Corporate Sustainable Management, Dividend Policy and Chaebol

: This study empirically examines the relationship between corporate sustainable management (CSM) and dividend policy. Among the various motivations related to dividends, this study examines the relationship between CSM and dividend policy based on the agency and signaling theory. After examining the relationship between CSM and dividend policy, we investigate whether belonging to a large business group (chaebol group) has a signiﬁcant effect on the relationship between CSM and dividend policy. The analysis period is from 2011 to 2018, and the ESG ratings of the Korea Corporate Governance Service are used as proxies for CSM. The empirical results show that CSM and dividends have a signiﬁcant relationship in the positive direction. This means that ﬁrms with excellent CSM activities have higher dividend levels than those that do not. Furthermore, the association between CSM and dividends is more negative for ﬁrms belonging to a chaebol group. This indicates that the positive relationship between CSM and dividends in a ﬁrm that belongs to a chaebol group is weakened. This means that the relationship between CSM and dividends in the group belonging to the chaebol group is weakened. It belongs to the group of conglomerates, meaning that the relationship between the amount of dividends and CSM weakened. Our study focuses on CSM as a determinant of dividends, and examines the effects of belonging to a chaebol group in the relationship between CSM and dividends. Given that resolving the interest incompatibility between investors and managers is the focus of corporate governance, dividend policies can be used as a method for resolving the interest incompatibility between investors and managers.


Introduction
A dividend policy is a financial decision to divide the business performance of a firm into dividends, distributed to shareholders and internal reserves for future reinvestment. The question of how the dividend policy affects corporate value has been of interest for many scholars [1]. Miller and Modigliani [2] put forth the dividend irrelevance theory, which states that dividend policies are irrelevant to corporate value under the assumption of a perfect capital market, where rational investors exist. After Miller and Modigliani's [2] dividend irrelevance theory was published, various studies in which the assumption of a perfect capital market was mitigated have been conducted.
Agency theory approaches dividend policies from the viewpoint of the agency problem. Jensen and Meckling [3] argued that since the lower the internal equity ratio, the higher the agency cost, the monitoring function of the external capital market can be strengthened by increasing dividends. Fama and Jensen [4] and Jensen [5] also claimed that dividend policy could be used as a means to reduce agency costs for managers. Signaling theory indicated that under information asymmetry, managers can use dividends to deliver positive information on corporate prospects to the capital market [6]. Kim et al. [7] and Grullon et al. [8] also contended that dividends were used as a means of signaling effect.
In addition, there are cases where firms belonging to a chaebol group have been criticized due to the fact that the controlling shareholder exercises 'voting rights' exceeding his/her 'ownership' to expand his/her wealth, ignoring the interests of minority shareholders. This problem becomes even more serious in the absence of mechanisms to control the discretion of the controlling shareholder [25,27,28].
Using 4718 Korean firm-year observations over the 2011-2018 period, our regression results show that the relationship between CSM and dividends was a significant positive relationship. This means that firms with more active CSM activities pay more dividends. With regard to the determinants of dividends, CSM activities can be interpreted from the viewpoint of the agency problem and signaling theory. Furthermore, the association between CSM and dividends is more negative for firms belonging to a chaebol group, which, in turn, indicates that the positive relationship between CSM and dividends in a firm belonging to a chaebol group is weakened.
When compared to previous studies related to dividend policy, this study makes the following contributions. First, this study provides evidence that CSM activities can have effects as a determinant of dividend policy. Second, this study is meaningful in that it directly examines the role of corporate governance in the relationship between CSM and dividend policy. This study suggests that corporate governance through whether or not a firm belonging to a chaebol group works as an alternative to reducing the agency problem in the relationship between CSM and dividends.
This study proceeds as follows. Section 2 reviews the theoretical background and develops hypotheses, and Section 3 presents the research design. In addition, Section 4 reports the empirical results and Section 5 concludes the study.

CSM and Dividends
Agency problems occur due to the separation of ownership and management. Therefore, when shareholders' rights are limited or relatively weak, managers can exacerbate agency costs by making decisions that favor their private interests rather than those of shareholders. The more distributed the ownership of a firm is, the more seriously the agency problem becomes between shareholders and managers leading to increases in agency costs. However, agency problems can occur not only between shareholders and managers, but also between majority shareholders and minority shareholders. Generally speaking, agency theory concerns the existence of an agency problem that arises as a result of information asymmetry between insiders and outsiders of the firm. In this context, the insiders are management and outsiders are shareholders, both being important stakeholders in the business [29].
