Economics Letters 119 (2013) 344–347
Contents lists available at SciVerse ScienceDirect
Economics Letters journal homepage: www.elsevier.com/locate/ecolet
How do powerful CEOs view corporate social responsibility (CSR)? An empirical note P. Jiraporn a,∗,1 , P. Chintrakarn b a
Pennsylvania State University, School of Graduate Professional Studies, Malvern, PA, USA
b
Mahidol University International College (MUIC), Nakhon Pathom, Thailand
highlights • We use agency theory to explain how powerful CEOs view corporate social responsibility (CSR). • Strong CEO power leads to more investments in CSR. • However, when the CEO is powerful to a certain point, he reduces CSR investments significantly.
article
info
Article history: Received 12 February 2013 Received in revised form 5 March 2013 Accepted 12 March 2013 Available online 26 March 2013 JEL classification: G30 G34
abstract We explore how powerful CEOs view investments in corporate social responsibility (CSR). The agency view suggests that CEOs invest in CSR to enhance their own private benefits. On the contrary, the conflict resolution view argues that CSR investments are made to resolve the conflicts among various stakeholders. Using Bebchuk et al. (2011) CEO Pay Slice (CPS) to measure CEO power, we show that the association between CEO power and CSR is non-monotonic. When the CEO is relatively less powerful, an increase in CEO power leads to more CSR engagement. However, as the CEO becomes substantially more powerful, he is more entrenched and no longer invests more in CSR. In fact, when CEO power goes beyond a certain threshold, more powerful CEOs significantly reduce CSR investments. © 2013 Elsevier B.V. All rights reserved.
Keywords: CEO power Corporate social responsibility CSR Agency theory Agency conflict
1. Introduction There has been a fierce debate in the literature about the role of corporate social responsibility (CSR). On the one hand, the agency view posits that CSR brings about private benefits to managers but do not necessarily enhance shareholders’ wealth. Hence, managers tend to over-invest in CSR (Bernea and Rubin, 2010). Managers enjoy publicity and gain media exposure when they engage in CSR activities, contributing to their personal reputation building. On the contrary, the conflict resolution view argues that managers engage in CSR to resolve the conflicts among various stakeholders.
∗
Corresponding author. Tel.: +1 4847533655. E-mail addresses:
[email protected],
[email protected] (P. Jiraporn).
1 Part of this research was conducted while the first author served as Visiting Associate Professor of Finance at The National Institute of Development Administration (NIDA) in Bangkok, Thailand. 0165-1765/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.econlet.2013.03.026
Different groups of stakeholders have divergent objectives. For instance, employees may demand large employee benefits, while shareholders view these benefits as reducing their investment returns. By engaging in CSR, managers alleviate the tension between the various groups of stakeholders. To distinguish between the agency view and the conflict resolution view, we explore the effect of CEO power on CSR. The agency view suggests that CSR investments are made to enhance the private benefits of managers. Therefore, when the CEO is more powerful, according to this view, he is expected to engage more in CSR (to enhance his own private benefits). On the other hand, if the conflict resolution is correct (CSR investments are made not for personal benefits), then the extent of CSR engagement is not expected to vary with CEO power. We measure CEO power by using Bebchuk et al.’s (2011) CEO pay slice (CPS), which represents the CEO’s total compensation as a fraction of the combined total compensation of the top five executives (including the CEO) in a given company. Our
P. Jiraporn, P. Chintrakarn / Economics Letters 119 (2013) 344–347 Table 1 Summary statistics. CEO pay slice (CPS) is as defined by Bebchuk et al. (2011). ROA is net income divided by total assets. SGA is the selling, general, and administrative expenses.
CSR score CEO pay slice (CPS) Total assets ROA Total debt/Total assets Dividend/Total assets R&D/Total assets SGA/Total assets
Mean
Median
S.D.
