Corporate social responsibility (CSR) initiatives and market reactions: The role of CEO risk preferences

Transcription

1 Corporate social responsibility (CSR) initiatives and market reactions: The role of CEO risk preferences By TRUNG NGUYEN 1 PhD. Candidate Finance Department Strome College of Business Old Dominion University Norfolk, VA tn1nguye@odu.edu ANH DANG PhD. Candidate Marketing Department Strome College of Business Old Dominion University Norfolk, VA adang@odu.edu 1 Corresponding author.

2 Corporate social responsibility (CSR) initiatives and market reactions: The role of CEO risk preferences ABSTRACT There has been no consensus in the literature regarding the effects of corporate social responsibility (CSR) on the firm s abnormal stock returns. Some studies posit that equity investors react negatively to CSR initiatives whereas other studies suggest the opposite. The present study aims to explain the inconclusive effects through an examination of CEO risk preferences reflected by CEO debt-based compensation. In particular, we investigated the cumulative abnormal returns of 905 event-year observations of 163 U.S. firms over the period. Our findings reveal that the effects of CSR initiative announcements on abnormal stock returns are more positive for risk-averse CEOs, compared to risk-seeking CEOs. The study also shows that bondholders do not have such a favorable reaction towards risk-averse CEOs CSR initiatives. In brief, the paper lends support to the importance of CEO pay structure in the CEO s commitment to CSR initiatives and market reactions to those initiatives. Keywords: CEO risk preferences, corporate social responsibility, inside debt, event study.

3 Corporate social responsibility (CSR) initiatives and market reactions: The role of CEO risk preferences INTRODUCTION Corporate social responsibility (CSR) refers to behaviors and practices that extend beyond immediate profit maximization goals and are intended to increase social benefits or mitigate social problems for constituencies external to the firm (Marquis et al. 2007, p. 925). It includes corporate philanthropy, cause-related marketing, minority support programs, socially responsible employment, to name a few (Sen & Bhattacharya 2001). CSR investments are becoming more and more prevalent. A global CEO survey reveals that more than 60% of CEOs considers CSR as a core business, rather than miscellaneous activities, in their firm (Horoszowski 2016). Hence, the effects of CSR investments and performance on security returns have received great research attention (e.g. Galema et al. 2008; Goss & Roberts 2011). According to stakeholder theory, the firm s performance depends not only on explicit contracts such as payments to bondholders but also on implicit contracts such as product quality and environment protection (Cornell & Shapiro 1987). When firms breach implicit contracts, stakeholders will convert those implicit contracts into explicit claims (e.g., stricter regulations from government agencies). These explicit claims are generally costlier for firm to satisfy than their implicit counterparts. Since CSR activities are designed as a tool to fulfil implicit contracts such as improving community or society, firms with stronger CSR reputation have lower costs for conducting those activities and accordingly enjoy a better financial performance (Cornell & Shapiro 1987). Consistent with this viewpoint, prior research has shown that corporate social 3

4 performance results in positive stock returns through enhanced productivity and analysts recommendations (Antunovich et al. 2000; Filbeck & Preece 2003). Meanwhile, opponents of stakeholder theory believe that CSR investments are a source of expenses, instead of profits. Thus, investing in CSR will divert the firm s ability from conducting other core business activities which yield much better earnings for the shareholders (Friedman 1970). In other words, CSR investments will hurt shareholders value. Accordingly, CSR investments are expected to result in negative stock returns due to disagreement from investors. In line with this view, previous studies did not find support for positive abnormal returns from CSR investments. For example, Chung et al. (1999) found that firms with better CSR did not outperform other firms in terms of stock returns. Further, Brammer et al. (2006) showed that CSR investments, especially those related to employment, were negatively correlated with abnormal stock returns. The inconsistency in the reported effects of CSR investments on abnormal returns calls for another explanation. According to Smith (2003), the effects of CSR are contingent on how CSR initiatives are executed, not on whether they are executed. More specifically, the strategies through which the firm conducts CSR activities can be influenced by the CEO of the firm. To this extent, agency theory has the potential in explaining why investors in general can react discretionarily to different CSR activities depending on what kind of CEOs implementing the strategies. According to Jensen and Meckling (1976), agency theory posits that there is a contract under which someone (the agent) works on behalf of the other (the principal) and has the decision-making authority delegated by the principal. Hence, the CEO of the firm, who is the agent, has the tendency to take advantage of the firm s assets for his own interests. For example, the CEO may decide to invest in CSR just to satisfy his or her personal needs for attention and 4