Agency theory argues that dividends can be used as a means to alleviate the agency problem between shareholders and managers or between controlling shareholders and minority shareholders. Dividends can reduce the agency problem of equity capital by shifting control of firms' business activities and performance to the capital market [30]. The more dividend payments are made, the greater the need for firms to raise capital through increased stock issuance. When capital raising from the capital market increases, monitoring by institutional investors, the stock exchange, and capital providers is strengthened. Taking this into consideration, Fluck [31] and Myers [32] proposed the agency theory model of dividend behavior that suggests managers pay dividends to avoid the monitoring activities of shareholders.
Signaling theory argues that dividends can be used as a means of delivering superior information about the firm's future value to the capital market under asymmetric information. This theory indicates that an unexpected increase in dividends could be perceived as an optimistic signal of the firm's future prospects, leading to a rise in stock price, and an unexpected decrease in dividends could be interpreted as a pessimistic signal, leading to a drop in the stock prices. Therefore, the stock prices changes due to the signaling effects regarding the predicted future values rather than the dividend payment itself.
The effect of CSM on dividend policy can be explained with two routes. First, the cash flow of high CSM firms can be expected to increase in the future, and a firm's ability to generate earnings typically affects its dividend policy. Many previous studies have already reported that earnings were a major determinant of dividend policies [33]. In addition, CSM activities can create new opportunities in the market [34] and may have positive effects on the prices of goods and services, including social costs because CSM activities form corporate moral capitals. Moral capital is a product that appears in the process of judgment, evaluation, and determination of moral value by stakeholders and the social community. The moral capital created by CSM activities can have a sort of protective function similar to the nature of insurance [15]. Friendly relationships with stakeholders not only reduce transaction costs [18] and strengthen competitive advantages [19], but also mitigate the impact of unfavorable events on cash flows [35]. Eventually, moral capital is generated to the extent that it has potential for profit creation. Heal [36] reported that CSM activities increase earnings through friendly relations with stakeholders, efficient asset allocation, labor policies, and improved management activities. Moreover, the positive relationship between CSM and earnings has been identified in recent studies through meta-analysis [37]. Intangible benefits thanks to CSM activities can not only improve earnings, but also creates scenarios whereby investor and stakeholder utility is enhanced by being connected to dividend policies [38]. Benlemlih [20] also shows that since firms are using dividend policies to manage the agency problems caused by excessive investments in CSM activities, firms with more CSM activities are paying more dividends.
Second, high CSM firms can reduce capital costs. Firms with high CSM activities can reduce systemic risks [39], the costs of debt [40], or equity capital [41]. Firms that perform CSM activities can not only have excellent corporate governance [42], strengthening the investor and customer loyalty [37], while accumulating a positive reputation and moral capital, [43] but also alleviate the problem of information asymmetry among stakeholders [11]. In cases where the cost of capital decreases, a firm's incentive to hold cash increases because the opportunity cost of holding cash decreases. In addition, when the cost of capital decreases, investment incentives increase because restrictions on external financing are relieved. Accordingly, firms experiencing relief of financial constraints can select investment plans that are somewhat less profitable and reduce dividend payments.
Meanwhile, Kim (2009) [44] argued that CSR activities have a positive effect on corporate value. While reinforcing social responsibility is also a cost factor in the short term, it has a positive aspect of raising a company's reputation, thereby increasing sales and lowering costs. In particular, it is reported that firms with good cash flow, low debtto-equity ratios, and firms belonging to a chaebol group have a relatively greater effect of increasing their corporate value than those that do not. Kook and Kang (2011) [45] showed that strengthening social responsibility has a positive effect on corporate value, and that there is an incentive for firms to continuously carry out voluntary social responsibility activities. In other words, through reinforcement of social responsibility, the interests of shareholders and stakeholders can be promoted at the same time.
