CEO compensation and Corporate Social Responsibility
A distinction between overconfident CEOs and rational CEOs.By Dinja Becker1
University of Groningen Faculty of Economics and Business
MSc Finance
Supervisor: Prof. dr. L.J.R. Scholtens Date: 4 June 2019
Abstract
This study analyzes the influence of the compensation structure of CEOs on corporate social responsibility (CSR) by using a sample of US firms during the period 2002-2018. The study compares overconfident CEOs to rational CEOs to examine the impact of the compensation structure of these two groups on CSR. As Malmendier and Tate (2005) point out, an overconfident CEO overestimates his knowledge and abilities, and tends to overinvest in projects that are too risky. A rational CEO makes decisions that offer the highest satisfaction and does not invest in projects that are too risky, as confirmed in the study by Goel and Thakor (2008). I find that compensating overconfident CEOs by short-term incentive payments, such as bonus and salary, is positively related to CSR. Moreover, the salary of rational CEOs has a significantly positive influence on CSR. I conclude that the compensation structure of CEOs affects the firm’s social responsible activities.
Keywords: Overconfidence, CEOs, CSR, Compensation Word count: 11298
1 E-mail: d.a.becker@student.rug.nl, student number: S2574098, course: Master’s Thesis Finance, course code:
1. Introduction
Growing attention to behavioral finance literature leads to more knowledge about the personal characteristics of CEOs. In corporate finance, it is often assumed that individuals behave rationally; that rational individuals make decisions that offer them the highest satisfaction (see, e.g., Becker, 1962; Moscati & Tubora, 2011). However, behavioral factors – such as overconfidence and cognitive limitations - could lead CEOs to act in a less than rational way. This study distinguishes between two types of CEOs. The first type is the overconfident CEO who overestimates his knowledge and abilities, and tends to overinvest in projects that are too risky, as mentioned by Malmendier and Tate (2005). The other type is the rational, risk-averse CEO who makes decisions that offer the highest satisfaction and does not invest in projects that are too risky, as confirmed in the study by Goel and Thakor (2008).
CEO characteristics influence a firm’s social responsible activities (see, e.g., Chen, Mack & Tang, 2018; Manner, 2010). Bénabou and Tirole (2010) define corporate social responsibility (CSR) as giving up some profit in social interest in which a firm must voluntarily go beyond its legal and contractual obligations. Empirical research demonstrates that firms with overconfident CEOs display less corporate social responsibility (see, e.g., Chen, Qian, Shen & Tang, 2015; McCarthy, Oliver & Song, 2016). In addition, overconfident CEOs tend to be less dependent on the support of their stakeholders; they are less driven to serve the demand of stakeholders, with the result that they engage in fewer socially responsible activities, as demonstrated in the study by Chen et al., (2015). Other studies (see, e.g., Bouslah, Linares-Zegarra, M’Zali & Scholtens, 2018; Jian & Lee, 2015; Mahoney & Thorn, 2006) demonstrate that the compensation structure of CEOs influences a firm’s socially (ir)responsible activities. A CEO could be compensated by short-term incentive payments, such as salary and bonus, and by long-term incentive payments, such as stock options.
Social esteem increases the career opportunities of CEOs and consequently their bargaining power, which could lead to a higher salary, as indicated by Milbourn (2003). Hence, the agency theory predicts a positive relationship between CEO compensation and CSR.
Multiple studies focus on the relationship between CEO compensation and CSR. To my knowledge, none attempts to make a distinction between overconfident CEOs and rational CEOs. The different behavioral characteristics of the CEOs could cause a different relationship between CEO compensation and CSR for these two groups. Therefore, the question arises whether the components of CEO compensation influence the socially responsible activities of firms with overconfident CEOs and rational CEOs. To address this question, I use data from 4,151 CEOs of companies listed on the S&P500 during the period January 2002 to January 2018.
Based on stock options exercised by CEOs, I make use of the Holder 67 measure to indicate the overconfidence level of a CEO. This method is based on the work of Malmendier and Tate (2005). They suggest that a CEO can be classified as overconfident if his stock options are more than 67% in the money at least twice during the sample period. An option is in the money if the exercise of the option results in a positive pay-off (see, e.g., Coval & Shumway, 2001; Macbeth & Merville, 1979). If the CEO does not meet the requirements for classification as overconfident, he is regarded as rational. This is the first paper that makes a distinction between these two groups and investigates their effect on the relationship between CEO compensation and CSR. I address the sensitivity of the Holder 67 measure by repeating the analysis using different levels for the benchmark.
The study finds that compensating overconfident CEOs by short-term incentive payments, such as bonus and salary, are significantly positively associated with CSR. Moreover, in the case of firms with rational CEOs, salary is significantly positively related to CSR. An explanation for this is that socially responsible firms attract highly skilled CEOs who demand more base salary, which is also suggested by Sun and Yu (2015). Furthermore, I find that CEO stock options and total compensation do not have a significant impact on the relationship between overconfident CEOs and CSR. The results are consistent after applying robustness checks.
2. Literature
This section starts by defining the main concepts and their interaction to clarify the research. This is followed by a synthesis of the literature. Table A.1 in the Appendix presents a short overview of the empirical literature, which includes the sample period and region of previous studies, their measurement of overconfidence, and their main results. A significant amount of research on CSR has already been carried out. This section deals primarily with overconfident CEOs, since the literature on rational CEOs associated with CSR is limited. There are only a few studies that have investigated the effect of CEO compensation on CSR. A closer look at the impact of CEO overconfidence on CSR can provide insights about the effect of the compensation of overconfident CEOs on CSR. The hypotheses are based on the literature.
2.1 Background
A key concept in this study is overconfidence. The empirical literature gives several definitions of overconfidence. Daniel, Hirshleifer, and Subrahmanyam (2001) define overconfident CEOs as CEOs who overestimate their expertise. Scheinkman and Xiong (2003) define overconfidence as the belief of an agent that he owns more accurate information than others. Malmendier and Tate (2005) define overconfident managers as those who overestimate the returns of their investments. As there are many definitions of overconfidence, this research defines an overconfident CEO as one who overestimates the value of the firm and tends to overinvest in projects that are too risky. A CEO who is not overconfident is classified as a rational, risk-averse CEO who makes decisions that are prudent and logical and does not invest in projects that are too risky. Some studies use the terms “hubris” and “overconfidence” interchangeably with regard to CEOs (see, e.g., Chen et al., 2015; Chen et al., 2018). For the purposes of making comparisons between studies, CEO hubris is considered as the same as CEO overconfidence.
Firms with overconfident CEOs tend to engage in risky projects and therefore display less CSR. Corporate social responsibility is composed of three elements; environmental, social, and corporate governance. Bénabou and Tirole (2010) define CSR as giving up some profit in the social interest in which a firm must go voluntarily beyond its legal and contractual obligations. They discuss three views on CSR. The first view is the “doing well by doing good” view. It states that CSR takes a long-term perspective and strengthens a firm’s market position, which increases long-term profits. The second view is about delegated philanthropy, which ensures that firms engage in CSR on behalf of their stakeholders, because stakeholders demand that the firm engages in philanthropy. The last view relates to insider-initiated corporate philanthropy, which reflects the executives’ desire to engage in philanthropy.
pay-off structure for stock options, as pointed out by Hull (2009). Stock options incentivize CEOs to invest more in risky projects, because the return of such projects is more volatile, which could lead to a higher pay-off. Consequently, they invest less in CSR activities, since these activities generate a high pay-off in the long term, not in the short term, as identified in the study by Purnamasari et al., (2015). However, when overconfident CEOs’ total compensation consists primarily of stock options, their compensation is volatile, and they may change their behavior and invest in less risky projects.
