Corporate Social Responsibility and CEO Compensation Structure

Transcription

1 Corporate Social Responsibility and CEO Compensation Structure Khondkar Karim a Manning School of Business, University of Massachusetts Lowell Eunju Lee b Manning School of Business, University of Massachusetts Lowell Sanghyun Suh c Manning School of Business, University of Massachusetts Lowell June 2015 Abstract We examine how a firm's corporate social responsibility (CSR) performance affects the structure of CEO compensation. Traditional agency theory suggests that CEOs actively engage in CSR for their own interests at the expense of shareholders, and our finding of a positive association between a firm's CSR performance and CEO total compensation seems to support this argument. However, a more detailed investigation reveals that the proportion of equity-based compensation is relatively high for socially responsible firms, whereas the proportion of cash-based compensation is relatively low. This suggests that a firm's social performance improves the CEO compensation structure in a way to motivate managers to maximize shareholders' values in the long run. The positive association between CSR and the proportion of equity-based compensation is more pronounced in certain CSR categories such as employee relations, and the main results are driven by the post-sox period. Overall, our results highlight the positive impact of CSR performance on the design of the CEO compensation package. Keywords: Corporate Social Responsibility (CSR), CEO compensation, CEO compensation structure, equity-based compensation, cash-based compensation a Manning School of Business, 1 University Avenue, University of Massachusetts Lowell, Lowell, MA 01854, Tel: , khondkar_karim@uml.edu. b Corresponding author. Tel: , eunju_lee@uml.edu. c Tel: , sanghyun_suh@uml.edu.

2 Corporate Social Responsibility and CEO Compensation Structure Abstract We examine how a firm's corporate social responsibility (CSR) performance affects the structure of CEO compensation. Traditional agency theory suggests that CEOs actively engage in CSR for their own interests at the expense of shareholders, and our finding of a positive association between a firm's CSR performance and CEO total compensation seems to support this argument. However, a more detailed investigation reveals that the proportion of equity-based compensation is relatively high for socially responsible firms, whereas the proportion of cash-based compensation is relatively low. This suggests that a firm's social performance improves the CEO compensation structure in a way to motivate managers to maximize shareholders' values in the long run. The positive association between CSR and the proportion of equity-based compensation is more pronounced in certain CSR categories such as employee relations, and the main results are driven by the post-sox period. Overall, our results highlight the positive impact of CSR performance on the design of the CEO compensation package.

3 1. Introduction Prior literature shows that corporate social responsibility (CSR) has great impact on various aspects of accounting and finance. These studies show that a firm's CSR activity improves operating and financial performance (McGuire et al. 1988; Waddock and Graves 1997; Hillman and Keim 2001; Jiao 2010), reduces the cost of capital (Sharfman and Fernando 2008; El Ghoul et al. 2011), and increases stock price (Shane and Spicer 1983; Hamilton 1995; Karpoff et al. 2005). It is also found that firms engaging in CSR are less likely to manage their earnings (Kim et al. 2012). On the other hand, a few studies do not find evidence on the effect of CSR on firm performance. Nelling and Webb (2009) find no evidence of the positive impact of CSR on firms' performance, and Brammer et al. (2006) show that socially responsible firms tend to have lower returns. The effect of CSR activity has also been investigated in the field of corporate governance. Some recent studies examine the association between CSR performance and CEO compensation, and this issue has been studied in two ways. One strand of this literature investigates the impact of CEO compensation on CSR. These studies examine if executive incentives play an important role in firms' engagement in CSR activity, and they find mixed evidence. McGuire et al. (2003) find that high levels of CEO compensation lead to poor CSR performance. They argue that performance pressures driven by such incentives make CEOs more focused on short-term targets, preventing them from engaging in CSR activities. 1 Mahoney and Thorn (2006) claim that executive compensation structure is an important factor in explaining firms' CSR performance, finding a negative effect of salary on CSR and a positive effect of stock options on CSR. The other literature examines the impact of CSR on CEO compensation. Classic agency theory introduced by Jensen and Meckling (1976) argues that CEOs tend to pursue their own interests rather than to maximize shareholders' value. This supports managers' tendencies to invest in CSR for the purpose of hiding their wrongdoings such as corporate misconduct (Hemingway and Maclagan 2004). In the same vein, CSR can be used as a tool for CEOs to increase their own benefits such as increasing reputation and strengthening bargaining power, leading to overinvestment in CSR activities (Milbourn 2003; Barnea and Rubin 2010). This story suggests a 1 Agency theory suggests that CEOs may use CSR activities to pursue their own interests. In this respect, the negative impact of CEO compensation on CSR performance could suggest that highly (poorly) paid CEOs are less (more) motivated to improve firms' social performance for their own benefits. 1