When the results of the previous studies above are putting together, it can be seen that the relationship between CSM activities and dividend policy can have both of two directionalities. From the viewpoint that capital costs related to capital raising are reduced as a result of actively carrying out CSM activities so that the opportunity cost of holding cash is lowered, and opportunities for external financing can be facilitated, dividends will decrease. On the other hand, the future cash flow of firms with high CSM is expected to increase and dividends will increase thanks to the augmentation in profit-generating capacity [46]. In addition, from the viewpoint of corporate governance, there is a possibility of increasing dividends as part of an effort to reduce agency costs. Therefore, in this study, a hypothesis is established as follows regarding the relationship between CSM activities and dividends.
Hypothesis 1 (H1). CSM activities and dividends will not be related to each other.

Chaebol, CSM, and Dividends
A firm with a sound governance structure can be said to have a high quality of final decision-making because shareholders' rights protection, board of directors, disclosure, audit organization, and management errors are all properly institutionalized. Therefore, it is inferred that a firm with a sound governance structure can actively carry out CSR activities, which are an important management issue in modern society [47].
In a study on the relationship between corporate governance and dividends, La Porta et al. [24] presented the managerial opportunism hypothesis, which advances the idea that the better the governance, the more dividends will be increased to reduce the agency problem of managers. The substitute hypothesis is that since relatively fewer agency problems will exist with better governance, dividends will not be paid excessively to additionally reduce agency costs. They reported that since governance and dividends appear to have a positive relationship, the managerial opportunism hypothesis was supported. Meanwhile, Jiraporn and Ning [48] argued that the substitute hypothesis was supported because the degree of protection for shareholders' rights and dividends appears to have a negative relationship.
The general focus of discussions related to corporate governance is to resolve disagreements between managers and investors.
Dividend policy can be utilized to achieve this goal. According to the agency theory, dividend payments can be used as a means to alleviate the agency problem between shareholders and managers. Jensen [5] argued that firms should raise funds from investors for high-profit investment proposals, but in the case, where no appropriate investment proposal is found, paying free cash flows as dividends to shareholders would be the most effective method to alleviate conflicts of interest between managers and investors. If the corporate governance is sound, the company is expected to implement appropriate dividend policy. That is, even when seen from the viewpoint of agency theory, firms with sound corporate governance are expected to choose a dividend policy that seeks to increase corporate value.
Whether the chaebol group's role is positive or negative may differ depending on whether the aspect of conflicts between managers and shareholders is seen or the aspect of conflicts between controlling shareholder and minority shareholders is seen. First, from the viewpoint of the agency problem between shareholders and managers, Kim et al. [49] argued that the frequency of manager replacement is relatively high in chaebol group and such replacement mainly shows a form of movement within the corporate group, and claimed that chaebol groups work as a controlling device for affiliated company managers. In addition, after the financial crisis in 1998, the management behavior of Korean chaebol groups changed [50], and chaebol groups performed desirable management activities such as not raising large-scale funds using excessive investments or debts [51] and have shown high profitability [50]. In addition, even when chaebol groups carried out overinvestments, such investments were found to have generated high profitability in the long term [52]. On the other hand, in the case of firms belonging to a chaebol group, the controlling shareholders have been criticized in that they expanded their wealth and sacrificed the interests of minority shareholders by exercising their "voting rights" exceeding their "ownership". In particular, this negative problem can be more serious in cases where there is no mechanism to control the discretionary power of the controlling shareholders [25,27,28]. That is, managers are more likely to play a role for chaebol groups that can influence manager replacement rather than taking actions for shareholders and may carry out internal transactions and adjust transfer prices between affiliated firms in order to achieve the optimal tax burden of the entire chaebol group.
To sum up the results of the above prior study, it can be seen that corporate governance works as an alternative to solve the agency problem between managers and capital market participants and alleviate information asymmetry problem [3,4]. The corporate governance termed chaebol can be interpreted differently according to the two perspectives. According to signaling theory, excellent corporate governance is expected to increase future cash flows, thereby increasing dividends. In addition agency theory suggests that where there is no adequate investment opportunity, dividends can be increased as part of an effort to limit opportunistic actions of managers to pursue private gains. Firms will try to show that agency costs are minimized by providing information on the excellence of governance through dividend policy. In chaebol groups, there may be a large agency problem due to the conflict of interest between controlling and minority shareholders. In other words, in firms belonging to a chaebol group, the incentive for controlling shareholders to exploit the wealth of minority shareholders can be large. In this case, the dividend police can be used by firms belonging to a chaebol group as a means to alleviate the agency problem between controlling and minority shareholders [53,54].