25th
75th
0.188 0.322 12,660 5.76% 0.216 0.013 0.023 0.190
0.000 0.330 2,533 5.43% 0.204 0.006 0.000 0.141
2.501 0.131 56,005 7.88% 0.173 0.036 0.047 0.195
−1.000
1.000 0.419 7901 9.28% 0.327 0.018 0.027 0.290
0.209 829 2.29% 0.064 0.000 0.000 0.021
results reveal that, when the CEO is relatively less powerful, an increase in CEO power raises CSR engagement. However, as CEO power reaches a certain point, a further increase in CEO power results in a significant decline in CSR engagement. In other words, the relationship is non-monotonic. Because we find a significant relationship between CEO power and CSR, the evidence appears to be more consistent with the agency view. Although the agency view predicts only a positive relationship (more CEO power, higher CSR), it seems that, when CEO power is sufficiently potent, he is so entrenched that any further investments in CSR would not enhance his private benefits any more (perhaps, because he can expropriate in other ways when he is so deeply entrenched). As a result, CSR investments decline significantly when the CEO is highly powerful. Apparently, deeply entrenched CEOs view CSR as depriving them of the free cash flow they could otherwise exploit. 2. Sample selection and data description 2.1. Sample selection We obtain CSR ratings from Kinder, Lydenberg, and Domini’s (KLD’s) database. We calculate CEO pay slice (CPS), using the compensation data from the EXECUCOMP database. Data on firm characteristics are obtained from COMPUSTAT. The final sample consists of 4489 firm-year observations from 1995 to 2007, representing 1370 unique firms (unbalanced across the years). 2.2. Measuring CEO power using CEO pay slice (CPS) One way to capture CEO power more objectively is to examine his relative compensation among top executives. Bebchuk et al. (2011) argue that the CEO’s pay slice (CPS) captures the relative significance of the CEO in terms of abilities, contribution, or power. As such, CPS provides a useful proxy for the relative centrality of the CEO in the top management team. This particular measure of CEO power is especially interesting because Bebchuk et al. (2011) find that CPS has strong explanatory power for a rich set of critical corporate outcomes, including firm value as measured by Tobin’s q, accounting profitability, and stock market reactions to acquisition announcements. We follow their approach and define CPS (i.e., CEO’s pay slice) as the CEO’s total compensation as a fraction of the combined total compensation of the top five executives (including the CEO) in a given company. Total compensation includes salary, bonus, other annual pay, long-term incentive payouts, the total value of restricted stock granted that year, the Black–Scholes value of stock options granted that year, and all other total compensation (EXECUCOMP item TDC1). 2.3. Measuring CSR To measure CSR, we use the KLD score, which has been widely used in the literature. KLD includes strength ratings and concern ratings for 13 dimensions; community, diversity, corporate gov-
345
ernance, employee relations, environment, human rights, product, alcohol, gambling, firearms, military, tobacco, and nuclear power. KLD assigns strengths and concerns in the first 7 dimensions, whereas the final 6 dimensions are just exclusionary screens and firms can only register concerns in those categories. For instance, a company can receive credit for a strong environment policy at the same time a concern is registered for its environment record. We do not include the exclusionary screen as part of the total CSR score. The total of the strengths minus the concerns is the composite CSR score (Goss and Roberts, 2011). 2.4. Control variables For control variables, we include firm size (total assets), profitability (ROA), leverage (total debt/total assets), R&D spending (R&D/total assets), dividend payouts. We also include the ratio of the selling, general, and administrative (SGA) expenses to total assets. SGA is largely subject to managerial discretion and may represent the extent of the agency conflict. These control variables are selected based on the recent studies in this area of research (Jo and Harjoto, 2011; Bernea and Rubin, 2010; Borghesi et al., 2012). To control for possible variable across industries and time, we also include industry and year dummies.2 3. Results Table 1 shows the summary statistics. The CSR score averages 0.188 with a standard deviation of 2.501. The average CEO pay slice (CPS) is 0.322 with a standard deviation of 0.131. Our average CPS is similar to that in Bebchuk et al. (2011). Table 2 displays the regression results. We report t-statistics using robust standard errors clustered at the firm level. Model 1 includes CPS and all control variables. The coefficient of CPS is not significant. Model 2 adds the quadratic term of CPS. The coefficient of CPS is now significantly positive, whereas the coefficient of the quadratic term is significantly negative. Thus, it appears that the association between CEO power and CSR is non-monotonic. At the lower range of CEO power, there is a positive relationship. At the higher range of CEO power, the relationship turns negative. The result corroborates the agency hypothesis when CEO power is relatively low. More powerful CEOs promote CSR to increase their own private benefits. When the CEO commands much more power, however, the agency view may not appear to be supported (because the relationship turns negative). We conjecture that, at the higher range of CEO power, powerful CEOs are so entrenched that the additional private benefits gained through CSR are probably trivial compared to what they could gain through other means of expropriation. If this is the case, then the CEO is more likely to view CSR investments as taking away the free cash flow that he could otherwise exploit. As a consequence, CSR investments diminish significantly. Alternatively, the evidence can be interpreted this way. When CEO power is really powerful and entrenched, he can afford not to care, not only about shareholders, but also about other stakeholders. None of these stakeholders would be powerful enough to remove the CEO. Therefore, it is less necessary for the powerful CEO to please them through CSR engagement. Fig. 1 shows a graphical representation of the relationship between CEO power and CSR engagement. The point where the relationship turns from positive to negative is when CPS is approximately 0.32. This point is obtained by differentiating the CSR function (Model 2 Table 2) with respect to CPS and solving for the maximum point when the first derivative is set to zero.