In general, there are two contradictory theories of CSR, namely the stakeholder theory and the agency theory. The stakeholder theory was initially suggested by Freeman (1984), who defines a stakeholder as “any group or individual that can affect or is affected by the achievement of a corporation’s purpose”. The theory states that the firm should not only act in the interests of its shareholders, but also in the interests of its stakeholders. Donaldson and Preston (1995) divide the stakeholder theory into three categories. First, the normative stakeholder theory explains why firms should take stakeholder interests into account. Second, the descriptive stakeholder theory interprets whether and how firms do take stakeholder interests into account. Third, the instrumental stakeholder theory examines the benefits for a firm when they take into account stakeholder interests. Based on the stakeholder theory, CEO’s take less compensation to mitigate conflicts of interest among stakeholders and to increase stakeholder loyalty. Moreover, Cai et al., (2011) argue that CEOs who care about social and ethical standards desire less compensation. Therefore, this theory suggests a negative relationship between CEO compensation and CSR.
as the compensation when the executives misbehave (see, e.g., Gale & Hellwig, 1985; Innes, 1990).
2.2 Literature review
Malmendier and Tate (2005) adopt an empirical approach in the behavioral finance literature on measuring CEO overconfidence. They argue that the timing of exercising executive stock options depends on the moneyness of the options, the risk aversion of the CEO, and the diversification of his or her portfolio. These authors created three measures of overconfidence:
Holder 67, Longholder, and Net Buyer. According to the first measure, Holder 67, a CEO is
classified as overconfident if his stock options are more than 67% in the money at least twice during the sample period and if he still does not exercise his options. In terms of the second measure, Malmendier and Tate (2015) identify two requirements for classifying a CEO as a ‘Longholder’: first, a CEO should hold a stock option until the year of expiration and second, the options are at least 40% in the money at the start of the year of expiration. Since most CEO stock options have time to maturity of ten years, and after four years the stock options are in most cases fully vested, holding these in the money options until the year of expiration is a risky strategy and requires overconfidence on the part of the CEO about the firm’s future performance. The last measure is the Net Buyer, which identifies a CEO as overconfident if he habitually buys the stock of his company during the five first sample years. To summarize, both the Holder 67 and Longholder measures use the timing of the option exercise to determine the overconfidence level, while the Net Buyer measure uses the CEO’s yearly purchases of his company’s stock to determine the overconfidence level.
According to Chen et al., (2018), a significant negative relationship exists between CEO hubris and CSR. This result is based on a sample of 397 US firms from 2001 to 2010. They compute an aggregate measure of CSR, based on KLD data, and construct a press-based measure for CEO hubris. This finding is strengthened by the study of McCarthy et al., (2016), who have also proved a significant negative relationship between CEO overconfidence and CSR. They measure CEO overconfidence in the way recommended by Malmendier and Tate (2005). Another finding is that female CEOs engage in more CSR than male CEOs, which is supported by Wu, Zhu, Zou, and Yang (2018). There is also a positive relationship between the tenure of the CEO and CSR, and a negative relationship between age and CSR.
Chen et al., (2015) make a distinction between CSR strengths and CSR concerns. They use press coverage to measure the confidence level of a CEO and they use KLD data to obtain their CSR measure. A significant negative relationship is identified between CEO hubris and CSR strength, and a significant positive relationship between CEO hubris and CSR concern. Moreover, the authors find that firm size is significantly positively related to CSR strength and CSR concern, which is also demonstrated in the study by Manner (2010).
concerns. They also make a distinction between normal and abnormal CSR, where normal CSR is the optimal level of CSR, and abnormal CSR is the deviation from the optimal level of CSR. A significant positive relationship is found between normal CSR and CEO compensation, and a significant negative relationship between abnormal CSR and CEO compensation. This finding suggests that CEOs are punished when they deviate from the optimal CSR investment level.
Similar research was carried out by Mahoney and Thorn (2006), who focus on a Canadian sample of 69 firms for the period 1995 and 1996. They test the impact of CEOs’ salary, bonus and stock options on CSR strengths, CSR weaknesses, and total CSR; while Jian and Lee (2015) have tested the reverse – the influence of CSR on CEO compensation. CSR strengths have a positive relationship with CEO bonus and a negative relationship with stock options. CSR weaknesses have a significant positive relationship with CEO salary.
Several studies demonstrate a significant positive relationship between CEO stock options and total CSR (see, e.g., Deckop, Gupta & Merriman, 2006; Mahoney & Thorn, 2006). However, the study by Chen et al., (2018) demonstrates an insignificant relationship between CSR and long-term pay focus, consisting of stock options and restricted stocks. I expect a positive relationship between CEO stock options and CSR for the following reason. Compensating CEOs primarily by stock-based compensation results in higher exposure to the firm’s risk. Therefore, the CEO will invest in less risky projects, and will consequently invest in CSR. I expect that the results will be the same for both an overconfident CEO and a rational CEO. This reasoning leads to the first hypotheses:
Hypothesis 1a: There is a positive relationship between stock-based compensation for overconfident CEOs and CSR.
Hypothesis 1b: There is a positive relationship between stock-based compensation for rational CEOs and CSR.
Deckop et al., (2006) focus on short-term incentive pay that consists of both bonus and salary. They indicate a significant negative relationship between short-term pay and CSR. The study by Mahoney and Thorn (2006) finds a positive relationship between CSR strengths and CEO bonus. I also anticipate a positive relationship between CEO bonus and CSR, based on the following two explanations. First, a higher CEO bonus could lead to dissatisfaction among stakeholders. Overconfident CEOs regard their status as essential, and are susceptible to people’s opinion of them. Therefore, to maintain the CEO’s status and to prevent a dissatisfaction among stakeholders when an overconfident CEO is compensated by an increased bonus, the firm engages in responsible activities to satisfy stakeholders. Second, a CEO receives a bonus only when certain requirements are met in a particular year. Nowadays, firms are focusing more on sustainability.
require that the firm should engage in more socially responsible activities in a particular year; and the CEO receives the bonus when the goal has been achieved. In summary, CEOs could be rewarded for CSR performance in the short term. This expectation results in the second hypotheses, namely:
Hypothesis 2a. There is a positive relationship between a bonus of overconfident CEOs and CSR.
Hypothesis 2b. There is a positive relationship between a bonus of rational CEOs and CSR.
Mahoney and Thorn (2006) do not find a significant relationship between CEO salary and CSR. Salary is fixed and therefore independent of performance. Armstrong and Jacquart (2013) indicate that CEOs who are intrinsically motivated achieve satisfaction by doing their task and therefore require less extrinsic motivation, such as compensation. It can therefore be assumed that a CEO with intrinsic motivations to engage in CSR would be more likely to accept less salary. On the other hand, a CEO without these intrinsic motivations to engage in CSR would not accept less salary. Hence, I expect CEO salary to have a significant negative impact on CSR. This leads to the following hypotheses:
Hypothesis 3a. There is a negative relationship between the salary of overconfident CEOs and CSR.
Hypothesis 3b. There is a negative relationship between the salary of rational CEOs and CSR.
Cai et al., (2011) arrive at results that are similar to those of Jian and Lee (2015), namely that CEOs of socially responsible firms receive lower total compensation than CEOs of socially irresponsible firms, and this is strengthened by the study of Benson, Faff, and Rekker (2014). These studies did not include the overconfidence level of CEOs in their research. Based on the stakeholder theory, I expect that socially responsible firms are more aware of determining total CEO compensation, for both overconfident CEOs and rational CEOs. Socially responsible firms are more conscious of potential conflicts of interests among stakeholders and therefore the CEO is willing to accept lower compensation, as argued by Cai et al., (2011). The following hypotheses test these predictions:
Hypothesis 4a. There is a negative relationship between total compensation and investing in CSR by overconfident CEOs.