5 image reinforcement, regardless of whether such an investment is effective and beneficial for the firm (Petrenko et al. 2016). Such an agency problem from CSR investments can cause shareholders to react negatively to CSR investments (Krüger 2015). However, how can shareholders identify which CSR initiatives are purely for the long-term performance of the firm and which ones are due to the CEO s personal stakes? Given Petrenko et al. s (2016) suggestion, in this study, we propose that shareholders or investors in general rely on CEO s risk preferences to react to CSR initiatives. In other words, they will consider CEO risk preferences to justify whether a CSR initiative is trustworthy and beneficial to the long-term performance of the firm. Specifically, investors will support CSR initiatives by risk averse CEOs who are more likely to invest in those CSR activities deemed essential for the firm s long-term reputation and survival. Meanwhile, risk-taking CEOs care less about the future performance of the firm and thus do not hesitate to invest in those activities in order to gratify themselves immediately. Further, to control for agency problems, the board of director can design the CEO s compensation to alter the degree of CEO risk aversion (Jensen & Meckling 1976). Accordingly, shareholders or investors in general infer the CEO s risk preferences through his/her compensation structure. As pension and deferred compensation (i.e. inside debt) data became publicly available after SEC s 2006 disclosure requirement, research scholars have used such information to better estimate the degree of CEO risk preferences (Sundaram & Yermack 2007; Edmans & Liu 2011; Wei & Yermack 2011). Because of the unfunded and unsecured nature of pension and deferred compensation, a CEO whose compensation structure tilts towards such instruments is exposed to default risk similar to that faced by creditors. Therefore, he or she will display a higher level of risk aversion. In this case, the CEO will be more risk-averse and more careful in selecting CSR 5

6 initiatives to ensure long-term performance of the firm. As a result, equity market reactions to the CSR initiatives announced by risk-averse CEOs will be more positive than they are to those made by risk-taking CEOs. Accordingly, we argue that shareholders are not against CSR initiatives; they simply do not advocate those initiatives from risk-taking CEOs since those initiatives will be harmful to their return on investments. In addition, we predict that bondholders also support CSR initiatives from risk-averse CEOs. Prior research has also shown that bondholders, along with other capital providers, are more likely to set lower capital cost to firms with lower risks. Further, Buysse and Verbeke (2003) suggest that CSR activities help firms avoid extremely negative events and thus reduce the probability of financial fallout. Also, those activities from risk-averse CEOs are even less risky since they are less subject to agency costs, i.e., the CEOs are less likely to engage in CSR for the sake of their personal benefits (Jensen & Meckling 1976). Thus, bondholders or creditors in general will support those CSR initiatives from risk-averse CEOs. The present paper is one of the first to use event study analysis to examine the potentially positive impact of CEO risk preferences on shareholders and bondholders reactions to CSR initiatives, the promises to spend a certain amount of money or effort for CSR activities. More specifically, we empirically investigate whether CEO risk preferences statistically affect the cumulative abnormal stock and bond returns around the CSR initiatives announcements made by U.S based public companies. We get the announcement dates from the news releases of the CSRwire website 2. In 2016, CSRwire has more than 70,000 readers per month and around 60,000 news alert subscribers (CSRwire 2016b). CSRwire streams news through leading 2 More information can be found at 6

7 websites such as Google, Reuters, Yahoo! News, and reference sources such as LexisNexis and Bloomberg. Accordingly, CSRwire is a leading information source of CSR news and has shown its credibility through a few selected academic papers (Sethi 2005; Griffin & Sun 2013; Naiker 2013 among others). Since our focus is CSR initiatives, we include only news releases about firms initiating a CSR campaign or investment, and we exclude news about firms CSR achievements such as awards or certification. The news sample comprises of 1,840 CSR initiatives by U.S. public firms within an eight-year period from July 2007 to December In order to ensure that our results are not tempered by other confounding events, we manually check the event dates in the Factiva database. After merging the refined CSR data with other data sources, we obtain the final sample of 905 event-year observations. For each of the event dates in the final sample, we calculate the cumulative abnormal stock returns within 2-day, 3-day, and 5-day windows around the announcement. We use the three- and four-factor models to estimate the abnormal stock returns (Fama & French 1996; Carhart 1997). Furthermore, we compute the cumulative abnormal bond returns using the twofactor bond model suggested by Wei and Yermack (2011). Following previous studies, we use CEO relative leverage and CEO relative incentive to capture the degree of CEO risk preferences (Sundaram & Yermack 2007; Edmans & Liu 2011; Wei & Yermack 2011). Specifically, CEO relative leverage equals CEO inside debt divided by CEO inside equity and then scaled by the firm leverage, and CEO relative incentive is the ratio of the change in CEO inside debt to the change in CEO equity scaled by the change in the firm leverage. In addition, our multivariate analyses also control for other CEO, firm and CSR characteristics. The findings of our study reveal some important insights. First, our study shows that the abnormal stock returns on CSR initiatives of risk-averse CEOs are significantly more positive, 7

