Introduction

Our paper contributes to the open debate on the financial determinants of Responsibility (CSR) activities by studying whether firms' dividend policy has an impact on the level of corporate social performance (hereafter, CSP). This issue, has, to the best of our knowledge, not been investigated so far and we attempt to fill this gap in the literature. Indeed, some previous studies have tried to understand to what extent the implementation of CSR activities may impact dividend policy (Kim and Jeon, 2015; Cheung et al., 2016; Samet and Jarboui, 2017). The literature seems to underline a positive effect of CSP on dividend payout. However, whereas the impact of dividend policy for other stakeholders is of particular interest, the reverse causality has never been studied before. We therefore specifically study to what extent the level of dividends paid to investors as financial stakeholders has an impact on the level of CSP, which can be considered as "the ability to manage and satisfy the different corporate stakeholders" Igalens and Gond, 2005. Contrary to previous studies, this allows us to examine for the first time whether the payment of dividends to financial stakeholders is contradictory to satisfying non-financial stakeholders through a high level of CSR activities.

Literature provides mixed results on the explanation of dividend decisions, but all traditional theories converge in that they all postulate that firms have the goal of satisfying shareholder claims. For instance, dividends may be paid to mitigate the problem of shareholder expropriation (Easterbrook, 1984; Jensen, 1986) or to convey information to investors (Bhattacharya, 1979; Miller and Rock, 1985). However, according to the instrumental stakeholder theory (Donaldson and Preston, 1995; Jones, 1995), firms must take all stakeholder claims into account, and not only financial ones, in order to maximize their value. Clarkson (1995) explains that firms have an interest in honoring both financial and non-financial stakeholder claims, because the survival of the firms depends on their propensity to meet stakeholder claims. Firms that pay a high level of dividends in order to satisfy their financial stakeholders may then also try to maintain their reputation as a "desirable partner" Jones, 1995 and thus retain the participation of non-financial stakeholders by implementing CSR policies, because of the importance of non-financial stakeholders in the wealth creation process.

On the other hand, stakeholders may be afraid that the payment of dividends decrease the firm's ability to satisfy their claims (Shapiro, 1990; Holder et al., 1998), since dividends distributed to shareholders are amounts of money that cannot be used to satisfy stakeholder claims. In line with the slack resources theory, lower slack resources should lead to lower CSP, because managers can easily invest in CSR policies when slack resources are important (Waddock and Graves, 1997; Barnea and Rubin, 2010). Since high dividends should lead to less cash, it may lead to a diminished ability to implement CSR policies. Thus, a high level of dividend payout may also have a negative effect on the level of corporate social performance.

These different preceding theoretical predictions lead us to expect either a positive or a negative impact of the level of dividends paid on the level of CSP. Considering the rationale under these two points of view, we combine them in expecting the relationship between dividend policy and corporate social performance to be curvilinear. Hence, we posit that initially, a higher level of dividends may be followed by higher CSP. Given that relationships with stakeholders are important for success (Clarkson, 1995; Donaldson and Preston, 1995; Jones, 1995), after dividends payments are made, managers may try to maintain their reputation with non-financial stakeholders by implementing CSR policies in order to satisfy these non-financial stakeholders, according to distributive justice (Harrison and Wicks, 2013). Once a certain amount of dividends paid has been reached, however, a high dividend payout leads to a reduced capacity to subsequently invest in CSR activities. Indeed, when dividends paid are too high, firms may not have enough financial slack, then limiting the possibility for satisfying subsequent stakeholder claims. Thus, once a threshold is reached, we hypothesize that the relationship between dividend payout and CSP becomes negative. We therefore postulate that the relationship between the level of dividends paid and corporate social performance is U-inverted shaped.

Based on a worldwide sample of almost 7,000 firm-year observations between 2006 and 2010, results show that until a certain point, higher dividends are linked with higher subsequent CSP, in line with the distributive justice argument. When dividends are particularly high (superior to the 75^th percentile), then, the relationship between dividends and CSP becomes negative. In addition, we investigate the impact of dividend policy on different pillars of CSP and find that the relationships between dividend policy and both environmental and social pillars are also U-inverted shaped. Overall, our results are robust to the use of different measures of dividends and of CSP, and to different model specifications.

This study contributes to the literature in revealing the existence of an optimal amount of dividend paid in order to reduce the conflict between the satisfaction of financial and non-financial stakeholder claims. When restrained, the level of dividend paid does not appear to be detrimental to non-financial stakeholders. The positive effect of the level of dividends paid CSP seems to indicate that when the amount of dividend paid is reasonable, firms are able to satisfy both financial and non-financial stakeholders. However, dividends seem to be detrimental to non-financial stakeholders when the level of dividends becomes too high, with dividend yields greater than 5.58%.