3. Data and methodology
First, I discuss the research design of my model, after which I explain how I constructed the sample. Next, I describe the measurement of CSR, the main independent variables, and the control variables and discuss the descriptive statistics. The methodology is presented in section 3.7, while the robustness tests are dealt with at the end.
3.1 Research design
To investigate the differences between rational CEOs and overconfident CEOs, I need to use a model that includes CEO characteristics. Moreover, CSR is strongly determined by firm characteristics, and the model should therefore also include firm characteristics. Following prior studies on the determinants of the CSR level (see, e.g., Callan & Thomas, 2010; Jian & Lee, 2015; Mahoney & Thorn, 2006), I create the following standardized model, mathematically denoted, to test the various hypotheses:
"#$%& = ( (*%&, ,%&, -%&) (1) where "#$%& refers to the ESG score for a specific firm at a particular point in time. *% is a
dummy variable and indicates whether the CEO is overconfident or not. ,%& relates to the CEO compensation of a CEO at a specific point in time, measured by different compensation variables. -%& refers to the control variables at a particular point in time. The following regression model tests the relationship between the compensation of CEOs and CSR:
"#$%& = / + 123+ 43#56578%&+ 49:;<=>%&+ 4?#@;,ABC@D;<%&+
4EF;@56";GCH<>5@D;<%&+ 4I";<@7;6J57D5K6H>%&+ L%&+ M%& (2) The term L%& refers to the year fixed effects and M%& represents the error term. Salary, bonus, stock options, and total compensation are the variables of interest in this equation. The other variables are control variables, which are described in detail in section 3.5. The control variables consist of CEO characteristics and firm characteristics.
The Hausman test is performed to determine whether a random effect model or a fixed effect model is required. The dummy variable significance test shows that adding year fixed effects to the model is needed to control for differences across years; Jian and Lee (2015) have also added year fixed effects to their model. Industry fixed effects are not included in the model, because the industry to which a firm belongs does not change over time. Callan and Thomas (2010) do not use fixed effects in their models. Further limitations of the fixed effects model are that the model cannot test binary variables, unless the indication of the binary variable is specified; otherwise, the fixed effects model omits these variables automatically. Therefore, I do not include the variable gender in the model, since it is a binary variable. However, to examine whether the results are different based on gender, I estimate the models separately for men and women in the sensitivity analysis.
3.2 Sample construction
The sample is restricted to companies listed on the constituent list of the S&P500. The choice to select companies located in the US is based on a number of reasons. First, data about companies listed on the S&P500 are relatively easy to obtain, since I only have access to certain databases. In addition, using the S&P500 results in a large sample. Furthermore, the main related study is that by Mahoney and Thorn (2006), who use Canada in their study. Canada has a different institutional context than the US, and I could therefore detect different results in my sample.
The dataset ranges from 2002 to 2018, as data regarding the ESG score have only been available since 2002. This range makes it possible to compare the pre-crisis period and the post-crisis period with a view to analyzing the effect of the 2008 financial post-crisis. By taking a long sample period, I can observe changing overconfidence behavior over the years. A normality test is performed to check the distribution of the residuals. The distribution is presented in Figure A.13 in the Appendix and the data seems normally distributed.
3.3 Measurement of CSR
Data on CSR are obtained from the Thomson Reuters Eikon Database, which provides an ESG score to measure the level of CSR in a transparent and objective manner. The database provides ratings on over 7,000 companies, dating back to 2002. The Thomson Reuters ESG scores are an improvement on the ASSET4 rating, which is used in empirical research, and are a robust measure of companies ESG performance and capacity. The ESG score is based on the following ten ESG topics: resource use, emissions, innovation, management, shareholders, CSR strategy, workforce, human rights, community, and product responsibility. Another rating to measure CSR is KLD’s social and environmental rating, which provides scores relating to environmental strengths and environmental concerns of firms. However, due to a lack of access, the KLD ratings are not used for this research. This study therefore does not distinguish between CSR strengths and concerns; the focus is on the total ESG score.
3.4 Main independent variables
The overconfidence level of a CEO is measured by using the theory of Malmendier and Tate (2005). I use their Holder 67 measure; due to a lack of access, I cannot use their Longholder and Net Buyer measures. The Holder 67 measure indicates that a CEO is overconfident if his stock options are more than 67% in the money at least twice during the sample period and he still holds the options rather than exercising it. A limitation of this measure is that a CEO is only classified as overconfident or rational; other classifications are not possible. The use of a discrete variable could give biased results. I test different threshold percentages to prove the robustness of the measure. I use the following formulas to calculate the moneyness of the options (see, e.g., Campbell et al., 2011; Humphery-Jenner, Lisic, Nanda & Silveri, 2016):
NOH75PH O56=H CH7 OH>@HQ ;C@D;< = J56=H ;( OH>@HQ =<HRH7,D>HQ ;C@D;<> S=GKH7 ;( OH>@HQ =<HRH7,D>HQ ;C@D;<> NOH75PH >@7DAH C7D,H = #@;,A C7D,H − NOH75PH O56=H CH7 OH>@HQ ;C@D;<
U;<H8<H>> ;C@D;<> =NOH75PH O56=H CH7 OH>@HQ ;C@D;< NOH75PH >@7DAH C7D,H
Other main independent variables are the compensation variables. The total compensation of a CEO consists of short-term incentive payments (e.g., bonus and base salary) and long-term incentive payments (i.e., stock options). CEO bonus and base salary are respectively measured as the dollar value of the bonus and base salary for a CEO during a particular fiscal year. The value of stock options refers to the value of the total stock options of a CEO that are exercised during a fiscal year and is calculated as the difference between the strike price and the exercise price of the stocks on the exercise date.
3.5 Control variables
The expected impact of the control variables is presented in Table A.4 in the Appendix. I base the sign of the variables partly on my expectations of the impact of the variables discussed in this section. In addition, I base the sign of the coefficients on the findings from previous studies that focus on the relationship between CEO compensation and CSR.
Empirical studies are examined to determine the control variables that could have a significant influence on CSR. Firm performance, firm size, and firm risk are included as firm measures. In the light of the debates in the literature about the proper variable to measure firm performance in this type of study, I measure firm performance by multiple variables: Tobin’s Q, the return on assets (ROA), and the market-to-book ratio (MtB). Using ROA and MtB is in accordance with the studies of Jian and Lee (2015), Chen et al., (2018), and Dimson, Karakas and Li (2015). Tobin’s Q is less frequently used as a measure for firm performance (see, e.g., Hirshleifer, Low & Teoh, 2012; Karim, Lee & Suh, 2018).
Assuming that the book value of the liabilities of the firm is equal to the market value, Tobin’s Q is calculated by the ratio of the market value of assets divided by the book value of assets; where the market value of the assets is calculated by adding the total assets and the market value of equity and subtracting the book value of equity. The book value of assets equals the total value of assets. Return on assets is measured by the ratio of EBITDA to book value of total assets. To calculate the market-to-book ratio, I divide the market value of equity by the book value of the equity. The market value of equity is calculated as the firm’s common shares outstanding multiplied by the closing price of the fiscal year. Book value of equity is calculated as total assets, minus total liabilities, minus preferred stock, plus deferred taxes, plus convertible debt. I expect that firm performance has a positive impact on CSR, since better performing firms are able to distinguish themselves and are more innovative.
and weekends). Assuming that firms with high volatility invest in risky projects and less in socially responsible projects, this results in a negative relationship between firm risk and CSR.