8 compared to those from risk-taking CEOs. Specifically, the abnormal stock returns of riskseeking CEOs CSR initiatives are negative whereas those of risk-averse CEOs are positive. Such a difference in abnormal stock returns from two types of CEO risk preferences is robust through different time windows (2 days, 3 days, 5 days) in both the three-factor model and fourfactor model. Shareholders responses do not vary across three types of CSR initiatives that are sustainability and environment, corporate philanthropy, and socially responsible investing. Moreover, whether firms publicize their monetary investments or not does not change shareholders reactions. Surprisingly, our findings indicate that bond returns on CSR initiatives are generally negative. Our further examination indicates that bondholders react more negatively to those CSR initiatives from risk-averse CEOs than those from risk-seeking CEOs. The present paper contributes to the literature in several important ways. First, our findings lend empirical support to the irreplaceable role of CEO risk preferences, or more specifically CEO inside debt, in the context of corporate social responsibility. Furthermore, through a unique dataset, the paper helps explain the literature inconsistency regarding shareholders reactions to CSR initiatives. Since CSR initiatives can produce two opposing effects obviously short-term expenses and potentially long-term gains, it is not an easy answer for shareholders when facing CSR initiative announcements. Yet, those CSR initiatives from risk-averse CEOs can make the latter effect more certain. Thus, shareholders react positively to those initiatives. By emphasizing the role of CEO risk preferences, our study sheds new light on future research with respect to market reactions to corporate social responsibility. Particularly, shareholders do not simply look at CSR initiatives per se during their decision making process. Instead, shareholders consider the potential of those initiatives reflected by CSR risk- 8

9 preferences. Therefore, future research should involve thorough examinations of shareholders decisions when there is CSR news specifically or any other news of the firm in general. Second, CEO characteristics have recently been advocated as determinants of CSR initiatives. Yet, prior research has just touched CEO static personalities/characteristics which are beyond the control of the board of directors. For instance, Petrenko et al. (2016) suggest that CEOs with greater needs for attention and image reinforcement tend to engage in more CSR activities. In addition, Waldman et al. (2006) find that CEO leadership styles such as transformational leadership will drive the firm s CSR investments. In this study, by examining the role of CEO risk preferences determined by compensation structure, we provide practical implications for the board of directors in developing compensation strategy for the CEO in a way that encourages the CEO to invest in efficient rather than self-serving CSR initiatives. Third, our empirical work mitigates the potential endogeneity problem inherent in prior studies. Instead of looking at the correlation between the firm s CSR performance ratings and investors reactions (e.g. Goss & Roberts 2011; Minor & Morgan 2011), our study is among the few recent papers that examine the market reactions right after firms announce their CSR news (e.g. Krüger 2015). According to Krüger (2015), short-term event study can significantly reduce the issue of reverse causality in CSR research. Furthermore, we look at CSR initiative announcements that are early in the CSR investment process and are uncertain about the future. Meanwhile, news about firms CSR achievements or negative events may already embrace information about the market reactions. Therefore, by investigating stakeholder reactions to the firm s CSR activities at the debut stage, the present paper indicates that managers have to keep track of their CSR outcomes and at the same time pay attention to what they propose or promise to investors earlier in their CSR initiatives. 9

10 Finally, whereas prior studies tend to look at the reactions of either shareholders or bondholders (Krüger 2015), our empirical work examines both of them. Additionally, there has been limited research on bondholders reactions to CSR initiatives (Goss & Roberts 2011). In our study, CSR initiatives, especially from risk-averse CEOs, receive support from shareholders, yet disproval from bondholders. Such a disapproval is, however, consistent with several studies in the extant literature. In particular, Menz (2010) found that there was a positive relationship between CSR performance and credit spreads. Additionally, Goss and Roberts (2011) show that for most firms, creditors do not consider CSR activities as risk-reducing. Further, risk-averse CEOs may spend more on CSR campaigns and may be strayed away from other potential activities. Such an investment strategy may cause difficulty in paying debt back to bondholders later. Future research therefore should go deeper into bondholders decision-making process when encountering CSR investment news. The rest of the paper is organized as follows. Section 2 and section 3 describes the methodology of the paper and the sampling process. Section 4 and section 5 provide empirical findings, robustness check, and discussion. The paper concludes with section 6. RESEARCH DESIGN The present research uses event-study methodology to investigate the short-term market reactions to CSR initiatives made by U.S. public firms around the announcement date. OLS regressions with clustered standard errors at the firm level are conducted to test the effect of CEO risk preferences on firms cumulative abnormal returns. The baseline regression model of our study is as follows: CAR i = β 0 + β 1 X i + k j=1 γ j Z i,j + ε i (1) 10