4 positive association between CSR and CEO compensation. That is, firms with high CSR performance are expected to pay high levels of CEO compensation. On the other hand, CSR can be negatively associated with CEO compensation. CEOs of socially responsible firms may tend to receive less compensation for various reasons. First, their strong performance in CSR makes them proud of being "an exemplary CEO" and internally rewarded by doing the right thing. In this case, CEOs of high CSR firms are willing to trade their financial benefits for such satisfaction (Potts 2006; Rekker et al. 2014). Second, low levels of CEO compensation could be the natural consequence of a firm's CSR activity, because socially responsible firms may reduce the compensation gap between CEOs and other employees to resolve conflicts between managers and other stakeholders (Cai et al. 2011). We extend the above studies by investigating the relation between CSR performance and CEO compensation structure. In particular, we shed light on the effect of a firm's social performance on the composition of the CEO compensation package. The CEO compensation structure, measured by the proportions of cash- and equity-based compensation over total compensation in this study, is a crucial factor in inducing executives to take on riskier investments and pursue long-term profits in alignment with the interests of shareholders. Given this, the relation between CSR performance and the components of CEO compensation can be interpreted in several ways. First, the impact of CSR on the composition of CEO compensation suggests the types of compensation packages preferred by CEOs who work for firms with different levels of CSR activities. For instance, if CEOs are entrenched enough to be involved in CSR activities to pursue their own interests, they may prefer cash-based compensation based on short-term performance rather than incentive-based compensation. In contrast, if CEOs are willing to sacrifice their financial compensation for doing the right thing, such confidence may make them receive more equity-based compensation based on long-run performance. If a negative association between CSR performance on CEO compensation is led by a firm's effort to improve its relations with employees and shareholders, low levels of CEO total compensation for socially responsible firms will be more likely to be driven by a lower proportion of cash-based compensation. While prior studies primarily focus on interpreting the relation between CSR and overall levels of CEO compensation from the perspective of agency theory, we consider all the possibilities of how the relation differs depending on the components of compensation. 2

5 Second, the association between CSR performance and CEO compensation implies how different levels of firms' CSR activities affect the board of directors' decision to design the compensation package. Optimal contracting theory argues that the ideal structure of the CEO compensation package from the perspective of the board of directors would be the one that motivates CEOs to align their interests with the interests of firm shareholders (Core and Guay 1999; Fama and Jensen 1983; Jensen and Meckling 1976; Murphy 1985). In this sense, our results will answer the research question of whether a firm s social performance leads to the right mix in the compensation package to improve firm value or act as an impediment to mitigating the agency problem. Third, the relation between CSR and CEOs' cash-based or equity-based compensation provides insights into its impact on firm value. Even though CSR performance influences CEO compensation, its impact on firm value may vary depending on the structure of the compensation. If CEOs strategically use CSR performance for their own interests, as suggested by Barnea and Rubin (2010), and it leads to increasing cash-based compensation based on short-term performance targets, agency problems will be amplified, resulting in decreases in firm value. On the contrary, if the compensation structure becomes more incentive-based with a firm's CSR activity, increases in CEO compensation with CSR performance will reduce agency problems by aligning CEO incentives with shareholders' interests and increase firm value in the long run. We revisit this issue by investigating if CEOs' active engagement in CSR affects the composition of the compensation package in a way that reduces or increases firm value. Before we analyze the effect of a firm's CSR performance on CEO compensation structure, we first examine if firms' disclosure of CSR information affects CEO compensation and its components. We find that firms releasing their CSR information tend to have higher levels of CEO total compensation compared to firms without CSR information. This suggests that CSR disclosure improves the quality of CEOs by motivating CEOs to work harder to earn more compensation. Meanwhile, this result turns out to be different when we look into the components of CEO compensation, and, in particular, the proportion of cash-based compensation is lower for firms with CSR disclosures. The results show that firms' CSR disclosures affect CEO total compensation and its structure in a way that provides CEOs incentives to pursue shareholders' interests. 3

6 When we investigate the association between CSR performance and CEO total compensation, we find that firms with high CSR performance tend to pay CEOs more in total compared to firms with low CSR performance. The results seem to be consistent with Jensen and Meckling (1976), suggesting that CEOs tend to improve firms' social performance to pursue their own interests. However, further analysis on CSR performance and compensation components shows that such high levels of total compensation for high CSR firms are mainly composed of equity-based compensation. Socially responsible firms tend to have a relatively low proportion of cash-based compensation and a high proportion of equity-based compensation, while socially irresponsible firms have a relatively high proportion of cash-based compensation and a low proportion of equity-based compensation. This provides evidence against CEOs' self-interested behavior of using CSR for their own interests, because a firm's CSR performance affects the structure of CEO compensation in the direction of mitigating the agency problem and motivating managers to maximize shareholders' value. It also shows that, even if high levels of CEO compensation for socially responsible firms are attributed to CEO entrenchment, changes in the compensation structure will end up attenuating agency problems. The results are more or less pronounced depending on certain dimensions of CSR and the passage of the Sarbanes-Oxley Act of 2002 (SOX). While the negative relation between CSR and the proportion of cash-based compensation is more pronounced in CSR areas of environment and human rights, the positive relation between CSR and the proportion of equity-based compensation is more pronounced in CSR areas of environment, diversity and employee relations. It is also found that these results are primarily driven by the post-sox period. Overall, our results indicate that the boards of directors of high CSR firms are more likely to design the structure of CEO compensation to align managers' interests with those of shareholders. These findings are consistent with Haugen and Senbet (1981) and Agrawal and Mandelker (1987), who argue that equity-based compensation, such as restricted stocks and stock options, plays an important role in inducing managers to increase shareholders' value and, thus reduces agency problems. This is the first study to examine the effect of firms' CSR activities on CEO compensation structure with a focus on the intensity of each compensation component. While prior studies investigate on CEO total compensation or the levels of the compensation components, our study 4