In addition, in a family firm with a high concentration of ownership, dividends can be increased compared to a distributed firm because the level of agreement between management and shareholders is high [33]. If positive factors of dividends such as decreases in agency costs due to dividends and information effects of dividends are large, dividend payments will increase the corporate value. On the other hand, from the viewpoint of the substitute hypothesis that when governance is better, dividends will not be paid excessively to further reduce agency costs because there will be relatively fewer the agency problems [24,48]. The chaebol and dividend payments can act as a selective control mechanism to alleviate the agency problem that arises within the company. If there is a controlling shareholder who can substantially control a firm, that controlling shareholder can serve as an effective surveillance mechanism to control the waste of internal resources by managers on unprofitable businesses. Therefore, the chaebol group does not need to increase dividend payments as a means to alleviate the agency problem related to free cash flow proposed by Jensen [5]. Compared to the non-chaebol group, the chaebol group has enhanced post-regulation and market surveillance functions. Additionally, because the chaebol group is large and has a high market share, external stakeholders require more detailed and accurate information.
Furthermore, since many financial analysts, institutional investors and foreign investors monitor the chaebol group, it is possible to actively monitor business activities. Accordingly, firms belonging to a chaebol group will have a lower level of information asymmetry compared to the non-chaebol group [55]. In the case of belonging to a chaebol group, the top management increases the inefficiency and opacity in overall management by exercising more voting rights than ownership [56,57].
Domestic and foreign chaebol groups have a variety of governance structures and complex investment structures, such as circular investment.
A domestic conglomerate group has properties such as a complex structure, such as a variety of investment governance and circular investment. Due to these characteristics, it is difficult for external investors to collect investment structure data of affiliates, access decision-making information at the corporate group level, or analyze direct or indirect impact on individual firms.
In this regard, information asymmetry may occur more than that of non-chaebol groups [54]. Therefore, the second hypothesis is as follows: Hypothesis 2 (H2). The relevance of CSM activities and dividends will differ depending on whether or not they belonging to a chaebol group.

Empirical Models
The regression model to verify the effect of CSM on dividends is as shown by Equation (1). CSM is measured by the evaluation score of the Korea Corporate Governance Service (KCGS). Since 2003, KCGS has been conducting corporate governance evaluation based on high transparency and expertise. Since 2011, KCGS has been evaluating the sustainability management level of Korean listed firms every year through ESG evaluation that includes social responsibility and environmental management. The ESG evaluation of the Korea Corporate Governance Service aims to assist listed companies to check the current level of sustainability management and use it for improvement. The ESG evaluation model of the KCGS is an independent evaluation model developed in accordance with international standards such as OECD Corporate Governance Principles and ISO26000, as well as faithfully reflecting the Korean legal system and business environment. The ESG ratings provided by the KCGS is a measure according to international standards and is used as a proxy for CSM in various prior studies and is proving its reliability. CSM's main measurement indicators are centered on non-financial indicators. The detailed indicators are organized around stakeholders, but there are a number of indicators centered on shareholders. Environmental management indicators consist of environmental strategy, environmental organization, environmental management, environmental performance, and stakeholder response. Social management indicators consist of relationships with workers, relationships with suppliers and competitors, and contributions to consumers and local communities. Corporate governance consists of the protection of shareholder rights, the composition and operation of the board of directors, disclosure, the audit organization, and the distribution of management errors.
The regression model to verify the relationship between CSM and dividends is as shown by Equation (1).
The regression model to verify the effect of whether or not belonging to a chaebol group on the relationship between CSM and dividends is as shown by Equation (2).