11 The left-hand-side of the model, i.e. the cumulative abnormal return, is calculated as the sum of several daily abnormal returns for a specified time window around the announcement date. To ensure robustness of our model, we use three time windows including two days, three days, and five days. We also estimate the daily abnormal stock return using the three-factor and four-factor models (Fama & French 1996; Carhart 1997). The corresponding inputs into these two models are the market risk premium, the SMB, the HML, and the momentum factors respectively. We calculate the coefficient estimates of each factor, using the estimation period from t = -320 to t = -20 relative to the CSR announcement date of a particular firm. With respect to the bond market s reaction, we use the two-factor model to estimate the cumulative abnormal bond return (Wei & Yermack 2011). Following previous studies, we employ the representative bond approach to account for the possibility that a certain firm may have multiple outstanding bond issues at a specific time (Dhillon & Johnson 1994; Bessembinder et al. 2009; Wei & Yermack 2011). Accordingly, we employ the investment-grade and the speculative-grade corporate bond indices to predict a particular firm s bond return. In this process, the estimation period is from t = -150 to t = -20. In both situations, changing the estimation period does not affect the overall results of our study. The daily abnormal stock/bond return is simply the difference between the actual return and the return predicted by the corresponding model. With respect to the right-hand-side of the baseline model, Xi,t represents the level of CEO risk preferences. For robustness, we use two different proxies for CEO risk preferences CEO relative leverage and CEO relative incentive (Sundaram & Yermack 2007; Edmans & Liu 2011; Wei & Yermack 2011). Specifically, CEO relative leverage equals CEO inside debt divided by CEO inside equity and then scaled by the firm leverage while CEO relative incentive is the ratio of the change in CEO inside debt to the change in CEO equity scaled by the change in the firm 11

12 leverage. Besides, Zj refers to the set of control variables including CEO, firm and CSR characteristics. Regarding CEO characteristics, we control for CEO age, CEO cash compensation, CEO tenure, as well as CEO vega-to-delta ratio (Core & Guay 1999; Cassell et al. 2012). For firm characteristics, we incorporate firm size, firm sale growth, firm market-to-book ratio, firm financial leverage, firm capital expenditures (CAPX), and firm research & development expenses (R&D) into the model specifications 3 (Coles et al. 2006). In the robustness section, we also add the following four CSR characteristics as dummies: CSR financial commitment, CSR environmental concern, CSR corporate philanthropy, and CSR socially responsible investment. Continuous variables are winsorized at the 1% and 99% levels to control for the potential outliers (Cassell et al. 2012). A detailed description of the variables is provided in Appendix A. Furthermore, all right-hand-side variables are lagged by one period relative to the CSR event dates. According to equation (1), the emphasis of our study is the effect of CEO risk preferences on cumulative abnormal returns. In other words, a significantly positive value of β1 will indicate that investors in general will react more positively to CSR announcements made by risk-averse CEOs. Meanwhile, if the coefficient is insignificant, inside debt does not differentiate investors reactions to CSR investment announcements. DATA COLLECTION We utilize different secondary sources to derive data for our sample. For CSR initiative announcements, we collect news releases from CSRwire ( - a leader in posting news about CSR (Griffin & Sun 2013). CSR newswire provides a huge sample of CSR investment disclosures for a wide range of industries and time available. In particular, the 3 We follow previous studies to set missing R&D expenses equal to zero. 12

13 website releases around 250 page views per month (Griffin & Sun 2013). Further, we decide to use CSRwire instead of KLD ratings because the markets immediately utilize new information and thus investors reactions to CSR investment disclosures from CSRwire within a few days are more relevant than to the overall year-based KLD ratings. As mentioned in Krüger (2015), KLD newsletters may suffer from the selective-reporting bias as KLD may intentionally focus on certain news that draws the public attention. When screening news releases, we include only those events related to the firm s CSR initiatives, i.e., commitments to spending a certain amount of money or effort on CSR. Such invested money or effort triggers perceived short-term risks through increased money outflows or decreased earnings and thus arouse market reactions. However, not all news releases about CSR investments reveal a specific dollar amount to be invested. For example, in 2013, 1,200 employees of Newell Rubbermaid Inc. spent hours building homes for low-income families. Even though their effort is not explicitly monetized, these employees could have generated more products and thus profits for the firm if those hours had been used for production rather than CSR activities. Therefore, such news is included in our sample. On the other hand, news such as awards and certification is about the firm s achievement but not about future money outflows or expenses. As a consequence, this news category is excluded from our analysis. To emphasize the magnitude of the dollar amount, in addition, we have the CSR financial commitment take a value of one if the disclosed value is greater than or equal $200,000. Using different thresholds such as $500,000 or $1,000,000 still provides consistent results (not tabulated) 4. 4 These additional findings are available upon request. 13

14 Further, we classify CSR initiatives into three categories as mentioned in the previous section: environmental concerns, corporate philanthropy, and socially responsible investing. Environmental concerns refers to those activities that are aimed at protecting an organism or an ecological community. It also includes those activities that help to satisfy firms current needs without compromising the resources needed for their future needs. An example of this category is Walmart s announced investments in renewable wind power in Corporate philanthropy activities are those donations to charities, communities, and nonprofit organizations. As an illustration, Holland American Line donated $25,000 to Seattle Children s Hospital in Socially responsible investing initiatives are those that integrate personal values and societal concerns with investment decisions (CSRwire 2016a). For example, in 2011, Entergy Corporation committed $1 million assistance to help elderly and disabled customers deal with high utility bills. It should be noted that there are some other CSR categories identified in KLD ratings which have been used by prior research. Examples are corporate governance, diversity, and human rights (Krüger 2015). However, these relate to policy adjustments and generally do not involve those efforts and monetary investments that may hurt the expenses of the firm. We therefore exclude those initiatives. Our taxonomy is in line with the main categories identified as CSR drivers of market returns in the literature. Each of these three types is distinctly different from one another and thus has been studied independently in prior research. Griffin and Sun (2013) focus exclusively on green investments when investigating stockholder reactions to CSR news. Similarly, Godfrey (2005) examines the relationship between corporate philanthropy and shareholder wealth. To capture the potential effects of CSR characteristics, we include these three categories as dummies in the robustness check. 14