Governance is captured by board size. Yermack (1996) finds that firms with smaller boards are more effective monitors. Therefore, I anticipate that board size has a significant negative impact on the ESG score. Moreover, CEO tenure and CEO age are added as human capital measures. A positive relation between CSR and CEO age is expected, since CEOs tend to become more ethical as they get older (see, e.g., Franke, Singhapakdi & Vitell, 1999; Terpstra, Rozell & Robinson, 1993). A significant positive relationship between CEO tenure and CSR is foreseen, which is consistent with the results found in the study by McCarthy et al., (2016). For year controls, I create dummies for every year in the sample period. The use of a fixed effect model ensures that industry dummies cannot be added to the model.
3.6 Summary statistics
Table 1 presents the descriptive statistics of the relevant variables in this research. This section highlights certain striking observations. A total of 4,151 observations are made; 2,717 CEOs are overconfident, and 1,434 CEOs are rational; the majority of the CEOs in the sample are overconfident. This is consistent with the finding of Goel and Thakor (2008), who demonstrate that an overconfident manager has a higher probability of becoming a CEO than a rational manager, and therefore, there are more overconfident CEOs than rational CEOs.
The mean of the ESG score is 56.67. This score means that, on average, the companies listed on the S&P500 between 2002 and 2018 had an ESG score of 56.67. A high degree of skewness exists in the distribution of stock options. Since stock options are part of total compensation, total compensation also exhibits high skewness. This could be explained on the basis of the significant differences in stock option compensation – some CEOs are compensated by $2.276 million in stock options, whereas other CEOs receive no stock options whatsoever. We can also see that some CEOs receive no base salary. In view of the media attention focused on CEO salaries, several CEOs require no base salary, as they do not want their company represented in a bad light. These CEOs are usually compensated in other ways.
The correlation matrix in Table A.3 in the Appendix presents the correlation coefficients of the ESG score, the compensation variables, and the control variables. We can observe a significant negative correlation between overconfident CEOs and the ESG score, which is consistent with empirical studies (see, e.g., Chen et al., 2018; McCarthy et al., 2016). CEO salary is significantly positively correlated with the ESG score, while CEO bonus is significantly negatively correlated with the ESG score. This significance is remarkable, because the study by Mahoney and Thorn (2006) did not find a significant correlation between bonus and the ESG score.
compensation. There are two variables in the sample which are almost perfectly correlated: CEO stock options and total compensation. The correlation coefficient of 0.999 is explained by the fact that a large part of total compensation consists of CEO stock options. Moreover, men invest less in CSR, which is consistent with the finding of McCarthy et al., (2016). Women receive more salary than men, which is remarkable, but this could be because the sample consists of only 2.24% female CEOs.
Table 1
This table provides an insight into the descriptive statistics for the socially responsibility score, the CEO compensation characteristics, and CEO and firm characteristics. The data relate to firms listed on the S&P500 over the period 2002-2018. There are 4,151 observations for each variable. The definition of the variables is presented in Table A.2 in the Appendix.
Mean St. dev. Min Max Skewness Kurtosis
Social responsibility score ESG score 56.67 16.828 8.0168 95.675 -0.170 2.226 CEO compensation characteristics CEO salary (*$1,000) 1,080 562 0 5613 1.824 12.959 CEO bonus (*$1,000) 568 1750 0 32000 8.688 114.69
CEO stock option (*$1,000) 7317 40034 0 2276678 44.97 2499 CEO total comp. (*$1,000) 8974 40146 0 2277447 44.56 2467
CEO characteristics
CEO age (in years) 55.373 6.278 27 81 -0.014 3.451
CEO tenure (in years) 10.363 6.741 0 54 2.067 10.723
3.7 Panel data framework
My dataset contains both time series and cross-sectional elements, and I therefore deal with a panel of data. I could use a pooled OLS technique or a panel data framework to estimate the regressions. However, a panel of data has several advantages over pooled OLS; it is possible to address more complex problems, to investigate how variables change over time, and there is no omitted variable bias. Therefore, I use a panel data framework to measure the impact of compensation of both overconfident CEOs and rational CEOs on CSR. An Anova test is performed to analyze the differences between these two groups. The outcome of this test is presented in Table A.14 in the Appendix – the significant p-values indicate the existence of differences between the two groups concerning the variables, except for CEO bonus and market-to-book ratio.
3.8 Robustness tests
This section presents the different robustness tests that I perform in this study. For the sake of simplicity, I consider only the overconfident CEOs; not the rational CEOs. The outcome of these tests is discussed in section 4.4. The use of the fixed effects model ensures that the results do not suffer from endogeneity. Overconfidence is determined by the Holder 67 measure of Malmendier and Tate (2005); a CEO is overconfident if his stock options are more than 67% in the money at least twice during the sample period. Like Campbell et al., (2011), I use different percentages – 60% and 80% – to test the sensitivity of the benchmark.
The impact of the compensation variables of CEOs is tested on the total ESG score that consists of three pillars – environmental, social, and governance. Table A.5 in the Appendix demonstrates that the mean of the social pillar score is quite high relative to the mean of the governance pillar score – 59.755 and 53.107 respectively. This suggests that, on average, firms devote more attention to social activities than to governance activities. Therefore, I test the model on the three terms separately, to determine whether the different dependent variables lead to the same results. Moreover, Chatterji, Levine and Toffel (2009) also test their model on multiple dependent variables relating to CSR concerns and strengths, and they demonstrate that their model impacts the dependent variables differently. The environmental pillar consists of the resource use, emissions, and innovation variables. The social pillar consists of the workforce, human rights, community and product responsibility variables. The governance pillar consists of the management, shareholders, and CSR strategy variables.
As an additional robustness test, I divide the sample into two subsamples to test the existence of a structural change during the financial crisis. The first subsample covers the pre-crisis period, from 2002 until 2008. The year 2008 is also included, since it is expected that the start of the crisis in September 2008 does not have an immediate effect on CEO compensation. The second subsample includes the post-crisis period, the years 2009 to 2018.
4. Results
This section presents the results of the main analysis. First, I discuss the results of the model for the overconfident CEOs, followed by the results of the model for rational CEOs. Section 4.3 discusses the differences between the results of the two groups. In section 4.4, I present the results of the robustness tests.
4.1 Results for overconfident CEOs
The models in Table 2 include a binary variable which ensures that only overconfident CEOs are selected. Therefore, the models in Table 2 present the effect of compensation of overconfident CEOs on the firm’s ESG score. Models 1 and 2 include only the control variables, which is the baseline model. Models 3 and 4 add the CEO’s salary, bonus and stock options variables, which is the fully specified model. Models 5 and 6 test the effect of total compensation.
The literature provides mixed results relating to the influence of CEO stock options on CSR. Hypothesis 1a predicts that stock options of overconfident CEOs are positively related to the ESG score. Models 3 and 4 in Table 2 both demonstrate that stock options of overconfident CEOs do not have any impact at all, resulting in acceptance of the null hypothesis and implying that compensating an overconfident CEO with stock options does not have an effect on the firm’s ESG score, which is consistent with the findings of Chen et al, (2018). However, this result is not in line with the outcomes of several other studies (see, e.g., Deckop et al., 2006; Jian & Lee, 2015), which demonstrate a significant positive relationship between CEO stock options and total CSR. The different way in which these various studies measure CSR could explain the different results. Moreover, other studies focus on the entire group of CEOs, not only on overconfident CEOs.