The CSM in Equation (1) represents the CSM level. The larger the value, the more excellent the CSM activity. CSM is the interest variable in Hypothesis 1, and the predicted sign of β 1 is in the positive direction. The higher the CSM score is, the higher the dividend level is expected to be. In Equation (2), CHAEBOL indicates whether or not belonging to a chaebol group. CSM × CHAEBOL is an interaction variable between CSM and CHAEBOL. CSM × CHAEBOL is the interest variable in Hypothesis 2, and the sign of β 3 is in the positive or negative direction. As control variables, SIZE, LEV, ROA, AGE, OWN, FOR, B_SIZE, B_RATIO, CASH, and FCF that affected the level of dividends in previous studies were selected. SIZE is the firm size and is measured by taking the natural logarithm of total assets. LEV is the debt ratio, representing the leverage or capital structures. ROA represents return on asset and AGE represents firm age. OWN and FOR are included to control corporate governance. B_SIZE means the size of the board of directors, and B_RATIO means the ratio of outside directors among registered directors. CASH stands for cash and liquid financial assets and FCF stands for free cash flow.

Samples and Data
The samples are for listed firms on the Korea Exchange from 2011 to 2018. Financial data and stock price are extracted from FN Data-Guide and TS-2000. For ensuring the homogeneity, the financial industry is excluded from the samples. Since firms that do not settle accounts at the end of December are concentrated on certain industries, and different settlement days may have different effects, only those corporations that settle accounts at the end of December were included in the samples. In our study, outliers of all variables except dummy variables were observed and adjusted to observations with outliers less than 1% at the bottom and less than 99% at the top. The final samples used in our study were 4718 firm-years. Table 1 shows the descriptive statistics of major variables. The dividend level (DIV_ASSET) was 1% on average, and the median was 0.6%. Given that the median was smaller than the average, the majority of firms had a dividend level lower than 1% of total assets. The average of the total evaluation grade (TOTAL_SCORE) is 7.237, and the average of the corporate governance evaluation grade (GOV_SCORE) is 6.609. The average of the social responsibility activity evaluation grade (SOC_SCORE) is 7.328, and the average of the environmental management evaluation grade (ENV_SCORE) is 7.346. The average of whether or not the firm belonged to a chaebol group (CHAEBOL) was 0.258, indicating that about 26% of the samples belonged to a large business group. Firm size is 27.128 on average, and the median is 26.904. The average debt-to-equity ratio (LEV) is 0.471, and the median is 0.475. The average return on asset (ROA) is 2.2%, and the average firm age (AGE) is 3.496. The majority shareholders' share ratio is 43.8% on average, and the average foreign ownership 10.3%. The size of the board of directors (B_SIZE) is 1.173 on average, and the average ratio of outside directors (B_RATIO) is 62%. The average ratio of cash and liquid financial assets is 14%, and the average free cash flow (FCF) is 1.2%.  Table 2 shows the results of Pearson correlation analysis of major variables. Dividends (DIV) show a significant positive relationship with CSM activities (TOTAL_SCORE, GOV_SCORE, SOC_SCORE, ENV_SCORE), return on asset (ROA), majority shareholders' share ratio (OWN), cash and current investment assets (CASH), and free cash flow (FCF). This means that the more excellent the CSM activity, the higher the return on asset, the higher the majority shareholder's share ratio, the higher the proportion of cash and current investment assets, and the larger the free cash flow, the higher the dividend level. Although this result was obtained without controlling other characteristics affecting the dividend payout ratio, it can be seen that the result is consistent with the hypothesis of this study that the more active the CSM activities, the higher the dividend level. On the other hand, chaebol group (CHAEBOL), firm size (SIZE), debt-to-equity ratio (LEV), firm age (AGE), and outside director ratio (B_RATIO) show significant negative relationships with dividends (DIV). This means that the firms belonging to a chaebol group, the larger the firm size, the higher the debt ratio, the higher the firm age, and the higher the outside director ratio, the lower the dividend level.  Table 3 is the results of regression analysis of Equation (1). The F value is shown to be significant at a level of 1%, indicating that the empirical model is suitable. As shown in Table 3, the regression coefficients (β 1 ) of CSM (TOTAL_SCORE) on dividends was shown to be 0.0006, which was a significant positive value at the level of 5%. Although not presented in the table, the regression coefficients (β 1 ) of CSM (GOV_SCORE, SOC_SCORE, ENV_SCORE) were shown to be 0.0001, 0.0002, and 0.0006, respectively, which were significant positive values at the levels of 10%, 10%, and 1%, respectively. That is, this is an empirical result showing that the more excellent the CSM activities of the firm, the higher the level of dividends of the firm. This result indicates that the more active a firm's CSM activities are, the more dividends the firm pays, thereby supporting hypothesis 1. That is, this result can be interpreted from the viewpoints of the agency theory and the signaling theory in relation to various motives for paying dividends [3,4,6,[26][27][28]. In addition, this supports the view that CSM activities can improve a relationship with stakeholders and increase earnings through efficient management. In addition, this supports the view that CSM activities can improve relations with stakeholders and increase profits through efficient management. Thus, an earnings increase may lead to a dividends increase.