The remainder of the article is organized as follows. In the next section, we present the data methodology. Then, the main empirical findings are discussed. The final section concludes.

Data and Methodology

Following Ioannou and Serafeim (2012), Cheng et al. (2014) or Eccles et al. (2014), we rely on corporate social performance scores from the Thomson-Reuters Asset4 database. We particularly focus on environmental and social scores, and exclude corporate governance and economic pillars from the overall score, to rule out direct conflicts between managers and financial stakeholders from our CSP proxy. Since the rating methodology may raise some concerns (Chatterji et al., 2014), we also use MSCI ESG Data to control for the robustness of the results. MSCI ESG Data, another worldwide leading ESG score provider, has also been used in preceding research such as Guenster et al. (2011) or Aktas et al. (2011).

Financial and control variables are from FactSet Fundamentals. For our control variables, we follow previous studies by Artiach et al. (2010) and Jiraporn and Chintrakarn (2013) and retain the following variables, which have been shown to be related to both CSR and dividend policy (Size, Leverage, and Returns On Assets). Hence, our model is as follows:

${CSP}_{i,t}={\alpha }_0+{\alpha }_1{Dividends}_{i,t-1}+{\alpha }_2{Dividends²}_{i,t-1}+{\alpha }_3{Size}_{i,t-1}+{\alpha }_4{Leverage}_{i,t-1}$


$+{\alpha }_5{ROA}_{i,t-1}+Year fixed effects+Industry fixed effects+Country fixed effects$


$+ {\epsilon }_{1,t}\left(1\right)$

It can be noticed here that quadratic models have already been used in the CSR literature. For instance, Barnett and Salomon (2012) have shown a curvilinear relationship between ROA and CSP. However, our work differs from this paper. Contrary to these authors, we aim to better understand the determimants of CSP, instead of investigating the impact of CSP of financial performance.

Table 1: Sample description.

Matching ESG data from Thomson Reuters Asset4, MSCI ESG Data and financial data over the 2006-2010 period leads to almost 7,000 worldwide observations. Sample description and descriptive statistics are presented respectively in Tables 1 and 2.

Empirical Results

Our model includes a one-year lag between the dependent and independent variables. We report the results of estimates of equation (1) in columns 1 to 3 of Table 3, where our variable of interest is div_yield, defined as Dividend per share divided by price per share. Our dependent variable is the overall CSP measure in column 1 (CSP_A4), environmental performance measure (Environment_A4) in column 2 and social performance measure (Social_A4) in column 3. Ordinary regressions and robust standard errors, clustered at the firm level have been used. The coefficients of div_yield are positive and highly significant for each model specification. The level of dividends paid seems then to have a positive impact on lead CSP, regardless of the CSP proxies.

This result shows that high dividend paying firms subsequently try to satisfy non-financial stakeholder claims by implementing CSR policies, since considering the involvement of each stakeholder must lead to higher financial performance (Clarkson, 1995). In addition, we see that the coefficients of div_yield² are significantly negative, which confirms a U-inverted shaped relationship. Hence, the positive impact of dividends on CSP becomes negative when dividends are high, since firms have fewer financial resources, which limits their ability to satisfy stakeholder claims. These results are robust to alternative measures of dividends (dividends by earnings, dividends by sales and dividends by total assets), not reported here for the sake of brevity but available upon request.

Table 3: Curvilinear relationship between dividend yield and CSP measured by ASSET4

This table shows the results of our ordinary least-squares regressions concerning the relationship between dividend yield and CSP. In columns 1 to 3, CSP is measured by Asset4 ratings: our dependent variable is overall CSP measure in column 1, environmental performance measure in column 2 and social performance measure in column 3. In column 4, our dependent variable is a dummy variable separating firms with high CSR ratings and low CSR ratings. In column 5, the CSP measure is the overall CSP measure proxied by MSCI ESG ratings. All regressions include year, industry and country dummies. Robust t-statistics are in brackets. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

Figure 1 presents the U-inverted relationship between div_yield and CSP_A4 graphically. The effect of div_yield on CSP becomes marginally negative when div_yield is around 5.58%. Based on Table 1, we note that 5.58% is above the 75^th percentile (3.36%). This reinforces our theoretical argument for testing for a non-linear relationship between the level of dividends paid and CSP: initially, div_yield has a positive impact on CSP_A4, but when div_yield becomes particularly high, the impact is negative. We also find a U-inverted relationship when we use Environment_A4 and Social_A4 as dependent variables.