Hypothesis 2a forecasts that the bonus of overconfident CEOs positively influences CSR. The CEO bonus coefficient is 0.001, which is positive and significant. This leads to a rejection of the null hypothesis at the 5% significance level, meaning that a 1% increase in bonus increases the ESG score by 0.1% on average. One can argue that this impact is negligible. However, it is realistic that a CEO’s bonus increases by, for example, 10%, resulting in a 1% increase in the firm’s ESG score, which is economically significant. Therefore, I conclude that bonus of an overconfident CEO is positively related to CSR. This finding is consistent with the study by Mahoney and Thorn (2006), who also found a positive relationship between CSR strengths and CEO bonus.
did not conclude that the salary of CEOs positively influences CSR; they did not find any significance at all. An explanation for this is that socially responsible firms attract highly skilled CEOs who demand more salary, which is also suggested by Sun and Yu (2015).
The last hypothesis tests the effect of total compensation of overconfident CEOs on a firm’s ESG score. Models 5 and 6 in Table 2 demonstrate that total compensation does not have any impact on a firm’s ESG score. This finding is not consistent with empirical findings (see, e.g., Cai et al., 2011; Jian & Lee, 2015; Rekker et al., 2014). This is caused primarily by the fact that the total compensation consists largely of CEO stock options, and CEO stock options do not have a significant impact on CSR either.
Table 2
This table presents the outcome of the fixed effects model for overconfident CEOs. The dependent variable is the ESG score of a particular firm. The definitions of the variables are presented in Table A.2 in the Appendix. The models are estimated by year fixed effects. Standard errors are reported in parentheses and are robust to heteroskedasticity. *** indicates significance at the 1% level, ** indicates significance at the 5% level, and * indicates significance at the 10% level.
(1) (2) (3) (4) (5) (6) Constant 15.578 77.986 14.081 90.535* 15.704 78.411 (17.952) (51.492) (16.319) (51.771) (17.996) (51.599) CEO salary 0.001 0.001** (0.000) (0.000) CEO bonus 0.001** 0.001** (0.000) (0.000)
CEO stock options 0.000 0.000
(0.000) (0.000)
CEO total comp. 0.000 0.000
(0.000) (0.000) CEO age 0.364 -0.946 0.421 -1.189 0.366 -0.951 (0.374) (1.074) (0.340) (1.078) (0.374) (1.075) CEO tenure 1.102*** 1.154*** 1.083*** 1.128*** 1.100*** 1.151*** (0.367) (0.359) (0.334) (1.107) (0.368) (0.361) Firm size 1.487 1.604 1.153 1.284 1.467 1.585 (1.070) (1.109) (1.072) (1.107) (1.075) (1.114) Tobin’s Q -0.057 -0.111 -0.103 -0.118 -0.064 -0.117 (0.257) (0.256) (0.260) (0.260) (0.259) (0.259) ROA 1.399 1.976 -0.292 0.217 1.347 1.909 (3.386) (3.452) (3.327) (3.400) (3.405) (3.469) Market-to-book -0.000*** -0.000*** -0.000*** -0.000** -0.000*** -0.000** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Firm risk 0.383 0.233 0.694 0.271 0.381 0.221 (1.148) (1.257) (1.114) (1.245) (1.149) (1.256) Board size -0.341* -0.409** -0.345 -0.410** -0.342* -0.410** (0.187) (0.182) (0.186) (0.182) (0.188) (0.183)
Year fixed effects No Yes No Yes No Yes
N 2,717 2,717 2,717 2,717 2,717 2,717
F 46.57*** 23.51*** 33.41*** 19.85*** 41.55*** 22.42***
4.2 Results for rational CEOs
The models in Table 3 indicate the results for rational CEOs. I construct these models in the same way as Table 2, but the models in Table 3 include a binary variable, which ensures that only rational CEOs are involved. The CEO tenure variable is deleted due to collinearity.
Hypothesis 1b predicts that stock options of rational CEOs are positively related to the ESG score. Models 9 and 10 both demonstrate that the coefficient of CEO stock option is significant. However, the significant coefficient of 0.000 means that CEO stock options do not influence CSR; therefore, no economic significance exists. Empirical research demonstrates a positive relationship between CEO stock options and CSR. The different results arise because certain variables are measured differently. Jian and Lee (2015) calculate CSR as KLD strengths minus KLD concerns. Corporate social performance (CSP) is established in the study by Deckop et al., (2006) by using six KLD dimensions. I use the ESG score to measure CSR, because I do not have access to KLD data.
Furthermore, Hypothesis 2b expects that the bonus of rational CEOs is positively related to CSR. Model 10 indicates that the bonus coefficient is insignificant. Therefore, I cannot reject the null hypothesis and conclude that the bonus of rational CEOs is not related to CSR. Mahoney and Thorn (2006) find a significant positive relationship between CEO bonus and CSR, which I also found for overconfident CEOs, as discussed in the previous section. Therefore, an explanation for this is that their sample consists mainly of overconfident CEOs. Moreover, the sample of Mahoney and Thorn (2006) consists of Canadian firms, which could also cause the difference.
In addition, Hypothesis 3b states that there is a negative relationship between the salary of rational CEOs and CSR. The salary coefficient for rational CEOs in Model 9 is not significant, but in Model 10 is significantly positive at the 5% level. This significance means that increasing salary for rational CEOs by 10% results, on average, in an increase in the ESG score of 2%, which is economically significant. This is inconsistent with my prediction and the findings of Mahoney and Thorn (2006). My model is based on their model, but I have added additional control variables, which could explain the different results. Consequently, I reject the null hypothesis and conclude that increasing CEO salary of a rational CEO leads to a higher ESG score.
Cai et al., (2011) measure CSR in the same way as this study, i.e. using total CSR instead of concerns and strengths. However, they use KLD data to measure CSR, whereas I use the ESG score, which could cause differences.
Table 3
This table presents the outcome of the fixed effects model for rational CEOs. The dependent variable is the ESG score of a particular firm. The definitions of the variables are given in Table A.2 in the Appendix. The models are estimated by year fixed effects. Standard errors are reported in parentheses and are robust to heteroskedasticity. *** indicates significance at the 1% level, ** indicates significance at the 5% level, and * indicates significance at the 10% level. (7) (8) (9) (10) (11) (12) Constant 22.550 160.285*** 34.387* 163.549*** 24.166 167.766*** (18.518) (57.094) (18.511) (51.771) (18.689) (58.688) CEO salary 0.003 0.002** (0.002) (0.002) CEO bonus -0.001* 0.000 (0.000) (0.000)
CEO stock options 0.000* 0.000**
(0.000) (0.000)
CEO total comp. 0.000 0.000**
(0.000) (0.000) CEO age 1.572*** -1.365 1.404*** -1.372 1.567*** -1.467 (0.208) (1.229) (0.209) (1.256) (0.209) (1.257) Firm size -4.445** -4.444* -4.960** -4.837** -4.575** -4.703** (2.050) (2.268) (2.043) (2.268) (2.044) (2.259) Tobin’s Q -2.077*** -1.132 -1.974*** -1.174 -2.138*** -1.239 (0.716) (0.933) (0.730) (0.928) (0.704) (0.909) ROA 8.640** 4.136 7.970** 3.729 8.499** 3.805 (3.753) (3.200) (3.605) (3.273) (3.733) (3.195) Market-to-book -0.011* -0.010* -0.011* -0.010* -0.011* -0.010* (0.007) (0.005) (0.007) (0.006) (0.007) (0.005) Firm risk 3.421 -2.229 2.773 -2.186 3.443 -2.258 (2.213) (2.468) (2.195) (2.446) (2.213) (2.449) Board size -0.210 -0.131** -0.222 -0.146 -0.216 -0.142 (0.213) (0.208) (0.210) (0.207) (0.211) (0.206)
Year fixed effects No Yes No Yes No Yes
N 1,434 1,434 1,434 1,434 1,434 1,434
F 10.80*** 6.80*** 9.26*** 6.77*** 11.25*** 7.37***
4.3 Comparison between results
This section discusses the differences between the results of overconfident CEOs and rational CEOs. In addition, I create a model that includes both the overconfident and rational CEOs, which I call the “entire group”. The results of the entire group are presented in Table A.6 in the Appendix. An advantage of using the entire group is that I can include the Holder 67 measure, because now it is not necessary to make a distinction between overconfident CEOs and rational CEOs. We can observe that the Holder 67 coefficient is significantly negatively related to CSR, consistent with previous studies (see, e.g., Chen et al., 2018; McCarthy et al., 2016).
