividend Yield but are correlated with Dividend Payout. The SIZE has a negative correlation with both dependent variables, Dividend Yield and Dividend Payout. The control variables and the independent variables are all correlated at low levels (the largest correlation coefficient is 0.2, showing that the multi-collinear is almost non-existent in the research model). In the correlation matrix table, the author did not conduct correlation analysis for CEO overconfidence because this is a dummy variable so the correlation coefficient will not make sense in the analysis (see in Table 4). Regression regression with table data in the next steps.

Regression
With the FEM, REM and GLS models used in this analysis. In particular, Hausman tests show that FEM model is more suitable than REM model with research data. However, the FEM model has problems with autocorrelation and heteroskedasticity. Therefore, the author uses the GLS model for final data analysis. The analytical results obtained are as follows: Regression analysis of the impact of Overconfidence on DYield shows that the variable Overconfidence has a positive impact on DYield (beta coefficient is 0.0086 and p-value is less than 0.05) (see in Table 5). This result indicates that overconfident CEOs will have a higher tendency to pay dividends per share than firms with overly confident CEOs. For listed companies in Vietnam in general, the more confident CEOs are, the more likely they are to pay shareholders a share of equity. It can be seen that the CEOs who are confident in the cash flow generated in the short term lead them to be willing to increase the Dividend Yield to encourage their shareholders and the problems of cash flow and solvency are overconfident for the CEO. This is a controllable factor. The results of this study are consistent with the previous study of Wu and Liu (2011), showing that the positive effect of Overconfidence on dividend payment.
For the control variables in the model, the regression results also show that firm size has a negative impact on Dividend Yield (beta coefficient is equal to −0.002 and p-value is less than 0.05). This result shows that with the expansion of the firm size, the capital source also increases, in addition to using loans, enterprises also use the capital from retained earnings to reduce interest expenses. This has resulted in the Dividend Yield ratio also falling to increase the available capital in its business. The analysis also showed that the variables AGE, LEV and GROWTH do not affect Dividend Yield (p-value is greater than 0.05). This result shows that the age of the enterprises, the ratio of leverage or the growth of revenue is not the reason for the change in the rate of Dividend Yield.
Results of regression analysis for the dependent variable Dividend Payout show that CEO overconfidence has a positive impact on Dividend Payout (beta coefficient is equal to 0.06 and p-value is less than 0.05) (see in Table 6). It can be seen that overconfident CEOs always have a higher dividend payout ratio than CEOs are not overconfident. This result once again shows that the CEO who is overconfident about the success of his/her executive decision in the short term will always tend to pay more dividends. It can be seen that representation theory has influenced the acts of paying dividends in enterprises. The overconfident CEO always makes decisions to pay more dividends to prove his/her management ability in the short term. Long-term investments that require large capital have not been considered by overconfident CEOs. It can be seen that overconfident CEOs are making under-investment decisions when they do not use diversification of investments but focus on risky projects to bring profits to shareholders.     Regarding the effect of controlling variables on dividend payout, it is shown that the number of years of establishment of a business has a positive effect on dividend payout. This result shows that the longer enterprises are established, the bigger the dividend payout. The longer the business operates, the more prestige and long-term business strategy tends to be more concerned. Therefore, these enterprises tend to pay dividends more to bring cohesion with shareholders as well as pay for shareholders' belief in the company. Besides, Leverage, Revenue growth and firm  size all have a negative impact on Dividend Payout. This result shows that the increasing debt ratio will make Dividend Payout ratio decrease when all businesses want to reduce the burden of interest expenses. This case Pecking Order Theory shows more consistent information asymmetric theory . In terms of revenue growth and firm size, having a negative impact on Dividend Payout shows that increased revenue tends to increase with scale. This shows that during the research period, all enterprises tend to expand, and this makes the revenue tend to increase. However, the increase in revenue during this period made Dividend Payout decrease because the large-scale investments were still compensated by the retained earnings.
Besides the general assessments for enterprises, the study also conducted data analysis in the form of state ownership and listed exchanges. The analysis results show that the State-owned enterprises (SOEs) and enterprises listed on the HNX does not have the impact of overconfident CEO on Dividend Yield (p-value is greater than 0.05). This result shows that SOEs with overconfident CEO and non-confident CEO have the same dividend yield. In the form of state ownership, besides the task of effective investment, there are other social responsibilities associated with the role of state enterprises. Therefore, CEOs who are overconfident and not overconfident does not make any difference when making Dividend Yield policy. At the same time, the analysis also indicates that non-state owned enterprises and listed companies on HOSE have the same impact of overconfident CEO on Dividend Yield. It can be seen that non-state enterprises and listing on HOSE tend to be similar to the situation of listed companies (see in Table 7).
The analytical results with the impact of the CEO being overconfident on Dividend Payout have similarities between the types of state-owned and non-state enterprises as well as the listing floor (CEOs are overconfident to work in the same direction on Dividend Payout) (see in Table 8)

Conclusion
Through quantitative analysis techniques, the research results are quite similar when they all show the direction of CEO overconfidence on enterprises' dividend payment policy. In other words, in the environment of enterprises listed on HNX and HOSE, overconfident CEOs tend to pay dividends more than CEOs who are not overconfident. At the same time, this study's results also show that enterprises with overconfident CEO often do not care about risks in the short term. Therefore, the study also recommends that enterprises with long-term strategies need to control overconfident CEO's decisions. Because the rate of retained earnings often decreases because the dividend payment policy is always high in a company with an overconfident CEO (Paying large dividends is beneficial to shareholders in the short term. However, retained earnings need to be retained in the long term to be distributed for investment rather than shareholders). Investors interested in short-term investments can also consider prioritizing businesses with overconfident CEOs to benefit in the short term (The CEO is overconfident to pay more dividends. This brings benefits to the shareholders).