Overall, it appears that a high dividend level seems to be followed by high CSP up to a threshold level, beyond which higher dividends negatively affect the future level of CSP. We therefore document a positive relationship between dividends and subsequent CSP, as long as the level of dividends paid is reasonable. But when the level of dividends paid reaches a high threshold, the relationship becomes negative, because paying too much dividends reduces slack resources needed to implement CSR policies.

Save View full size Expand inline Collapse inline

Figure 1: Relationship between dividend yield and CSP.

Note: This figure depicts average predicted values generated from Model 1 of Table 3. The line represents the average predicted values generated by the model, all else being equal, highlighting graphically the inverted-U relationship.

In order to check the sensitivity of our results against measures of dependent variables, we have used alternative variables for CSP. In Columns 4 of Table 3, a dummy variable

separating firms with high CSR ratings, i.e. above the mean of our sample, and low CSR ratings, has been tested. Our probit model shows that our previous results are not affected. In a similar way, we have re-estimated equation (1) with the CSR rating provided by MSCI ESG Research, another leading CSR rating agency, as an alternative CSR proxy. Here again, the U-inverted relationship between dividends and CSP remains unchanged.

We also dealt with endogeneity concerns. Indeed, results provided in Table 3 seem to indicate that a high level of dividends is followed in the next period by higher CSP, as long as a certain threshold is not reached. It is however possible to argue that the level of CSP also has an impact on a firm's dividend policy, as it has been demonstrated in some recent papers (Kim and Jeon, 2015; Cheung et al., 2016; Samet and Jarboui, 2017). For instance, according to the agency theory, dividend payment is a means to reduce the risk of shareholder expropriation by managers Jensen, 1986. Thus, it is possible to posit that high CSP leads to a higher dividend payout, since CSR policies may be linked with the extraction of managers' private benefits (Barnea and Rubin, 2010). Hence, we estimate two-stage least-squares regressions in order to deal with this potential endogeneity issue.

We instrument div_yield by countrysector, which is computed as the average dividend yield for the firm’s country and sector. We thus assume that a firm's dividend policy is impacted by sectorial and institutional determinants, as mentioned by Michel and Shaked (1986). Hence, investment opportunities are comparable for firms in the same sector, which may have an impact on the dividend level of firms in this sector. Moreover, the characteristics of some countries, such as their capital market mechanism or tax levels, may also have an influence on dividend policy. Concerning the quadratic terms, which may also be endogenous (div_yield²), we have used countrysector². Moreover, we have included all control variables and quadratic control variables as instruments. The estimations of the two-stage least-squares regressions are presented in Table 4. We observe that the coefficients on div_yield are still positive and highly significant for each specification, whereas coefficients on div_yield² are still negative and highly significant, confirming that our previous results do not suffer from endogeneity.

Table 4: Two-stage least-squares regressions (second step).

This table depicts results of the second step of our two-stage least-squares regressions. Our instruments are countrysector, (the average dividend yield for the country and the sector of the firm), countrysector², and all control variables and quadratic control variables. We include year, industry and country dummies. T-statistics are in parentheses. * Statistical significance at the 10% level. ** Statistical significance at the 5% level. *** Statistical significance at the 1% level.

Conclusion

This paper investigates the relationship between dividend policy and CSP. On one hand, dividends paid may lead to subsequent CSP in order to satisfy all stakeholders according to the distributive justice argument of stakeholder theory. On the other hand, economic theory predicts that dividends to shareholders paid may be at the expense of other stakeholders and CSP.

Based on a worldwide dataset and on different CSP ratings, we reconcile these two arguments in showing a strong U-inverted shaped relationship between dividends paid and CSP. The level of dividends paid is positively related to subsequent CSP until the level of dividends paid reaches a maximum. Due to how important it is to maintain good relationships with stakeholders in order to maximize a firm's value Clarkson, 1995Jones, 1995, high dividend paying firms seem to foster higher subsequent CSP, because these firms then invest in CSR activities in order to maintain the commitment of non-financial stakeholders. Indeed, a good reputation with these stakeholders is crucial for firms to achieve their financial goals (Clarkson, 1995), and they may interpret a high level of dividends paid as a decreased ability to act in their interests (Shapiro, 1990; Holder et al., 1998). When a certain high level of dividends paid is reached, however, this relationship is inverted, because after this threshold, dividends paid reduce the financial resources needed to implement CSR policies. Hence, the relationship becomes negative.