2 The industry dummies are based on the first two digits of the SIC codes.
346
P. Jiraporn, P. Chintrakarn / Economics Letters 119 (2013) 344–347
Table 2 Regression analysis of corporate social responsibility (CSR) on CEO Power. CEO pay slice (CPS) is as defined by Bebchuk et al. (2011). ROA is net income divided by total assets. SGA is the selling, general, and administrative expenses. Dependent variable
Model 1 (t-statistics) OLS CSR
Model 2 (t-statistics) OLS CSR
Model 3 (t-statistics) Fixed-effects CSR
Model 4 (t-statistics) First-stage CEO pay slice
Model 5 (t-statistics) Second-stage CSR
Constant
−5.199*** (−5.76) −0.213 (−0.49)
−6.147*** (−6.36)
−1.828 (−2.26)
0.042 (1.02)
−6.792*** (−6.14)
CEO pay slice
(CEO pay slice)2 Industry-average CEO pay slice CEO pay slice (Instrumented) (CEO pay slice)2 (Instrumented) Ln (total assets) ROA Total debt/Total assets Dividend/Total assets R&D/Total assets SGA/Total assets
5.015*** (3.04) −7.816*** (−3.52)
0.455*** (6.38) 3.495*** (5.33) −0.922** (−2.09) 1.383 (1.47) 5.503*** (3.54) 1.437*** (2.49)
0.472*** (6.60) 3.513*** (5.36) −0.923** (−2.11) 1.377 (1.48) 5.505*** (3.54) 1.495*** (2.58)
Yes Yes 20.64%
Yes Yes 20.97%
Industry-average CSR Year-average CSR Year dummies Industry dummies R2 or Pseudo R2 Shea’s Partial R2 * ** ***
2.536** (2.24) −3.420** (−2.17)
0.172* (1.85) 0.754* (1.76) 0.405 (1.20) −2.738* (−1.89) 0.266 (0.19) 0.408 (0.72) 0.599*** (17.81) −0.215*** (−2.55) No No 73.26%
0.989*** (21.13)
−0.006** (−2.50) −0.020 (−0.59)
0.055*** (2.76) −0.020 (−0.31) 0.034 (0.43) 0.024 (1.01)
Yes Yes 17.86% 45.29%
10.172*** (2.20) −15.286*** (−2.07) 0.461*** (6.40) 3.520*** (5.38) −0.906** (−2.01) 1.368 (1.44) 5.591*** (3.60) 1.418** (2.45)
Yes Yes 20.73%
Indicates statistical significance at the 10%. Indicates statistical significance at the 5%. Indicates statistical significance at the 1%.
This point is close to the mean and the median of CPS (0.322 and 0.330 respectively). Therefore, when CEO power is below the mean or median, an increase in CEO power leads to more CSR engagement. When CEO power exceeds this threshold, however, any further increase would result in a decline in CSR. To determine the marginal effect of CPS on CSR, we find that the first derivative of CSR with respect to CPS is 5.015 +[2 ×(−7.816)] CPS. Hence, the marginal effect of CPS on CSR is a function of CPS. The magnitude of the marginal effect on CSR ranges from 1.748 (evaluated at the 25th percentile of CPS) to 0 (evaluated at the median) and −1.535 (evaluated to the 75th percentile).3 We execute further tests to address endogeneity. First, it is possible that both CEO power and CSR are influenced by unobservable firm characteristics, thus creating a spurious relationship. We run a fixed-effects regression, which controls for time-invariant unobservable characteristics. The fixed-effects result is shown in Model 3 in Table 2.4 CPS retains a positive and significant coefficient, while the quadratic term of CPS still exhibits a negative and significant coefficient. The fixed-effects regression essentially removes the cross-sectional variation and concentrates only on the variation over time. The fixed-effects result suggests that, within
3 It is worth noting that the 75th percentile value for CPS is 0.419. So, there are very few firms on the downward sloping part of the graph. It appears that powerful CEOs with high CPS simply stop further investments in CSR, rather than significantly reducing them. 4 Because industry dummies are time-invariant, we cannot include them in the fixed-effects regression. Instead, we compute the average CSR score for each industry and include it in the regression to control for possible industry effects.