15 Because executive pensions and deferred compensations are explicitly available for only U.S.-based firms since 2006, our sample is restricted to U.S. corporations only. As a result, we obtain 1,840 news releases about CSR investment disclosures between July 2007 and December We then merge the sample with the other data sources to extract firm characteristics, CEO characteristics, and security price information. Specifically, we collect data about CEO compensation structure and CEO characteristics such as CEO age from the Execucomp database. Firm characteristics are extracted from the Compustat database. We derive daily stock prices from the Center for Research in Securities Prices (CRSP) database and daily bond prices from the TRACE database. Our final sample has 905 CSR events spanning from 2007 to EMPIRICAL FINDINGS Descriptive statistics including means, standard deviation, 25 th, 50 th, and 75 th percentiles for CEO, firm, and CSR characteristics are reported in Table 1. The mean and the standard deviation of CEO relative leverage are.79 and.72 respectively while those of CEO relative incentive are 5.94 and 2.58 respectively. The average CEO in the sample has approximately years of age and 5.44 years of tenure. In addition, the average CEO cash compensation (i.e. salary plus bonus) is around 11.88% of his/her firm s total assets. With respect to firm characteristics, the average firm size - the natural logarithm of one plus book value of assets - is The mean of the market-to-book ratio is 1.89, indicating that the average firm in the sample is overvalued by the market. On average, a firm spends roughly 1.73% and 4.74% of its total assets on R&D expenses and capital expenditures, respectively. Regarding firm sales, the average firm grows approximately by 6.00% per year. Although our sample size is smaller than 15

16 that of Krüger (2015) due to the elimination of missing CEO compensation data, the two samples share comparable statistics for firm variables Insert Table 1 about here In terms of CSR initiative events, the mean of financial commitment is.34, which shows that almost 66.00% of the news releases do not publicize or have the amount of invested money greater than or equal to $200,000. Among the three CSR categories, CSR corporate philanthropy is more popular than the other two (i.e. CSR environmental concerns and CSR socially responsible investment) as shown by their respective percentages, 68.40% versus 21.66% and 9.94%. Table 2 shows the pairwise correlations among the variables in the sample. According to the table, CEO relative leverage and CEO relative incentive are positively correlated at the 1% level. More specifically, the correlation between CEO relative leverage and relative incentive is.80, which is expected since they are alternative proxies for CEO risk preferences. Regarding the other correlation coefficients, despite being significant in a few cases, the statistics are far below.70, thus these correlations do not raise concern about multicollinearity. For instance, a manager working for a larger firm tends to have higher CEO risk preferences as the pairwise correlations between firm size and the two proxies of CEO risk preferences are.35 and.45, respectively. Both coefficients are statistically significant at the 1% level. 5 Krüger s (2015) CSR sample has around 1,400 events spanning from 2001 to

17 Insert Table 2 about here Univariate results of abnormal stock and bond returns of CSR initiatives are reported in Table 3. The second column of the table show the means of the cumulative abnormal returns for the full sample. On average, shareholders react positively whereas bondholders react negatively to CSR initiative news releases. However, the market reactions are statistically insignificant except for the five-day window cumulative abnormal bond returns. For the five-day window, CSR initiative events are associated with a reduction of.09% in bond prices and it is significant at the 5% level. With respect to cumulative abnormal bond returns, our results are consistent with Goss and Roberts (2011) as bondholders do not consider CSR activities risk-reducing investments, and thus react negatively to such events on average. Further examinations on the means of cumulative abnormal returns of the five subsamples based on CEO risk preferences reveal additional insights. On the one hand, the cumulative abnormal stock returns are consistently more positive for those CSR initiatives from the higher quintiles of CEO relative leverage in all of the two-day, three-day, and five-day windows. Specifically, mean comparison in the three-factor model shows that the cumulative abnormal stock returns of the fifth quintile (i.e. risk-averse CEOs) are significantly larger than those of the first quintile (i.e. risk-seeking CEOs) at the 5% and the 10% levels within the twoday and five-day windows, respectively. Similarly, in the four-factor model, the fifth quintile of CEO relative leverage has significantly more positive cumulative abnormal stock returns within both the two-day and five-day windows than the first quintile. Despite being insignificant, the 17