CSM and Dividend Policy (H1)
On reviewing the control variables, it could be seen that the level of dividends showed significant relationships in the positive direction with return on asset (ROA), foreign ownership (FOR), cash and liquid financial assets (CASH) holding ratios, and lagged dividends (DIV t−1 ). This means that the higher the return on asset, the higher the foreign ownership, the higher the lagged dividend, and the higher the cash and liquid financial assets holding ratio, the higher the dividend level. The level of dividends showed significant relationships in the negative direction with the firm size (SIZE) and the growth rate (GRW). This means that the larger the firm size and the higher the growth rate, the lower the dividend level. Table 4 shows the result of regression analysis of Equation (2). The regression coefficients (β 1 ) of CSM (TOTAL_SCORE) on dividends was shown to be 0.002, which was significantly positive at the levels of 1%. The regression coefficients (β 3 ) of CSM × CHAE-BOL (TOTAL_SCORE × CHAEBOL), which indicate the relationships between CSM and dividends according to whether or not they belong to a chaebol group were shown to be −0.002, which was significantly negative at the level of 1%. The regression coefficient (β 1 ) of CSM (GOV_SCORE × CHAEBOL) was shown to be -0.0001, with a negative direction, but it was not significant. Although not presented in the table, the regression coefficients (β 3 ) of CSM × CHAEBOL (SOC_SCORE × CHAEBOL, ENV_SCORE × CHAEBOL), which indicate the relationships between CSM and dividends according to whether or not belonging to a chaebol group were shown to be −0.0004, and −0.0006, respectively, which were significantly negative at the levels of 5%, and 5%, respectively. The study results as such indicate that in firms belonging to a chaebol group, the positive relationship between CSM and dividends is reduced compared to firms that do not. This means that the role of dividends as part of an effort to solve the agency problem and as a tool to signal the market is reduced in chaebol groups. It can be seen that the corporate governance structure of the chaebol plays an alternative role in the agency problem or the signaling theory.

Chaebol, CSM, and Dividend Policy (H2)
In addition, they are the empirical results supporting the viewpoint of the substitute hypothesis that the better the governance structure, the fewer the agency problems, so that dividends will not be paid excessively to further reduce agency costs [20,44]. It suggests that dividends are used less as a means of reducing agency costs due to the establishment of excellent corporate governance. In other words, it implies that the chaebols are operating as a mechanism to govern corporations in the relationship between CSM and dividends. Considering disclosure of additional firm related non-financial information reduces information asymmetry, and thus reduces agency cost between managers and shareholders or controlling shareholder and minority shareholder, therefore, it can be utilized as an alternative to dividend-payment in reducing the agency problem [25]. Table 5 shows the result of regression analysis of Equation (1), which scaled the dividend level with equity. The regression coefficients (β 1 ) of CSM (TOTAL_SCORE) on dividends was shown to be 0.182, which was significantly positive at the levels of 1%. H 1 was also supported when the dividend level was scaled with equity instead of total assets. Although not presented in the table, H 2 was supported. CSM activities can have an impact on earnings creation. CSM activities can have an impact on profit creation. Dividends can increase due to the augmentation in earnings-generating capacity from CSM activities. Such benefits are expected to flow through to investors in the form of increased earnings and strong dividend policy, resulting in a win-win scenario for firms, investors and stakeholders [15].  Table 6 shows the result of regression analysis of Equation (1) considering time series and cross-sectional dependencies. The methodology of Gow et al. [58] was applied to control the time series and cross-sectional dependencies. The regression coefficient (β 1 ) of CSM (TOTAL_SCORE) on dividends was shown to be 0.0006, which was significantly positive at the levels of 1%. That is, robust empirical results were shown because H 1 was supported even when the time series and cross-sectional dependencies were controlled. Although not presented in the table, H 2 was supported.