We find the same results for specific pillars of CSP such as environmental and social performance and they prove to be also robust to different model specifications and to different measures for both variables of interest and endogeneity concerns. Our study contributes to the literature concerning the determinants of CSP by revealing the U-inverted impact of the dividend policy on firms' CSP, and showing that there is an optimal dividend policy, around 5.5 percent of dividend yield, that seems to satisfy, all else equal, both shareholders and stakeholders.

References

• Artiach T., Lee D., Nelson D., Walker J., 2010. The Determinants of Corporate Sustainability Performance. Accounting & Finance 50, 31-51
• Barnea A., Rubin A., 2010. Corporate Social Responsibility as a Conflict between Shareholders. Journal of Business Ethics 97, 71-86
• Barnett M.L., Salomon R.M., 2012. Does It Pay to Be Really Good? Addressing the Shape of the Relationship between Social and Financial Performance. Strategic Management Journal 33, 1304-1320
• Bhattacharya S., 1979. Imperfect Information, Dividend Policy, and "the Bird in the Hand" Fallacy. The Bell Journal of Economics 10, 259-270
• Chatterji A., Durand R., Levine D.I., Touboul S., 2014. Do Ratings of Firms Converge? Implications for Managers, Investors and Strategy Researchers. Forthcoming, Strategic Management Journal
• Cheng B., Ioannou I., Serafeim G., 2014. Corporate Social Responsibility and Access to Finance. Strategic Management Journal 35, 1-23
• Cheung A.W., Hu M., Schwiebert J., 2016. Corporate Social Responsibility and Dividend Policy. Accounting & Finance
• Clarkson M.B.E., 1995. A Stakeholder Framework for Analyzing and Evaluating Corporate Social Performance. Academy of management review 20, 92-117
• Donaldson T., Preston L.E., 1995. The Stakeholder Theory of the Corporation: Concepts, Evidence, and Implications. Academy of management review 20, 65-91
• Easterbrook F.H., 1984. Two Agency-Cost Explanations of Dividends. The American Economic Review 74, 650-659
• Eccles R.G., Ioannou I., Serafeim G., 2014. The Impact of Corporate Sustainability on Organizational Processes and Performance. Management Science 60, 2835-2857
• Guenster N., Bauer R., Derwall J., Koedijk K., 2011. The Economic Value of Corporate Eco-Efficiency. European Financial Management 17, 679-704
• Harrison J.S., Wicks A.C., 2013. Stakeholder Theory, Value, and Firm Performance. Business Ethics Quarterly 23, 97-124
• Holder M.E., Langrehr F.W., Hexter J.L., 1998. Dividend Policy Determinants: An Investigation of the Influences of Stakeholder Theory. Financial Management, 73-82
• Igalens J., Gond J.-P., 2005. Measuring Corporate Social Performance in France: A Critical and Empirical Analysis of Arese Data. Journal of Business Ethics 56, 131-148
• Ioannou I., Serafeim G., 2012. What Drives Corporate Social Performance? The Role of Nation-Level Institutions. Journal of International Business Studies 43, 834-864
• Jensen M.C., 1986. Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. The American Economic Review 76, 323-329
• Jiraporn P., Chintrakarn P., 2013. How Do Powerful Ceos View Corporate Social Responsibility (Csr)? An Empirical Note. Economics Letters 119, 344-347
• Jones T.M., 1995. Instrumental Stakeholder Theory: A Synthesis of Ethics and Economics. Academy of management review 20, 404-437
• Kim J., Jeon Y., 2015. Dividend Policy and Corporate Social Responsibility: A Comparative Analysis of Multinational Enterprise Subsidiaries and Domestic Firms in Korea. Emerging Markets Finance and Trade 51, 306-319
• Michel A.J., Shaked I., 1986. Country and Industry Influence on Dividend Policy: Evidence from Japan and the USA. Journal of Business Finance & Accounting 13, 365-381
• Miller M.H., Rock K., 1985. Dividend Policy under Asymmetric Information. The Journal of Finance 40, 1031-1051
• Samet M., Jarboui A., 2017. Corporate Social Responsibility and Payout Decisions. Managerial Finance 43, 982-998
• Shapiro A.C., 1990. Modern Corporate Finance: Macmillan New York
• Waddock S.A., Graves S.B., 1997. The Corporate Social Performance–Financial Performance Link. Strategic Management Journal 18, 303-319