18 difference-in-mean statistics for the three-day windows are consistently positive under both models. Overall, the univariate results strongly lend support to our central prediction that equity investors prefer CSR initiatives from risk-averse CEOs (with a higher level of inside debt) to those from risk-seeking CEOs (with a lower level of inside debt) Insert Table 3 about here Regarding bondholders reactions, on the other hand, our two-factor model indicates that the returns from the higher quintile of CEO relative leverage tend to be smaller than they are from the smaller quintile and the difference is significant at the 5% level in the two-day window. Out of the three cases, in particular, the difference in average bond returns between risk-averse and risk-seeking CEOs are negative for the two-day and three-day estimates. Furthermore, it is even statistically significant at the 5% level in the case of the two-day abnormal bond returns. Though it is against our main prediction with respect to bondholders response, this piece of evidence is consistent with previous studies (Menz 2010; Goss & Roberts 2011). To this extent, bondholders may perceive that risk-averse CEOs, due to their debt-based compensation structures, are induced to concentrate more on CSR campaigns and deviate from the firm s primary objective, i.e. maximizing firm value. From the bondholders perspective, hence, firms pursuing such a peripheral investment strategy may have difficulty repaying their corporate debts in the long run. Finally, replacing CEO relative leverage with CEO relative incentive yields consistent results 6. 6 These results are available upon request. 18

19 Insert Table 4 about here In this section, we then conduct multivariate regression analyses using both three-factor and four-factor models to test the impact of CEO compensation structure on market reactions to CSR initiatives. In any reported table, we capture CEO risk preferences using both CEO relative leverage and CEO relative incentive. Similar to our univariate analysis, we use three different time windows (i.e., two days, three days, and five days) to ensure the robustness of our results. In table 4, cumulative abnormal stock returns estimated from the three-factor model are regressed on CEO-level, firm-level, and CSR-level characteristics. The standard errors are clustered at the firm level. In the first three specifications, we regress cumulative abnormal stock return on CEO relative leverage, along with control variables such as CEO age and tenure. The effect of CEO relative leverage is significantly positive at the 1% level in the two-day window and at the 5% level for the other two windows. The last three specifications use CEO relative incentive instead of CEO relative leverage. Consistently, the effect is significantly positive across the three time windows. Such a finding conforms to our main prediction that shareholders prefer CSR events announced by risk-averse CEOs. Regarding the coefficients of other control variables, most of them are not statistically significant. For instance, CEO vega-to-delta ratio has a positive effect in most of the models, yet it is statistically insignificant. With respect to firm characteristics, our findings show that firms with bigger size, higher level of sale growth, and higher level of R&D expenses are associated with lower cumulative abnormal stock returns around the CSR initiative event dates. Such results are statistically significant at either the 5% or the 1% level. In terms of R&D expenses, the negative coefficient reflects that abnormal stock returns on CSR initiatives 19

20 tend to be lower for firms with higher R&D expenses. This can be explained by the fact that given limited resources, investors weigh R&D activities more important than CSR investments which are primarily used to avoid negative externalities Insert Table 5 about here In table 5, we perform the same procedures as those in the previous table, but we use the four-factor model s cumulative abnormal stock returns as our dependent variable. The empirical findings are similar to those in the three-factor model and thus confirm our prediction. Specifically, the effects of both CEO relative leverage and CEO relative incentive are significantly positive in the two-day window (p <.01). For the other two time windows, the coefficient estimate is statistically significant at the 5% level for CEO relative leverage and at the 10% level for CEO relative incentive. The other control variables have the same signs as those in the three-factor model regressions Insert Table 6 about here Our multivariate regressions of two-factor model cumulative abnormal bond returns on CEO risk preferences are documented in table 6. Similar to the previous tables, Model I, II, and III regress bond returns on CEO relative leverage and model IV, V, and VI use CEO relative incentive as the main independent variable. Consistent with our univariate results, the coefficient 20

21 estimates of CEO risk preferences are negatively related to cumulative abnormal bond returns across the six specifications. In particular, such a negative effect of CEO relative leverage is statistically significant in the first two windows (p <.05) and that of CEO relative incentive is significant in the two-day window (p <.10). Furthermore, most of the control variables are insignificant except for firm capital expenditures. The coefficient estimates of firm capital expenditures are positive across the six models and significant in three models at least at the 10% level, which suggests that firms with higher capital expenditures are perceived more positively by bondholders when they announce their CSR activities Insert Table 7 about here To enhance the robustness of our results, we continue to examine whether our aforementioned findings are affected by CSR characteristics or not. In this part, we report the same multivariate results but controlling for CSR financial commitment and the other three CSR categories environmental concerns, corporate philanthropy, and socially responsible investing, respectively. For brevity, we only report the results for the four-factor model stock returns, and the two-factor model bond returns within three days around the announcement dates. Table 7 shows the multivariate results with the addition of CSR financial commitment. In the first two models, the dependent variable is the cumulative abnormal stock returns derived from the fourfactor model. In the last two models, the dependent variable is the cumulative abnormal bond returns derived from the two-factor model. According to this table, the main effects of both CEO relative leverage and CEO relative incentive are still significant even with the inclusion of CSR 21