FCF, CSM, and Dividend Policy
In order to examine the effect of the agency problems on the downward inelasticity of costs, Chen et al. [59] used surplus funds and the tenure of office of the representative as the characteristic of companies where more the agency problems can occur to determine firms with high possibilities of occurrence of the agency problems. Therefore, in this study, free cash flow was selected as a characteristic of firms where the agency problems may occur to carry out additional analysis. In Table 7, the regression coefficients (β 3 ) of CSM × FCFDUM that show the relationships between CSM and dividends according to the levels of free cash flow was shown to be 0.0007, which was significantly positive at the levels of 5%. As a result of further analysis, H 2 was supported, but we did not report it in the paper. This means that the free cash level strengthens the positive relationship between CSM and dividends. That is, this indicates that firms with a high level of free cash increase dividends as part of effort to reduce agency costs because their agency costs are larger [59]. Table 6. The relationship between CSM and dividend policy (using the methodology of Gow et al. [58]).  [58]). ***, and * represent significance at the 0.01, and 0.1 level, respectively. Please see Abbreviations for variable definitions.

Board Size, CSM, and Dividend Policy
The size of the board of directors is a major factor that determines corporate governance [60]. Various studies have been conducted on the effective size of the board of directors in enhancing the quality of the responsibility of managers [61]. Larger boards of directors have collective expertise and are better able to perform their duties [62]. Jensen [63] argued that boards of directors with eight or more members can be efficient in achieving effective performance. Ezat and El-Masry [64] stated that a large board of directors would increase the timely reporting of financial statements. On the other hand, he argues that as the size of the board of directors grows, communication problems arise, resulting in poor performance, reduced participation rates, and more conflicts of interest before reaching an agreement. Therefore, this study further analyzed how the size of the board of directors affects the relationship between CSM and dividends [65].
In Table 8, the regression coefficients (β 3 ) of CSM × B_SIZEDUM (TOTAL_SCORE × B_SIZEDUM) representing the relationships between CSM and dividends according to the sizes of the board of directors was shown to be 0.0004, which was significantly positive at the levels of 10%. As a result of further analysis, H 2 was supported, but we did not report it in the paper. This means that the size of the board of directors strengthens the positive relationship between CSM and dividends. In other words, this result supports the managerial opportunism hypothesis that the larger the board of directors, the better the governance, and the better the governance structure is, the more the dividends will be increased to reduce the agency problems of managers. Foreign investors prefer firms with high reliability of accounting information when making investment decisions. Jeon [66] reported that the smaller the absolute value of discretionary accruals, the higher the earnings persistence coefficient, and the smaller the standard deviation of earnings per share, the more the foreign ownership increases. Kim and Cho [67] reported that foreign investors use the quality of accounting incomes as an important indicator for investment decision making. In addition, when evaluating the intrinsic value, foreign investors can identify firms with excellent earnings quality based on their international experience and expertise, and faithfully perform the role of monitoring corporate management for invested firms to induce qualitative improvement of accounting incomes [68]. As such, foreign ownership operates as a corporate governance.
In Table 9, the regression coefficients (β 3 ) of CSM × FORDUM (TOTAL_SCORE × FORDUM) representing the relationships between CSM and dividends according to foreign ownership was shown to be 0.0013, which was significantly positive at the levels of 5%. As a result of further analysis, H 2 was supported, but we did not report it in the paper. This is an empirical result indicating that firms with high foreign ownership strengthen the relationship between CSM and dividends. That is, this result supports the managerial opportunism hypothesis that firms with high foreign ownership have excellent governance and that the better the governance, the more the dividends will be increased to reduce the agency problems of managers.

Conclusions
This study analyzed the effects of CSM on dividend policy using 4718 firm-years samples from 2011 through 2018. In addition, the effects of CSM on dividends according to whether or not belonging to a chaebol group were studied. Jensen [5] stated that continuous dividend payments can reduce the agency problems by minimizing the firm's free cash flow. In cases where a firm has free cash flow, the manager can have an incentive to expand investments to investment proposals that would not create value. Therefore, the agency problems due to overinvestment can be reduced by minimizing free cash flow through dividends. Miller and Rock [6] argued that managers can use dividends as a means to deliver superior information on corporate prospects to the market.