7 on the proportions of cash- and equity-based compensation draws clearer implications on the impact of CSR on the structure of CEO compensation. In addition, our results imply the extensive future use of CSR information among stakeholders and investors, because a firm's CSR performance is not only an indicator of its financial and operating performance but also a measure of the degree of efficiency of a firm's corporate governance system. Outside investors can extract information about a firm's corporate governance from its social performance. The remainder of this paper is organized as follows. Section 2 reviews relevant literature and explains the hypotheses that will be tested, and Section 3 describes data and methodology. In Section 4, we discuss our empirical results of the effect of CSR on CEO compensation structure, and Section 5 addresses additional tests including robustness checks. Finally, Section 6 summarizes and concludes the paper. 2. Related Literature and Hypothesis Development Our main focus is on how a firm's social performance affects CEO compensation. When it comes to testing this, we presume that such relation may vary depending on the components of CEO compensation. Given this conjecture, we analyze three dimensions of CEO compensation: total compensation, the proportion of cash-based compensation, and the proportion of equity-based compensation. The effect of CSR on CEO total compensation is anticipated in two ways. A firm's good performance in CSR makes CEOs confident about doing the right thing. Thus, CEOs will be internally rewarded by firms' CSR activities, which leads to relatively low levels of total compensation (Potts 2006; Rekker et al. 2014). Alternatively, lower levels of CEO compensation can be interpreted as the product of a firm's attempt to improve its relations with employees. On the contrary, CEOs may intend to exploit firms' social performance for their own benefits (Milbourn 2003; Barnea and Rubin 2010). In this case, they will attempt to increase their reputation and obtain more bargaining power by improving firms' CSR performance. Such opportunistic behavior will ultimately lead to high levels of CEO total compensation. Given all these possibilities for the relation between a firm's CSR performance and CEO total compensation, we propose the following hypothesis: 5

8 H1. A firm's CSR performance is significantly associated with CEO total compensation. We also expect that a firm's CSR performance will affect the structure of CEO compensation. A firm's good CSR performance indicates that the firm is good at harmonizing different interests among stakeholders and attempts to improve its relation with employees. Given this, socially responsible firms are likely to have a higher proportion of equity-based compensation structure than socially irresponsible firms. Linking this with a negative relation between CSR performance and CEO total compensation, we predict that the negative association between CSR and total compensation will be driven by decreases in the proportion of cash-based compensation. In a traditional agency setting, CEOs prefer a less risky compensation structure with greater cash-based compensation rather than equity compensation based on long-term performance (Harris and Raviv, 1979; Westphal 1998). If CEOs overinvest in CSR activities to increase their own benefits and their entrenched behavior leads to high levels of total compensation, the positive relation between CSR performance and total compensation is expected to be driven by a relatively high proportion of cash-based compensation compared to the proportion of equitybased compensation. 2 Given these possible effects of firms' social performance on the components of CEO compensation, we establish the following hypotheses: H2. A firm's CSR performance is significantly associated with the proportion of cash-based compensation. H3. A firm's CSR performance is significantly associated with the proportion of equity-based compensation. 3. Data and Methodology 3.1. Data description 2 This is somewhat consistent with managerial power theory, which explains the effects of CEO's power on the setting of their compensation contracts (Bebchuk and Fried, 2003; Bebchuk and Fried 2004; Core, Holthausen and Larcker 1999; Hartzell and Starks, 2003). 6

9 We collect annual CSR data from the Kinder, Lydenberg, and Domini Stats database (KLD STATS) for the period KLD STATS includes social and environmental performance of more than 3,000 listed companies, and the database contains the company ID (CUSIP), binary indicators for strength and concern activities in thirteen CSR categories, and summary counts for all the areas. The CSR rating criteria consist of strength and concern indicators for seven qualitative issue areas and six controversial business issue areas. The seven major qualitative issue areas include community, corporate governance, diversity, employee relations, environment, human rights, and product, and the six controversial business issues are alcohol, gambling, firearms, military, nuclear power, and tobacco. 3 Using the KLD STATS dataset, we obtain several CSR measures used in our analysis. Our CSR measures are based on six main categories, community, diversity, employee relations, environment, human rights, and product. 4 We first calculate total strength and total concern scores separately by adding up strength and concern indicators across the six categories. After that, we obtain net CSR scores for each data observation by calculating differences between these total strength and concern scores. The net CSR score for each of the six CSR categories is also used in the detailed analysis. We collect CEO compensation data from ExecuComp for our sample period. The database includes the details of CEO compensation such as salary, bonus, restricted stocks, stock options, pensions, and other annual compensation. Based on this dataset, we calculate total compensation as the sum of salary, bonus, total value of restricted stocks and stock options, long-term incentive payouts, and all other compensation given for the fiscal year. We also calculate cash- and equitybased compensation, which are the sum of salary and bonus and the sum of restricted stocks and stock options granted respectively during the fiscal year. Using these level data, we calculate the proportions of cash- and equity-based compensation, which are key compensation measures of this study. Following Core et al. (1999), Leone et al. (2006), and Rekker et al. (2014), we control for total number of directors, percentage of independent directors on board, and several CEO characteristics, such as CEO ownership, duality, age, and tenure, since these variables possibly affect CEO compensation and its structure. 3 The details of CSR categories and strength/concern items are described in Appendix 1. 4 We exclude CSR scores on corporate governance in this study, since its definition contains CEO compensation. 7