Furthermore, the table indicates that CEO salary is significant for the entire group, which is consistent with the results for the overconfident CEOs and rational CEOs separately. Since these findings are not in line with the study by Mahoney and Thorn (2006), I divide the sample into four subgroups in order to gain a better understanding of the results. I use firm size to create the four subgroups, because it is expected that larger firms have more resources and are therefore better able to invest in social responsible activities. The results are presented in Tables A.7 and A.8 in the Appendix. We can observe that CEO salary is significant only for firms that are larger than the average firm in the sample, since only Models 3 and 4 indicate significant CEO salary coefficients. This finding applies to firms with both overconfident CEOs and rational CEOs.
It is remarkable that the coefficient CEO bonus is significant only for overconfident CEOs, and not for the entire group and rational CEOs. One reason that firms engage in social responsible activities is to satisfy stakeholders. Higher CEO bonus could lead to dissatisfaction among stakeholders. Overconfident CEOs regard their status as essential, and they are sensitive to the opinions of others. Therefore, in order to maintain the CEO’s status and to prevent dissatisfaction among stakeholders when an overconfident CEO is compensated by a higher bonus, the firm engages in social responsible activities to satisfy stakeholders.
taking the natural logarithm of a firm’s total assets. However, they use KLD data to measure CSR, whereas I use the ESG score, which could cause the different results.
To summarize, the main difference between the groups is that the coefficient CEO bonus is significant only for overconfident CEOs; not for the entire group and rational CEOs. CEO salary has a significantly positive relationship with CSR, for both overconfident CEOs and rational CEOs. The next section discusses the robustness tests for overconfident CEOs. 4.4 Additional analyses and robustness tests
This section discusses the results of the robustness tests described in section 3.8. The main findings of the previous sections indicate that compensating overconfident CEOs by bonus and salary are significantly positively associated with corporate social responsibility. For firms with rational CEOs, only salary is significantly positively related to a firm’s socially responsible activities; other compensation variables do not have an effect. Therefore, I perform the robustness tests only for overconfident CEOs, and not for rational CEOs.
Table A.9 in the Appendix presents the results of the sensitivity analysis of the Holder 67 measure. Model 1 indicates the results of the main model in which Holder 67 measures overconfidence. Models 2 and 3 indicate respectively the results where a CEO is indicated as overconfident when his options are at least twice for 60% (Holder 60) and 80% (Holder 80) or more in the money. The results of the model are similar when the Holder 60 and Holder 80 overconfidence measures are used. Therefore, the results of the main model are robust to the sensitivity of the benchmark Holder 67.
The results of three dependent variables are presented in Table A.10 in the Appendix. These dependent variables are the environmental pillar score, the social pillar score, and the governance pillar score. We can conclude that the results are similar when the environmental pillar score and the governance pillar score are used as the dependent variable. However, the CEO bonus and CEO salary coefficients are not significant when the social pillar score is used as the dependent variable. Therefore, the compensation structure of overconfident CEOs does not have an impact on the social activities of a firm, but affects the environmental and governance activities. This could be explained by the fact that the average social pillar score is already quite high relative to the environmental and governance pillar score, as can be observed in Table A.5 in the Appendix. Therefore, the board could require a CEO to focus on environmental and governance activities rather than on social activities. When the CEO achieves the targets, he receives a bonus.
two subsamples are created to test the differences between male CEOs and female CEOs. The models are presented in Table A.12 in the Appendix. I conclude that gender does not influence the relationship between CEO compensation and the ESG score, which can be explained by the fact that female CEOs constitute only 2.3% of the sample.
5. Conclusion
This study has examined the influence of the CEO compensation structure on CSR. Its objective was to compare overconfident CEOs to rational CEOs to examine the impact of the compensation structure of these two groups on CSR. The Holder 67 measure of Malmendier and Tate (2015) is used to measure the overconfidence level of a CEO. I find evidence that the compensation structure of both overconfident CEOs and rational CEOs does have an impact on CSR.
Previous studies offer mixed results about the relationship between CEO stock options and CSR. I find that there is no evidence of a significant relationship between CEO stock options and CSR, with regard to either overconfident CEOs or rational CEOs. This finding is consistent with the study by Chen et al., (2018). Other studies that do find a significant relationship (see, e.g., Deckop et al., 2006; Jian & Lee, 2015) use a different method for measuring CSR, which could cause different results. The second conclusion of this study is that the bonus of overconfident CEOs positively influences CSR, which is consistent with the findings of Mahoney and Thorn (2006). However, I find an insignificant relationship between the bonus of rational CEOs and CSR. The sample of Mahoney and Thorn (2006) could consist primarily of overconfident CEOs, which explains the differences.
Furthermore, in the case of both overconfident and rational CEOs, CEO salary is significantly positively related to CSR, especially for firms that are larger than the average firm in the sample. This result is not in line with either the third hypothesis or the literature. An explanation for this could be that socially responsible firms attract highly skilled employees who demand more salary, which is also suggested by Sun and Yu (2015). The last hypothesis expects a positive relationship between the total compensation of CEOs and CSR. However, the analysis follows that total compensation does not influence the relationship between CEOs and CSR, in the case of both overconfident CEOs and rational CEOs. This finding is not in line with previous studies (see, e.g., Cai et al., 2011; Jian & Lee, 2015; Rekker et al., 2014). In my sample, total compensation consists primarily of CEO stock options, which could explain the different results.
The results are robust to the financial crisis, to different methods of measuring CEO overconfidence and to lagged variables. In conclusion, compensating overconfident CEOs with short-term incentive payments such as bonus and salary is positively related to CSR. In addition, the salary of rational CEOs has a significantly positively influence on CSR.
References
Armstrong, J.S., Jacquart, P., 2013. The ombudsman: are top executives paid enough? An evidence-based review. Interfaces 43, 580-589.
Becker, G.S., 1962. Irrational behavior and economic theory. Journal of Political Economy 70, 1-13.
Bénabou, R., Tirole, J., 2010. Individual and corporate social responsibility. Economica 77, 1-19.
Benson, K.L., Faff, R.W., Rekker, S.A.C., 2014. Corporate social responsibility and CEO compensation revisited: do disaggregation, market stress, gender matter? Journal of Business Economics 72, 84-103
Bouslah, K., Linares-Zegarra, J., M’Zali, B., Scholtens, B., 2018. CEO risk-taking incentives and socially irresponsible activities. The British Accounting Review 50, 76-92.
Cai, Y., Jo, H., Pan, C., 2011. Vice or virtue? The impact of corporate social responsibility on executive compensation. Journal of Business Ethics 104, 159-173.