Meanwhile, studies on CSR are linked to major detailed topics of the corporate finance theory. Considerable studies have been conducted on the causal relationship between CSR activities and corporate value. In particular, the causal relationship between CSR activities and capital structure decisions/dividend policy has recently been attracting attention as a research topic. However, regarding the relationship between social responsibility activities and dividend policy, previous studies do not present consistent analysis results.
A study conducted in the early days after stakeholder theory [9] first suggested the logic that if more investments are made in social responsibility activities, the dividend payout ratio will decrease, and the results of empirical analysis indicate a negative relationship between social responsibility activities and dividend payout ratios. On the other hand, Rakotomavo [69] empirically shows that corporate dividends and social responsibility activities have a complementary relationship, and therefore, the level of social responsibility activities has a significant positive relationship with the dividend prediction errors. Benlemlih [20] also shows that high CSM firms pay more dividends, because firms use dividend policies to manage the agency problems caused by excessive investments in social responsibility activities. As such, previous studies do not provide consistent results regarding the relationship between social responsibility activities and dividend policies. Therefore, this study aims to empirically analyze the effect of the unique corporate governance structure in South Korea (chaebol) on the relationship between CSM and dividends.
This study is expected to have the following implications. This study is differentiated in that it examined the relationship from the viewpoint of financial decision-making termed CSM activities and dividends unlike previous studies. This study is expected to present additional determinants for various dividend policies of firms. In previous studies, factors in which dividend policies differ from firm to firm have been presented as various factors such as information asymmetry, the agency problems, dividend-related tax policies, and corporate financial characteristics. However, we suggest that CSM activities can also be a significant factor for the diversity of dividend policy. Furthermore, this study is significant in that it directly examined the role of a chaebol groups in the relationship between CSM activities and dividends. This study demonstrated that CSM activities can also affect the decision-making of managers belonging to a chaebol group. This study's findings provide evidence that expands the meaning and role of CSM activities. Given that resolving the interest incompatibility between investors and managers is the focus of corporate governance, dividend policy can be used as a method of resolving such incompatibility. Finally, as the relationship between CSM activities and dividends seems to be weakening in the chaebol group, the dividend level is decreasing as the importance of understanding of other stakeholders (employees, consumers, suppliers, local communities, etc.). In other words, if investment activities for non-financial stakeholders are defined as CSM, the increase in CSM activities in the chaebol group can be inferred as reflecting the preferences of stakeholders other than shareholders.
The limitations of this study are first, it did not consider all the variables that affect CSM activities and dividends. Second, CSM activities are defined as non-financial indicators. Although this study revealed the fact that non-financial activities are related to dividends, a financial indicator, further deliberation is considered necessary to examine the logic to connect the two. Third, attention is required in the interpretation of regression analysis in that regression analysis is not intended to draw and apply a conclusion, but to assume a direction or roughly analyze something due to limitations in the basic assumptions and the application of analysis results. Fourth, in this study, the relationship between corporate governance and dividends was described in terms of the alternative hypothesis. However, there is a possibility that the chaebol is considered less important from the viewpoint of the agency theory and the signaling theory, so that it is not a problem of corporate governance structure. It cannot be ruled out that this is the result of the selfishness of the stakeholder or for reasons not explained.
In addition, the conceptual definition of CSM activities is expanding, and measuring indicators are gradually being diversified and elaborated. Future studies seem necessary to consider these areas. Examining the effects of CSM activities by stakeholder could be also meaningful. Acknowledgments: The authors are grateful to the anonymous reviewers and editor for their comments and suggestions on this study.

Conflicts of Interest:
The authors declare no conflict of interest. the natural log of total assets LEV leverage, total debts/total assets ROA the return on assets, pretax income/lagged total assets AGE the natural log of the number of years between t−1 year and initial listing year OWN the ownership ratio FOR the foreign ownership ratio B_SIZE the board size, the natural log of the number of registered directors B_RATIO outside directors ratio, outside directors/registered directors CASH cash equivalent ratio, (cash and cash equivalents + current financial assets)/total assets FCF free cash flow/total assets DIV t−1 lagged dividend level, lagged dividend/lagged total assets GRW growth rate, (total assets-lagged assets)/lagged assets YD

Abbreviations
year dummy ID industry dummy