10 We obtain financial statements data from COMPUSTAT and stock return data from CRSP. Utilities and financial companies (SIC codes and ) are excluded from the sample. A list of the variables and their definitions are reported in more detail in Appendix Descriptive statistics Table 1 reports summary statistics for key variables in the entire sample. Panel A shows the statistics of the CSR variables. The mean of the total CSR scores (TCSR) is positive (0.23), which is consistent with Rekker et al. (2014) and Cai et al. (2011). The mean of CSR strength scores (CSR_S) (1.92) is greater than the mean of CSR concern scores (CSR_C) (1.69). This shows that firms disclosing their CSR information tend to have relatively good CSR performance. In terms of each CSR category, the mean of net CSR scores is positive in community (COMM), environment (ENV), and diversity (DIV), while the mean of net CSR scores is negative in employee relations (ER), human rights (HR), and product (PROD). Panels B and C report the statistics of compensation, CEO and firm characteristic variables. The mean dollar amount of equity-based compensation (EQUITY) ($3.2 million) is much higher than the mean dollar amount of cash-based compensation (CASH) ($1 million). Consistently, the mean proportion of equity-based compensation (PEQUITY) (47.97%) is higher than the mean proportion of cash-based compensation (PCASH) (32.6%). The statistics of the other compensation and firm characteristic variables, such as board size (BSIZE), CEO ownership (CEO_OWN), book-to-market (BM), and Tobin's Q (TOBINQ), are in general consistent with the statistics shown in prior studies (Cai et al. 2011; Rekker et al. 2014). [Insert Table 1 about here] Table 2 reports Pearson correlations between CSR and compensation and CEO-specific variables. The correlation between total CSR score and total compensation is significantly positive (0.17), and this positive correlation is consistent when we look into levels of cash- and equity-based compensation. This seems to support that CEOs use CSR activities as a tool for their private interests. In this case, we predict that CEO entrenchment will also lead to a relatively high proportion of cash-based compensation compared to the proportion of equity-based 8

11 compensation. However, an examination on the proportions of the compensation components leads to conflicting results. While the proportion of cash-based compensation is negatively correlated with the total CSR score (-0.15), the proportion of equity-based compensation is positively correlated with the total CSR score (0.15). This suggests that the positive correlation between CSR performance and total compensation is primarily driven by the association between CSR and equity-based compensation. That is, although the correlation between CSR and total compensation seems to highlight CEOs' abuse of CSR activities, the detailed investigation of CSR performance and compensation structure suggests that socially responsible firms tend to have a relatively low proportion of cash-based compensation and a high proportion of equitybased compensation. Overall, the correlation results suggest that high CSR firms have compensation structures that minimize CEOs' moral hazards and induce CEOs to maximize shareholders' values. Since these findings are based on pairwise correlations between CSR and compensation variables in this section, we analyze the relation between CSR and compensation in more detail using regressions in Section 4. [Insert Table 2 about here] 4. Empirical Results 4.1. CSR disclosure and CEO compensation Prior to our analysis on CSR performance and compensation structure, we examine if a firm's disclosure of CSR information affects CEO total compensation and the proportions of cash- and equity-based compensation. The release of firms' CSR information to the public reduces information asymmetry. This is supported by Verrecchia (2001), who argues that a firm's disclosure reduces information asymmetry. If this is the case, a firm's disclosure on CSR activity enhances shareholders' role to monitor CEOs and leads the board of directors to gain more bargaining power in determining CEO compensation. Therefore, CSR disclosures are expected to be negatively associated with CEO compensation. In addition, given enhanced power of 9