Callan, S.J., Thomas, J.M., 2011. Executive compensation, corporate social responsibility, and corporate financial performance: a multi-equation framework. Corporate Social Responsibility and Environmental Management 18, 332-351.
Campbell, T.C., Gallmeyer, M., Johnson, S.A, Rutherford, J., Stanley, B.W., 2011. CEO optimism and forced turnover. Journal of Financial Economics 101, 695-712.
Chatterji, A.K., Levine, D.I., Toffel, M.W., 2009. How well do social ratings actually measure corporate social responsibility? Journal of Economics and Management Strategy 18, 125-169. Chen, G., Mack, D.Z., Tang, Y., 2018. The differential effects of CEO narcissism and hubris on corporate social responsibility. Strategic Management Journal 39, 1370-1387.
Chen, G., Qian, C., Shen, R., Tang, Y., 2015. How CEO hubris affects corporate social (ir)responsibility. Strategic Management Journal 36, 1338-1357.
Daniel, K.D., Hirshleifer, D., Subrahmanyam, A., 2001. Overconfidence, arbitrage and equilibrium asset pricing. The Journal of Finance 56, 921-965.
Deckop, J.R., Gupta, S., Merriman, K.K., 2006. The effects of CEO pay structure on corporate social performance. Journal of Management 32, 329-342.
Dimson, E., Karakas, O., Li, X., 2015. Active Ownership. Review of Financial Studies 28, 3225-3268.
Donaldson, T., Preston, L.E., 1995. The stakeholder theory of the corporations: concepts, evidence and implications. Academy of Management Review 20, 65-91.
Ferrell, A., Liang, H., Renneboog, L., 2016. Socially responsible firms. Journal of Financial Economics 122, 585-606.
Franke, G.R., Singhapakdi, A., Vitell, S.J., 1999. Antecedents, consequences and mediating effects of perceived moral intensity and personal moral philosophies. Journal of the Academy of Marketing Science 27, 19–35.
Freeman, R.E., 1984. Strategic management: a stakeholder approach. Journal of Management Studies 29, 131-154.
Gale, D., Hellwig, M., 1985. Incentive-compatibility debt contracts: the one-period problem. Review of Economic Studies 52, 647-663.
Goel, A.M., Thakor, A.M., 2008. Overconfidence, CEO selection, and corporate governance. The Journal of Finance 63, 2737-2784.
Hirshleifer, D., Low, A., Teoh, S.H., 2012. Are overconfident CEOs better innovators? The Journal of Finance 67, 1457-1498.
Hong, B., Li, Z., Minor, D., 2016. Corporate governance and executive compensation for corporate social responsibility. Journal of Business Ethics 136, 199-213.
Hurwicz, L., 1972. On informationally decentralized systems. Decision and Organization, 297-336.
Innes, R., 1990. Limited liability and incentive contracting with ex ante action choices. Journal of Economic Theory 52, 663-692.
Jensen, M., 1986. Agency costs of free cash flow, corporate finance, and takeovers. American Economic Review 76, 323-329.
Jensen, M., Meckling, W., 1976. Theory of the firm: managerial behavior, agency costs, and ownership structure. Journal of Financial Economics 3, 305-360.
Jian, M., Lee, K.W., 2015. CEO compensation and corporate social responsibility. Journal of Multinational Financial Management 29, 46-65.
Karim, K., Lee, E., Suh, S., 2018. Corporate social responsibility and CEO compensation structure. Advances in Accounting 40, 27-41.
Kitzmueller, M., Shimshack, J., 2012. Economic perspectives on corporate social responsibility. Journal of Economic Literature 50, 51-84.
Macbeth, J.D., Merville, L.J., 1979. An empirical examination of the Black-Scholes call option pricing model. Journal of Finance 34, 1173-1186.
Mahoney, L.S., Thorn, L., 2006. An examination of the structure of executive compensation and corporate social responsibility: a Canadian investigation. Journal of Business Ethics 69, 149-162.
Malmendier, U., Tate, G., 2015. Behavioral CEOs: the role of managerial overconfidence. Journal of Economic Perspective 29, 37-60.
Malmendier, U., Tate, G., 2005. CEO overconfidence and corporate investment. Journal of Finance 60, 2661-2700.
McCarthy, S., Oliver, B., Song, S., 2016. Corporate social responsibility and CEO confidence. Journal of Banking and Finance 75, 280-291.
Milbourn, T., 2003. CEO reputation and stock-based compensation. Journal of Financial Economics 68, 233-262.
Moscati, I., Tubaro, P., 2011. Becker random behavior and the as-if defense of rational choice theory in demand analysis. Journal of Economic Methodology 18, 107-128.
Purnamasari, V., Hastuti, T.D., Chrismastuti, A.A., 2015. CSR: the impact on long-term and short-term company performance. International Journal of Humanities and Management Sciences 3, 248-252.
Scheinkman, J.A., Xiong, W., 2003. Overconfidence and speculative bubbles. Journal of Political Economy 111, 1183-1219.
Sun, L., Yu, T.R., 2015. The impact of corporate social responsibility on employee performance and cost. Review of Accounting and Finance 14, 262-284.
Terpstra, D.E., Rozell, E.J., Robinson, R.K., 1993. The influence of personality and demographic variables on ethical decisions related to insider trading. Journal of Psychology 127, 375–389.
Yermack, D., 1996. Higher market valuation of companies with a small board of directors. Journal of Financial Economics 40, 185–211.
Appendix
Table A.1 – Brief overview of the empirical literature
Authors, Year, Title Period Region Measurement
overconfidence
Results
Bouslah, Linares-Zegarra, M'Zali & Scholtens 2017
CEO risk-taking incentives and socially irresponsible activities.
1992-2012 US – Significant positive relationship between CEO risk-taking incentives and the firm's socially irresponsible activities for the pre-crisis period.
– Stronger relationship for firms with higher prior level of socially responsible activities.
– No significant relationship for the post-crisis period. Callan & Thomas
2011
Executive Compensation, corporate social responsibility, and corporate financial
performance: a multi-equation framework
2003-2005 US – Significant positive influence of CSR on CEO pay.
Cai, Jo & Pan 2011
Vice or virtue? The impact of corporate social responsibility on executive compensation
1996-2010 US – Significant negative relationship between (lagged) CSR and total CEO compensation.
– Significant negative relationship between (lagged) CSR and cash compensation.
Chen, Mack & Tang 2018
The differential effects of CEO narcissism and hubris on CSR
2003-2010 US Press coverage – Significant positive relationship between CEO narcissism and CSR. – Significant negative relationship between CEO hubris and CSR.
– Hubristic CEOs are less likely to increase CSR activities when peers with a relatively lower level of CSR increase their engagement.
Chen, Qian, Shen, Tang 2015
How CEO hubris affects corporate social
(ir)responsibility
2001-2010 US Press coverage – Significant negative relationship between CEO hubris and CSR strength. – Significant negative relationship between CEO hubris and slack.
– Significant positive relationship between market competition, CEO hubris and CSR strength.
– Significant negative relationship between uncertainty, CEO hubris and CSR concern.
– Significant negative relationship between market competition, CEO hubris and CSR concern.
Dimson, Karakas, Li 2015
Active Ownership
1999-2009 US – CAR is +1.8% over the year following initial CSR engagement.
– CAR is +4.4% for successful engagements and gradually flatten out after a year.
– No market reaction to successful engagements. Ferrell, Liang, Renneboog
2016
Socially responsible firms
1999-2011 World – No evidence that CSR is associated with ex ante agency concerns. – CSR positively related to legal protection of shareholder rights. – CSR negatively related to controlling shareholders' expropriation of minority shareholders.