22 financial commitment. Therefore, our main predictions about the market reactions of CSR events and CEO risk preferences are not influenced by the events financial commitment characteristics. From a theoretical standpoint, in addition, disclosing the dollar value of the CSR activities can intrigue the negative (positive) reaction from equity (bond) investors. Thus, we should expect a negative (positive) and significant coefficient for CSR financial commitment, respectively. Nevertheless, the coefficient of CSR financial commitment is not significant, either economically or statistically, especially from the bondholders perspective. The effects of other control variables remain consistent with the previous analyses Insert Table 8 about here Next, we repeat the same analysis, yet with the three CSR categories instead. Table 8 shows the multivariate results with the addition of CSR environmental concern. Similarly, the main effects of both CEO relative leverage and CEO relative incentive on the cumulative abnormal returns are not tempered by the addition of CSR environmental concern. Although it is not statistically significant, in addition, both equity and bond investors react positively to CSR news releases about environmental concern, holding other things constant. Again, the effects of other control variables remain consistent with the previous analyses Insert Table 9 about here

23 In table 9, we report the multivariate results of regressing the cumulative abnormal returns on CEO risk preferences controlling for CSR corporate philanthropy. The documented evidence is still consistent with our main conclusion. In contrast to the consistent effects of CSR environmental concerns on the market reactions, the presence of CSR corporate philanthropy dummy yields different effects for equity and bond investors. Specifically, CSR events associated with corporate philanthropy are greeted with open arms by equity investors. However, the opposite result is observed for bondholders. In both cases, the coefficient estimates are not statistically significant. Again, the effects of other control variables remain consistent with the previous analyses Insert Table 10 about here Table 10 shows the multivariate results of regressing the cumulative abnormal returns on CEO risk preferences controlling for CSR socially responsible investment. Similar to the previous tables, this specific CSR category does not influence the main conclusion of our study, i.e. CEO risk preferences matter in CSR activities. In contrast to the effect of the other two CSR categories, CSR events associated with corporate philanthropy are perceived negatively by equity investors, whereas the opposite result is documented for bondholders. In both cases, the coefficient estimates are not statistically significant. Again, the effects of other control variables remain consistent with the previous analyses. In brief, the last four tests ensure the robustness of our main empirical results. Indeed, such findings about the effects of different CSR 23

24 classifications on the market reaction reported above, to a certain extent, are consistent with Krüger (2015). DISCUSSION Our empirical analysis reveals some interesting findings. First of all, the current study shows that shareholders react positively to the CSR initiatives by risk-averse CEOs, yet negatively to those by risk-seeking CEOs. This is consistent with Smith s (2003) emphasis on how, rather than whether, CSR investments are made. CEOs with inside debt are expected to invest in CSR initiatives that promise to build up the long-term image of the firm. CEOs without inside debt, whose compensation is not dependent much on the firm s long-run future, will simply invest in CSR just for the sake of their personal recognition (Petrenko et al. 2016). Therefore, equity investors favor those CSR initiatives of risk-averse CEOs who have higher inside debt. Bondholders, however, do not offer such positive support towards risk-averse CEOs CSR activities. Their reactions to CSR investments per se are not favorable as they do not consider these initiatives value enhancing or risk reducing. To this extent, bondholders may perceive that risk-averse CEOs, due to their debt-based compensation structures, are induced to concentrate more on CSR campaigns and deviate from the firm s primary objective, i.e. maximizing firm value. From the bondholders perspective, hence, firms pursuing such an investment strategy may have difficulty repaying their corporate debts in the long run. Nevertheless, the present study has several limitations. First, our sample only covers 163 public companies in the U.S. due to the limited information about pensions and deferred compensations. Hence, the essential role of CEO risk preferences in the context of CSR investment may lack some generalizability. Future research in this area can also investigate the difference in the market reactions to CSR events between firms with inside-debt disclosures and 24

25 their counterparts. Second, our empirical results for bondholders reactions may be affected by our bond return estimation method. According to Bessembinder et al. (2009), the representative bond approach is less robust than the firm level approach (i.e. portfolio approach) in capturing the abnormal bond returns when handling firms with multiple corporate bond issues. However, the latter requires additional inputs such as maturity and credit ratings for each bond issue, which are of limited access. This concern may potentially explain our counterintuitive findings of bondholders reactions to CSR announcements made by risk-averse CEOs. Third, since the debtbased compensation data are available for U.S. firms only, our sample is primarily restricted to U.S. firms. In the future, our results can be more generalizable if firms in other countries can be examined. Despite the aforementioned caveats, our research provides both theoretical and practical implications. With respect to CSR and finance literature, our study explains the controversy regarding the market reactions to CSR investments using CEO risk preferences. Previous studies tend to simplify the effects of CSR investments on market reactions when assuming that investors automatically have negative responses to CSR investments (Brammer et al. 2006). In fact, equity investors value those CSR activities that are strategically planned; thus, they will support those activities of CEOs with inside debt who are risk averse and thus more committed to CSR investments. Further, prior research primarily examines CEO s characteristics that are beyond the control of the investors. In this study, by examining CEO compensation structure, we suggest a governance mechanism through which investors can monitor CSR activities effectively. Regarding practical implications, our study also suggests that the board of directors should consider debt based compensation for their top executives to ensure the quality of the firm s CSR or other long-term investments. In terms of conducting CSR research, finally, our 25