12 shareholders and the board of directors driven by a firm's CSR disclosure, we expect that such negative relation between CSR disclosure and CEO total compensation will be mainly driven by decreases in the proportion of cash-based compensation based on short-term performance goals. On the other hand, a firm's disclosure on its CSR activity can lead to an increase in CEO compensation, since CSR information also reveals the overall quality of CEOs, resulting in improvement in their quality in the long run. Therefore, such environment induces CEOs to work harder to earn more compensation and bargaining power. This argument is in line with Hermalin and Weisbach (2012), who argue that a firm's enhanced disclosure positively influences both shareholders and CEOs through the corporate governance system, eventually leading to high levels of CEO compensation. By this logic, a firm's CSR disclosure will lead to an increase in CEO compensation. With respect to the structure of the compensation, we conjecture that the positive association between a firm's CSR disclosure and CEO compensation is mainly driven by increases in the proportion of equity-based compensation, since both shareholders and CEOs benefit from a reduction in information asymmetry in this case. In order to test this, we estimate the following regression model using a CSR disclosure indicator as a predictor variable: COMP it = α + β CSRDIS it 1 + θ γ + ε it (1) where COMP it is CEO total compensation ( TCOMP it ), the proportion of cash-based compensation (PCASH it ), or the proportion of equity-based compensation for firm i in year t (PEQUITY it ), CSRDIS it 1 is a CSR disclosure indicator which is 1 if firm i has CSR scores in year t-1 and 0 otherwise, and θ is a vector of lagged control variables including CEO and firm characteristic variables. 5 We run the above regression with year and industry fixed effects. Table 3 reports the regression results. When we use total compensation as a dependent variable, a coefficient on the CSR disclosure indicator is significantly positive (0.126). This indicates that firms with the disclosure of CSR information have high levels of CEO total compensation. As 5 The CEO characteristic variables include the logarithm of the number of directors on the board of directors(bsize), the percentage of outside directors on board (PIND), ownership by CEOs (CEO_OWN), CEO duality (CEO_DUR), CEO age (CEO_AGE), and CEO tenure (CEO_TEN ). The firm characteristic variables include the logarithm of sales (SALE), book-to-market (BM), the return-on-assets ratio (ROA), the Tobin's q ratio (TOBINQ), the leverage ratio (LEV), the liquidity ratio (LIQ), firm age (FAGE), market beta (BETA), aggregated returns for 12 months in years t-1 and t-2 (RET and LAGRET), and the standard deviation of aggregated returns over 5 years (STDRET). The detailed descriptions are reported in Appendix 2. 10

13 explained above, it demonstrates that the release of firms' CSR information motivates CEOs to exert effort and improves their quality. This eventually leads both shareholders and CEOs to benefit from firms' disclosures on their CSR activities. [Insert Table 3 about here] Under this hypothesis, we expect that this positive association will be attributable to a high proportion of equity-based compensation. In order to test this, we repeat the above regression using two components of CEO compensation as dependent variables, which are the proportions of cash- and equity-based compensation. Columns (2) and (3) report the regression results. The result in Column (2) shows that a coefficient on the CSR disclosure indicator is significantly negative (-0.085) when we use the proportion of cash-based compensation as a dependent variable. On the contrary, the coefficient in Column (3) is positive (0.011) in the regression of the proportion of equity-based compensation, but is not statistically significant. The results show that the positive association between CSR disclosure and total compensation does not hold when we examine the relation between CSR disclosure and the compensation components. Although the proportion of equity-based compensation is not significantly positively associated with CSR disclosure, the proportion of cash-based compensation significantly decreases with CSR disclosure. Given that cash-based compensation is based on short-term profits and its high levels can cause agency problems, our results imply that a firm's CSR disclosure does not affect its compensation structure in a way that amplifies the agency problem and reduces firm value. Overall, our results on a firm's CSR disclosure and CEO compensation highlight the positive impact of CSR disclosure on CEO compensation and its structure. A firm's disclosure on its CSR activity is positively correlated with CEO total compensation and is negatively correlated with cash-based compensation. These results suggest that a closer look into the proportion of each compensation component can offer insightful implications on the impact of a firm's CSR disclosure on CEO compensation. In the next section, we further investigate the impact of CSR performance on CEO compensation and its structure CSR performance and CEO compensation structure 11

14 We simply considered a firm's CSR disclosure rather than the degree of CSR performance in the previous section. However, investigating the effect of CSR performance on CEO compensation could lead to different results. Motivated by this, we examine how a firm's CSR performance is associated with total levels and each component of CEO compensation in this section. CSR is a combined measure of a firm's ethical consciousness for the communities and environment and its relationship with stakeholders such as employees, customers, and suppliers. Thus, a firm's high CSR score verifies its superior ability to manage relations with stakeholders. Some prior studies show that firms with good CSR performance tend to have lower levels of CEO compensation as part of their efforts to improve the relation with employees. In addition, CEOs of such a firm can be internally compensated for doing the right thing, which makes them accept relatively low levels of pay. Taken together, a negative relation between CSR performance and CEO total compensation is anticipated in this case. However, if CEOs actively engage in CSR activities for the purpose of increasing their reputation and strengthening their bargaining power, CEO compensation will increase with CSR performance. According to this story, we cannot rule out the possibility that CSR performance is positively associated with CEO compensation. We also conjecture that the relation between CSR and CEO compensation may differ depending on the components of CEO compensation. CEO compensation can be divided into cash- and equity-based compensation. Cash-based compensation is regarded as a short-term incentive, while equity-based compensation is known as an effective tool to mitigate agency problems by aligning CEOs' interests with those of shareholders. If a firm is socially responsible, its board of directors will be more likely to design the CEO compensation package in a desirable way for the effective corporate governance system, compared to boards of directors of socially irresponsible firms. Therefore, the proportion of equity-based compensation will be relatively higher for socially responsible firms, while the proportion of cash-based compensation will be relatively lower. If CSR performance is negatively correlated with CEO compensation, this argument enables us to predict that the negative association is driven by a relatively low (high) proportion of cash-based (equity-based) compensation. This story may not apply in the case that a positive association between CSR performance and CEO compensation is caused by CEO entrenchment. Given the prior finding that CEOs prefer 12