Hirshleifer, Low, Hong Teoh 2012
Are overconfident CEOs better innovators?
1993-2003 Holder67
Press coverage – Firms with overconfident CEOs have greater return volatility – Firms with overconfident CEOs invest more in innovation. – Overconfident CEOs achieve only greater innovation in innovative industries.
– Overconfidence of CEOs is related to riskier projects.
– Firms with overconfident CEOs obtain more patents and patent citations. – Firms with overconfident CEOs achieve greater innovative success for given R&D expenditures.
Jian & Lee 2015
CEO compensation and CSR
1992-2011 US – Significant negative relationship between CEO compensation and CSR investment.
– Significant positive relationship between CEO compensation and normal CSR.
– Significant negative relationship between CEO compensation and abnormal CSR.
– CSR positively associated with firm size, operating profitability and MtB ratio.
– CSR negatively associated with volatility in profitability, volatility in stock return, CEO age, board independence, director's stock ownership and
institutional ownership. Mahoney, Thorn
2006
An examination of the structure of executive compensation and CSR: A Canadian investigation Malmendier & Tate
2015
Behavioral CEOs: The Role of Managerial Overconfidence
1996-2012 US Longholder – In this paper they update the 'longholder' measure.
– Market reaction to merger announcement of overconfident CEOs is significantly more negative than the market reaction of announcements of rational CEOs.
– Larger drop in investment following a shock (financial crisis) among overconfident CEOs.
Malmendier & Tate 2005
CEO overconfidence and corporate investment
1980-1994 US Holder67 Longholder Press coverage
– CEOs who excessively hold company stock options do not earn significant abnormal returns over the S&P 500 on average.
Manner 2010
The impact of CEO
characteristics on corporate social performance
2002-2006 US – Bachelor degree in economics is significant negative related to strong/proactive CSP.
– Female CEOs are significant positive related to proactive CSP.
–CSP is negatively related to short-term CEO compensation, but not to long-term compensation.
Table A.2 – Description of variables
Variable Code Definition Units Source Freq
Socially responsible scores
ESG score ESGSCORE The score ranks from 0 to 100. The score measures a company’s ESG performance based on reported data in the public domain Eikon Year
Environmental pillar
score ENVIRONMENT The score ranks from 0 to 100. The score is based on the following categories: resource use, emissions and innovation. Eikon Year
Social pillar score SOCIAL The score ranks from 0 to 100. The score is based on the following categories: workforce, human rights, community and product responsibility. Eikon Year
Governance pillar score GOVERNANCE The score ranks from 0 to 100. The score is based on the following categories: management, shareholders and CSR strategy. Eikon Year
CEO compensation characteristics
CEO stock option STOCKOPTION Value realized from option exercises during the year. The value is based on the
difference between the exercise price and the market price of the stock on the exercise date.
Thousands Execucomp Year
CEO bonus BONUS The dollar value of a bonus earned by the executive officer during the fiscal year. Thousands Execucomp Year
CEO salary SALARY The dollar value of the base salary earned by the executive officer during the fiscal year. Thousands Execucomp Year
CEO total
compensation TOTALCOMP =Stock option + bonus + salary Thousands Execucomp Year
CEO characteristics
CEO overconfidence OVERCONF Dummy that indicates if a CEO is overconfident. Execucomp Year
CEO gender GENDER The gender of the CEO. Execucomp Year
CEO age AGE The age of the CEO. Execucomp Year
CEO tenure TENURE The date the individual left the as CEO minus the date the individual became CEO. Execucomp Year
Firm characteristics
Firm size SIZE =Log (Total Assets) Millions Compustat Year
Tobin's Q Q =Market Value Assets / Book Value Assets Millions Compustat Year
ROA ROA =EBITDA / Book Value Total Assets Millions Compustat Year
Market-to-Book MtB =Market Value Equity / Book Value Equity Millions Compustat Year
Table A.3 – Pearson’s correlation coefficient
The table presents the Pearson’s correlation coefficient for n = 4151
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) ESG score (1) 1 Holder67 (2) -0.195 *** 1 CEO salary (3) 0.303*** -0.0544*** 1 CEO bonus (4) -0.0320* 0.0233 0.292*** 1 CEO stock options (5) -0.0143 0.0606*** 0.0135 0.0338* 1
CEO total comp. (6) -0.0121 0.0608*** 0.0377* 0.0807*** 0.999*** 1
CEO gender (7) -0.129*** -0.0459** -0.0562*** 0.0183 0.00666 0.00679 1 CEO tenure (8) 0.0603 *** 0.0527*** 0.165*** 0.0143 0.00408 0.00659 0.0281 1 CEO age (9) -0.0605*** 0.191*** 0.0468** 0.00464 0.131*** 0.132*** 0.0535*** 0.426*** 1 Firm size (10) 0.472*** -0.191*** 0.514*** 0.119*** 0.00547 0.0166 -0.0824*** 0.0992*** -0.0849*** 1 Tobin's Q (11) -0.0510** 0.140*** -0.149*** -0.0352* 0.126*** 0.123*** -0.0330* -0.0205 0.133*** -0.427*** 1 ROA (12) 0.0205 0.106*** -0.00187 0.0235 0.0593*** 0.0601*** -0.00393 0.0141 0.0489** -0.246*** 0.466*** 1 Market-to-book (13) -0.0170 -0.0109 -0.0295 0.00579 0.00362 0.00352 -0.00234 -0.0531*** -0.112*** 0.00345 -0.0166 -0.0223 1 Firm risk (14) -0.169 *** 0.131*** -0.123*** 0.00658 0.0399* 0.0387* 0.0306* -0.0881*** 0.0278 -0.188*** 0.0283 -0.0910*** 0.00712 1 Board size (15) 0.252 *** -0.138*** 0.309*** 0.132*** -0.0203 -0.0109 -0.0557*** 0.0956*** -0.0738*** 0.448*** -0.198*** -0.0402** -0.00948 -0.127*** 1 ***significant at the 1% level
Table A.4 – Expected impact of independent and control variables
This table presents an overview of the expected impact of the main independent variables and control variables on CSR. I base the expected sign on my expectations and on the findings in the previous studies which focused on the relationship between CEO compensation and CSR.
These studies are presented in the third column.
Variable name Expected sign Source
CEO salary - Mahoney and Thorn (2006)
CEO bonus + Mahoney and Thorn (2006)
CEO stock options + Mahoney and Thorn (2006)
Deckop, Gupta and Merriman (2006) CEO total
compensation
- Benson, Faff and Rekker (2014) Jian and Lee (2015)
CEO tenure + Jian and Lee (2015)
CEO age + Callan and Thomas (2010)
Jian and Lee (2015)
Firm size + Deckop, Gupta and Merriman (2006)
Mahoney and Thorn (2006) Callan and Thomas (2010) Jian and Lee (2015)
ROA + Deckop, Gupta and Merriman (2006)
Mahoney and Thorn (2006) Jian and Lee (2015)
Market-to-book + Jian and Lee (2015)
Firm risk - Mahoney and Thorn (2006)
Callan and Thomas (2010) Jian and Lee (2015)
Table A.5 – Descriptive statistics of pillar scores
This table provides an insight into the descriptive statistics for three socially responsibility scores: the environmental pillar score, the social pillar score, and the governance pillar score. The data relate to firms listed on the S&P500 over the period 2002-2018. There are 4,151 observations for each variable. The definitions of the variables are presented in Table A.2 in the Appendix.
Mean St. dev. Min Max Skewness Kurtosis
Social
responsibility score
ESG score 56.67 16.828 8.0168 95.675 -0.170 2.226