firms, when the CEO acquires slightly more power, he invests more in CSR. However, when the CEO gains substantially more power, he tends to reduce CSR investments. Because the fixed-effects result is consistent with the OLS result, it does not appear that our conclusion is affected by endogeneity due to the omitted-variable bias. Second, we explore possible reverse causality. In particular, we estimate a two-stage least squares (2SLS) regression using industry-average CPS as an instrument.5 The logic is that CSR engagement in a particular firm might influence its CEO power. However, it is extremely unlikely that the CSR of a given firm is related to industry-wide average CPS because there are many firms in each industry (the average number of firms in each industry is 23.7). The 2SLS results are shown in Models 4 and 5. Model 4 is the first-stage regression where CPS is the dependent variable. The coefficient of industry-average CPS is positive and highly significant, as expected. Shea’s partial R2 is 45.29%, suggesting that the instrument is not weak. Model 5 is the second-stage regression with CPS instrumented from the first stage.6 CPS exhibits a significantly positive coefficient, whereas the quadratic term of CPS shows a significantly negative coefficient. Thus, the 2SLS results are consistent with the OLS and fixed-effects results. Our conclusion seems to be robust to endogeneity. To further control for endogeneity, we identify CPS in the earliest year for each firm in the sample. Then, we replace CPS in any given year with CPS in the
5 Industry classification is based on the first two digits of the SIC codes. 6 For the quadratic term of CPS in the second stage, we square the predicted value of CPS instrumented from the first stage.
P. Jiraporn, P. Chintrakarn / Economics Letters 119 (2013) 344–347
347
Table A.1 Two-stage least squares (2SLS) estimation based on CPS in the earliest year. Dependent variable
Model 1 (t-statistics) First-stage CEO pay slice
Model 2 (t-statistics) Second-stage CSR
Constant
−0.004 (−0.14)
−4.081*** (−7.77)
Yes 50.60%
3.745* (1.63) −5.490* (−1.68) Yes 27.00%
CEO pay slice (earliest year)
0.662*** (26.55)
CEO pay slice (instrumented) (CEO pay slice)2 (instrumented) Control variables included R2 or Pseudo R2 * ***
Fig. 1. CEO pay slice (CPS) versus CSR. The horizontal axis represents CEO pay slice. The vertical axis represents the predicted value of the CSR score. The graph is based on the regression equation in Model 2 Table 2. Except for CSR and CPS, all variables are held at their means.
earliest year. The rationale is that CPS in the earliest year could not have resulted from CSR in any subsequent years, making reverse causality unlikely. We then employ the two-stage least squares (2SLS) estimation, using CPS in the earliest year as our instrumental variable. The 2SLS results based on earliest CPS remain consistent, confirming that reverse causality is unlikely.7
Indicates statistical significance at the 10%. Indicates statistical significance at the 1%.
vate benefits to the CEO. However, when the CEO consolidates his power to a certain point, he is so entrenched and invulnerable that he no longer views CSR favorably and therefore reduces CSR investment significantly. Deeply entrenched CEOs can probably expropriate through other means and do not view CSR as necessary. Highly powerful CEOs seem to view CSR investments as depriving them of the free cash flow they could otherwise exploit. Appendix See Table A.1.
4. Conclusion Our study sheds light on the motivation behind CSR investments. The evidence shows that CEO power has a significant influence on CSR engagement. In particular, an increase in CEO power leads to more CSR engagement. Nevertheless, when CEO power reaches a certain threshold, the CEO reduces CSR investments significantly. The evidence appears to be consistent with the agency interpretation. As the CEO becomes more powerful, he increases CSR investments, suggesting that CSR confers at least some pri-
7 The results are shown in Appendix.
References Bebchuk, L., Cremers, M., Peyer, U., 2011. The CEO pay slice. Journal of Financial Economics 199–221. Bernea, A., Rubin, A., 2010. Corporate social responsibility as a conflict between shareholders. Journal of Business Ethics 71–86. Borghesi, R., Houston, J., Naranjo, H., 2012. What motivates corporate managers to make socially responsible investments? Working paper, University of Florida. Goss, A., Roberts, G.S., 2011. The impact of corporate social responsibility on the cost of bank loans. Journal of Banking and Finance 1794–1810. Jo, H., Harjoto, M.A., 2011. Corporate governance and firm value: the impact of corporate social responsibility. Journal of Business Ethics 351–383.