15 cash-based compensation to equity-based compensation, such positive relation between CSR performance and CEO compensation is expected to be driven by a relatively high (low) proportion of cash-based (equity-based) compensation. For this analysis, we estimate the following regressions of CEO compensation on CSR scores with year and industry fixed effects: COMP it = α + β TCSR it 1 + θ γ + ε it (2) where COMP it is CEO total compensation ( TCOMP it ), the proportion of cash-based compensation (PCASH it ), or the proportion of equity-based compensation for firm i in year t (PEQUITY it ), and TCSR it 1 is lagged total CSR score calculated as the sum of net CSR scores across six CSR categories (community, environment, diversity, employee relations, human rights, and product) in year t-1. 6 Our main interest in Equation (2) lies in the regression coefficient on the total CSR score, β. As addressed above, β is expected to be either negative or positive with the dependent variables of total compensation and the proportions of cash-based and equitybased compensation. In order to examine if test results are primarily driven by CSR strengths or concerns, we repeat the above estimation using CSR strength and concern scores as separate independent variables as follows: COMP it = α + β 1 CSR_S it 1 + β 2 CSR_C it 1 + θ γ + ε it (3) where CSR_S it 1 and CSR_C it 1 are total strength and concern scores for firm i in year t-1. Table 4 reports the estimation results of Equations (2) and (3). When we use CEO total compensation as a dependent variable in the first column, a coefficient on the total CSR score is significantly positive (0.012). It seems to suggest that CEOs tend to concentrate on CSR performance with the intention of increasing their own benefits. This positive relation between 6 We conjecture that CEOs' equity-based compensation is determined based on their performance for the previous year. In terms of cash-based compensation, however, Perry and Zenner (2001) and Comprix and Muller (2006) address that bonuses are based on the current year's performance, while salaries are determined before the year-end performance is released. Taken together, a firm's performance may contemporaneously affect the components of CEO compensation. Given this, we repeat all regression analyses using contemporaneous variables to ensure that our findings are robust. In unreported tables, the overall results are unchanged in terms of magnitude and significance. These results are available upon request. 13

16 CSR performance and CEO total compensation turns out to be mainly driven by a positive association between the CSR strength score and total compensation. When we use the CSR strength and concern scores separately as independent variables in the second column, coefficients on both scores are positive (0.021 and 0.003), but only the coefficient on the strength score is statistically significant. [Insert Table 4 about here] In Columns 3 through 6 of Table 4, we report regression results with the dependent variables of the proportions of cash- and equity-based compensation. If CEO entrenchment is the main cause of the positive relation between CSR performance and CEO compensation, we expect to find a positive association between CSR and the proportion of cash-based compensation and a negative association between CSR performance and the proportion of equity-based compensation. When we use the proportion of cash-based compensation as a dependent variable in the third column, the coefficient on the total CSR score is significantly negative (-0.002). On the other hand, when we use the proportion of equity-based compensation in the fifth column, the coefficient on the total CSR score is significantly positive (0.005). The results suggest that high levels of total compensation for socially responsible firms are mainly driven by a high proportion of equity-based compensation and a low proportion of cashbased compensation. This contradicts our prior expectation that CEOs may use CSR activity as a tool for their own benefits. Even though CEOs seem to engage in CSR activities for their own interests, a change in compensation structure prevents such behavior from increasing agency problems and hurting firm value. On the other hand, given the prior finding of the positive impact of a firm's CSR performance on its financial and operational performance, the positive relation between CSR and CEO compensation might simply indicate that CEOs are reasonably compensated for their strong performance in CSR. To sum up, we conclude that socially responsible firms have CEO compensation structures that mitigate agency problems and contribute to improving firm value. We also find that the associations between CSR and the proportions of cash- and equity-based compensation are mainly driven by CSR strength scores. The regression results with the independent variables of CSR strength and concern scores in the fourth and sixth columns show 14