26 empirical work offers some validity check for CSRwire news releases from an academic point of view. Future studies can consider CSRwire a reliable alternative to the KLD database in the context of short-term event studies. CONCLUSION Our research has yielded additional and valuable insights into the CSR and corporate finance literature. Through a sample of 905 event-year observations during the period, our study has shown that equity investors react more favorably to those CSR initiatives of CEOs with inside debt than they do to those CSR initiatives of CEOs without inside debt. Therefore, it gives rise to the importance of CEO pay compensation in reducing potential agency costs in CSR investments and thus enhancing CSR effectiveness. In other words, debt-based compensation arouses CEO long-term commitment to the firm. As they become risk-averse, these CEOs will engage in careful planning for CSR investments. We also found that bondholders reactions are not as positive as those of shareholders. In conclusion, our empirical work suggests that future research should look at CEO risk preferences and their underlying compensation structure when examining market returns on CSR initiatives. 26

27 REFERENCES Antunovich, P., Laster, D., Mitnick, S., Are high-quality firms also high-quality investments? Current Issues in Economics and Finance 6 Bessembinder, H., Kahle, K.M., Maxwell, W.F., Xu, D., Measuring abnormal bond performance. Review of Financial Studies 22, Brammer, S., Brooks, C., Pavelin, S., Corporate social performance and stock returns: UK evidence from disaggregate measures. Financial Management 35, Buysse, K., Verbeke, A., Proactive environmental strategies: A stakeholder management perspective. Strategic Management Journal 24, Carhart, M.M., On persistence in mutual fund performance. The Journal of Finance 52, Cassell, C.A., Huang, S.X., Sanchez, J.M., Stuart, M.D., Seeking safety: The relation between CEO inside debt holdings and the riskiness of firm investment and financial policies. Journal of Financial Economics 103, Chung, S.Y., Schneeweis, T., Eneroth, K., Corporate reputation and investment performance: the UK and US experience. Available at SSRN Coles, J.L., Daniel, N.D., Naveen, L., Managerial incentives and risk-taking. Journal of Financial Economics 79, Core, J., Guay, W., The use of equity grants to manage optimal equity incentive levels. Journal of Accounting and Economics 28, Cornell, B., Shapiro, A.C., Corporate stakeholders and corporate finance. Financial Management,

28 CSRwire, 2016a. Corporate social responsibilty categories by CSRwire. Retrieved on May 17, 2016 at CSRwire, 2016b. CSRwire coverage. Retrieved on May 17, 2016 at Dhillon, U.S., Johnson, H., The effect of dividend changes on stock and bond prices. The Journal of Finance 49, Edmans, A., Liu, Q., Inside debt. Review of Finance 15, Fama, E.F., French, K.R., Multifactor explanations of asset pricing anomalies. The Journal of Finance 51, Filbeck, G., Preece, D., Fortune s best 100 companies to work for in America: Do they work for shareholders? Journal of Business Finance & Accounting 30, Friedman, M., The social responsibility of business is to increase its profits. New York, Galema, R., Plantinga, A., Scholtens, B., The stocks at stake: Return and risk in socially responsible investment. Journal of Banking & Finance 32, Godfrey, P.C., The relationship between corporate philanthropy and shareholder wealth: A risk management perspective. Academy of Management Review 30, Goss, A., Roberts, G.S., The impact of corporate social responsibility on the cost of bank loans. Journal of Banking & Finance 35, Griffin, P.A., Sun, Y., Strange bedfellows? Voluntary corporate social responsibility disclosure and politics. Accounting & Finance 53, Horoszowski, M., % of CEOs are increasing investment in corporate social responsibility in Here's why. Retrieved on May 17, 2016 at 28

29 Jensen, M.C., Meckling, W.H., Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics 3, Krüger, P., Corporate goodness and shareholder wealth. Journal of Financial Economics 115, Marquis, C., Glynn, M.A., Davis, G.F., Community isomorphism and corporate social action. Academy of Management Review 32, Menz, K.-M., Corporate social responsibility: Is it rewarded by the corporate bond market? A critical note. Journal of Business Ethics 96, Minor, D., Morgan, J., CSR as reputation insurance: Primum non nocere. California Management Review 53, Naiker, V., Discussion of Strange bedfellows? Voluntary CSR disclosure and politics. Accounting & Finance 53, Petrenko, O.V., Aime, F., Ridge, J., Hill, A., Corporate social responsibility or CEO narcissism? CSR motivations and organizational performance. Strategic Management Journal 37, Sen, S., Bhattacharya, C.B., Does doing good always lead to doing better? Consumer reactions to corporate social responsibility. Journal of Marketing Research 38, Sethi, S.P., Investing in socially responsible companies is a must for public pension funds because there is no better alternative. Journal of Business Ethics 56, Smith, N.C., Corporate social responsibility: whether or how? California Management Review 45,