17 that the coefficient on the CSR strength score (-0.004) is significantly negative for the proportion of cash-based compensation, while it is significantly positive (0.007) for the proportion of equity-based compensation. The signs of coefficients on the board of directors and CEO-specific variables from a regression of total compensation are consistent with prior studies. The coefficient on board size (BSIZE) (0.156) is significantly positive, and the coefficient on CEO ownership (CEO_OWN) (-0.018) is significantly negative. These results are consistent with Core et al. (1999) and Lambert et al. (1993), who show that CEO compensation is higher when boards are less effective and CEO ownership is higher. The coefficient on CEO duality (CEO_DUR) (0.031) is consistently positive but insignificant for our sample. On the contrary, the coefficient on the percentage of independent directors on board (PIND) (0.839) is significantly positive, showing that the dependence of the board increases CEO total compensation. 7 The coefficients on CEO age and tenure (CEO_AGE and CEO_TEN) are significantly positive (0.01 and 0.011), confirming the findings of Fahlenbrach (2009) that older CEOs and CEOs with longer tenure tend to receive higher compensation. Coefficients on firm-specific variables are also in line with prior findings. The coefficients on sales and book-to-market are consistent with Core et al. (1999), suggesting that CEOs tend to earn larger compensation when firms are larger and have higher investment opportunities proxied by low book-to-market ratios. The positive coefficients on stock return and the standard deviation of stock return show that CEOs tend to earn larger compensation when firms' stock returns are higher and the standard deviation of stock returns is higher. The coefficient on firm age (FAGE) (-0.107) is consistent with Fahlenbrach (2009), indicating that younger firms tend to pay higher CEO compensation. Overall, we find that firms with high CSR performance have relatively high levels of CEO compensation compared to firms with low CSR performance. An in-depth analysis of the compensation components shows that socially responsible firms tend to have a higher proportion of equity-based compensation and a lower proportion of cash-based compensation than socially irresponsible firms. These results suggest that a firm's CSR performance affects not only total 7 Core et al. (1999) also find a negative association between total compensation and the percentage of inside directors on the board, claiming that there is no evidence that the percentage of inside directors leads to high CEO compensation. 15

18 levels of CEO compensation but also the structure of CEO compensation in a way that reduces agency problems and induces CEOs to align their interests with those of shareholders The effects of CSR on the relation between CEO compensation and firm value We have shown that a firm's CSR performance has a great impact on CEO compensation structure. The proportion of cash-based compensation tends to decrease with a firm's CSR performance, while the proportion of equity-based compensation tends to increase with the CSR performance. The results have direct implications for the implementation of a firm's CSR policy and its dynamics with CEO characteristics. More importantly, however, the results imply that the relation between CSR performance and CEO compensation may essentially affect firm value and shareholders' wealth. The effects of CSR and CEO compensation structure on firm value have been investigated separately in existing studies. The agency theory suggests that granting equity-based compensation provides CEOs with incentives to align their interests with those of shareholders and maximize firm value, and it is supported by many early studies (Bryan et al 2000; DeFusco et al. 1990; McConnell and Servaes 1990; Mehran 1995). A growing literature documents that a firm's CSR activity improves firm performance and increases shareholders' values. Although we find that these two factors that crucially affect firm value are closely correlated, it is not clear how one of them affects the relation between the other and firm value. In particular, a firm's CSR performance may reinforce or mitigate the effects of the components of CEO compensation on firm value. Investigating this issue enables us to draw clear implications on the channels through which a firm's CSR performance influences firm value. Motivated by this, we investigate how a firm's CSR performance affects the relation between CEO compensation components and firm value. To analyze this, we run the regression of a measure for firm value on the total CSR score and the proportion of cash- or equity-based compensation and their interaction term as follows: TOBINQ it = α + β 1 TCSR it 1 + β 2 PCASH it 1 or PEQUITY it 1 + β 3 PCASH it 1 TCSR it 1 (or PEQUITY it 1 TCSR it 1 ) + θ γ + ε it (4) 16

19 Following Faleye (2007), we use Tobin's Q as a proxy for firm value. This is calculated as market value of assets over book value of assets, where market value is the sum of book value of assets, market value of common stock, and deferred taxes minus book value of common stock. For explanatory variables, we include the proportions of cash- and equity-based compensation and their interaction terms with the total CSR score. We also add to the equation some other control variables that can possibly affect firm value, such as board size (BSIZE), the percentage of outside directors on board (PIND), capital expenditure (CAPEX), leverage (LEV), sales (SALE), and operating profitability (OPPROF). Given the previous findings addressed above, we expect to find a positive association between the total CSR score and Tobin's Q. Also, there will be a positive (negative) relationship between the proportion of equity-based (cash-based) compensation and Tobin's Q. If a firm's CSR performance reinforces the positive impact of equity-based compensation on firm value, the coefficient on the interaction term between the proportion of equity-based compensation and the CSR score will be significantly positive. In the same sense, if a firm's CSR performance deteriorates the negative impact of cash-based compensation on firm value, the coefficient on the interaction term between the proportion of cash-based compensation and firm value will be significantly negative. Table 5 reports the estimation results. The coefficient on the total CSR score is positive, although it is not significant when we regress Tobin's Q on CSR, the proportion of equity-based compensation, and their interaction term in column (2). With respect to the components of CEO compensation, the coefficient on the proportion of cash-based compensation (-0.353) is significantly negative, while the coefficient on the proportion of equity-based compensation (0.562) is significantly positive. The results in general support the previous findings that CSR performance and equity grants to CEOs have positive impacts on firm value. In column (2), the coefficient on the interaction term between the proportion of equity-based compensation and Tobin's Q (0.043) is significantly positive. This suggests that a firm's CSR performance strengthens the positive effect of equity-based compensation on firm value. Given our finding of the positive association between CSR performance and the proportion of equitybased compensation, the result highlights the positive aspect of CSR on both CEO compensation structure and firm value. Socially responsible firms tend to have a higher proportion of equitybased compensation, and the positive marginal impact of higher equity-based compensation on 17

20 firm value is stronger for these firms. This demonstrates that the dynamics between CSR performance and CEO compensation components boost firm value, which makes a firm's engagement in CSR activities more valuable. Overall, our results reveal that a firm's CSR performance not only increases the proportion of equity-based compensation but also reinforces its positive role in firm value. 5. Additional Tests 5.1. CSR categories and CEO compensation structure In the previous sections, we used the total CSR score as a measure of CSR performance, which is calculated as the sum of net CSR scores across six CSR categories. However, if a given CSR category contains certain issues that could affect CEO compensation, our findings on the effect of CSR on CEO compensation and its structure can be more or less pronounced. For example, firms with high CSR scores in employee relations may not exhibit a positive association between CSR and CEO total compensation, because they particularly attempt to improve employee relations by reducing CEO compensation. Motivated by this, we further explore the link between CSR performance and CEO compensation using the net CSR score on each category. We repeat the above analysis using net CSR scores on the six categories as independent variables as follows: COMP it = α + β 1 COMM it 1 + β 2 ENV it 1 + β 3 DIV it 1 + β 4 ER it 1 + β 5 HR it 1 + β 6 PROD it 1 + θ γ + ε it (5) where COMM it 1, ENV it 1, DIV it 1, ER it 1, HR it 1, and PROD it 1 are lagged net CSR scores on community, environment, diversity, employee relations, human rights, and product. The net CSR score is calculated as the difference between the strength and concern scores on a given CSR category. [Insert Table 6 about here] 18

21 Table 6 reports the estimation results of Equation (5). With the dependent variable of total compensation, we find that the net CSR score is significantly positively associated with CEO compensation in community, environment, and diversity. The coefficient on the net score in employee relations is insignificantly positive, but this does not provide ample evidence that firms focusing on improving employee relations tend to reduce CEO compensation. When we use the proportion of cash-based compensation as a dependent variable, the regression coefficient on the net CSR score is significantly negative in environment and human rights. Meanwhile, the net CSR score is significantly positively associated with the proportion of equity-based compensation in the categories of environment, diversity, and employee relations. The positive relation between the net CSR score in employee relations and the proportion of equity-based compensation demonstrates that firms paying attention to employee relations tend to have more incentive-based compensation structure for their CEOs. Overall, the results show that our findings on the effect of CSR on CEO compensation and its components are more pronounced in certain CSR categories. The positive association between CSR and the proportion of equity-based compensation is more pronounced in environment, diversity and employee relations, while the negative association between CSR and the proportion of cash-based compensation is more pronounced in environment and human rights The effect of CEO power on compensation structure We posit that the board of directors plays an independent and important role in designing the structure of CEO compensation and reducing agency problems. 8 However, if CEOs have power to be involved in the setting of their own compensation, they may exert their influence to design the compensation for their own interests, leading to an increase in the agency problems. If this is the case, our main results of the effect of CSR performance on CEO compensation may also depend on CEOs' involvement in the determination of their compensation components. Motivated by this, we examine the effect of CEO power on the relation between CSR performance and CEO compensation components. For this analysis, we use CEO duality as a 8 Many prior studies show that the compensation committee of the board of directors aims to design CEO compensation to mitigate the agency problem (Core and Guay 1999; Fama and Jensen 1983; Holmstrom 1979; Jensen and Meckling 1976; Murphy 1985). 19

22 proxy for CEO power on the determination of compensation. 9 The variable of CEO duality is a dummy variable indicating one if a firm's CEO is a chairman of the board of directors and zero otherwise. In Table 7, we add to the regression model in Table 4 an interaction term between CEO duality and CSR performance to test if our results are robust. [Insert Table 7 about here] Table 7 reports the estimation results. A coefficient on total CSR score is robustly positive when we add the interaction term between CEO duality and total CSR score (CEO_DUR TCSR) and use total compensation or the proportion of equity-based compensation as a dependent variable. Although a coefficient on the proportion of cash-based compensation is insignificantly negative, the overall results confirm that high levels of CEO compensation for socially responsible firms are mainly driven by a high proportion of equity-based compensation. When we run the regression of total compensation in Column (1), the coefficient on the interaction term (-0.003) is negative but insignificant. We also find that the coefficients on the interaction terms are not statistically significant when we use cash- and equity-based compensation as dependent variables in Columns (2) and (3). This suggests that our findings above are not affected by CEO duality. Overall, our results on the effect of CSR performance on CEO compensation structure are robust to the possible effect of CEO power on the setting of the compensation contract Sarbanes-Oxley Act of 2002 (SOX) An exogenous shock, such as the enactment of accounting and financial regulations, may affect the impact of CSR performance on CEO compensation. The implementation of SOX is one example of an exogenous shock. SOX requires firms to disclose more accurate information and meet the independence requirements of a firm's board of directors. This suggests that the passage of SOX paved the way for improving the transparency of accounting disclosures and the effectiveness of corporate governance system. Given this, the relations between CSR performance and the components of CEO compensation could be affected by this environmental 9 We have already controlled for this effect by adding CEO duality as one of the control variables in the main analysis. We use the same measure to analyze the effect of CEO power in more detail in this section. 20