Prosocial CEOs, corporate policies, and firm value

This paper examines how chief executive officers’ (CEOs’) prosocial tendency influences corporate policies and firm value. We use individuals’ involvement with charitable organizations as a proxy for prosocial tendency. We find that, compared to firms with non-prosocial CEOs, firms with prosocial CEOs have lower executive subordinate turnover, implement more employee-friendly policies, experience higher customer satisfaction, and engage in more socially responsible activities. We also find that firms with prosocial CEOs have higher value and lower risk, partly due to the corporate policies adopted by prosocial CEOs. These results are corroborated when we compare changes in corporate policies and firm value around different types of CEO turnovers: a prosocial CEO replacing a non-prosocial CEO versus other types. Our results thus suggest that prosocial CEOs are more likely to make corporate decisions that benefit others and increase firm value.

1. Our conversations with members of charity boards indicate that charitable organizations look for people who believe in their mission and values to join. Similarly, people are more likely to sacrifice their money, time, and other resources for charitable organizations if they share their mission and values. Therefore, there is likely to be a positive association between individuals’ prosocial tendencies and their involvement with charitable organizations.

2. The IRS defines “charitable” as follows: “the term charitable is used in its generally accepted legal sense and includes relief of the poor, the distressed, or the underprivileged; advancement of religion; advancement of education or science; erection or maintenance of public buildings, monuments, or works; lessening the burdens of government; lessening neighborhood tensions; eliminating prejudice and discrimination; defending human and civil rights secured by law; and combating community deterioration and juvenile delinquency” (https://www.irs.gov/charities-non-profits/charitable-purposes).

3. The position that CEOs hold most often in charitable organizations is board director. According to the IRS’s Governance and Related Topics – 501(c)(3) Organizations (2008), charities should generally not compensate persons for service on the board of directors, except to reimburse the direct expenses of such service. Therefore, it is reasonable to assume that being a board director of a charitable organization constitutes prosocial behavior.

4. We recognize that the extent of prosocial tendency likely exists on a spectrum. For expositional purposes, combined with the fact that we use an indicator variable to identify CEO’s prosocial behavior, we use the terms “prosocial CEOs” and “non-prosocial CEOs” throughout the paper.

5. Our controls for time-varying firm characteristics and year fixed effects also help alleviate the concern that our results are driven by changes in firm-level factors and the overall environment.

6. We acknowledge that we cannot fully rule out concerns over correlated omitted variables. However, as suggested by Glaeser and Guay (2017), compared to studies targeted to address identification issues, broad sample studies are often more generalizable and can use various approaches to narrow omitted variable concerns, making them valuable to the literature.

7. In this study, we focus on the effect of having a prosocial CEO on corporate decision-making and do not intend to disentangle the specific incentives that drive the CEO’s prosocial behavior.

8. Hernandez (2012) defines stewardship as “the extent to which an individual willingly subjugates his or her personal interests to act in protection of others’ long-term welfare.”.

9. We do not have a directional prediction for the association between prosocial leadership and performance-based executive subordinate turnover. On one hand, prosocial CEOs might be more likely to focus on firm goals and display less favoritism, which would lead to more performance-based executive subordinate turnover. On the other hand, firms with prosocial CEOs might have more generous employee policies, which would reduce such turnover.

10. Executives appear to recognize customer satisfaction as an important driver of a firm’s future performance (Chen et al. 2014). However, prior research has documented mixed evidence on the association between customer satisfaction and firms’ financial performance (e.g., Ittner and Larcker 1998; Luo and Homburg 2007). Because of this mixed evidence, it is unclear whether CEOs’ incentives to maximize firm value due to compensation or career concerns complicate the association between CEOs’ prosocial tendencies and customer satisfaction. The same applies to socially responsible activities, given that the evidence on the relation between CSR and firm performance is also mixed (Lys et al. 2015; Waddock and Graves 1997).

11. Consistent with this argument, based on a survey of 80 CEOs, Agle et al. (1999) document a positive univariate association between CEOs’ other-regarding values and the community aspect of CSR performance.

12. In addition, to the extent that a prosocial CEO is less likely to sacrifice firm value to maximize his or her own utility, a reduction in agency problems, such as shirking and asset expropriation, would increase firm value (Jensen and Meckling 1976). Moreover, CEOs’ prosocial activities may enhance firm value by helping CEOs learn new management skills and establish useful networks (for example, by sitting on charities’ boards) (Perry and Peyer 2005).

13. The earliest starting year of our data is 1992, which is when the ExecuComp and KLD databases begin.

14. BoardEx’s data sources include company websites for public, private, and not-for-profit organizations, annual reports and accounts, companies’ public filings, and select news outlets. To the extent that the data is partially based on managers’ own disclosure of their involvement in charitable activities and this disclosure is driven by managers’ desire for self-promotion, it should bias against finding our results, as these managers are less likely to care about employees, customers, or CSR. In addition, we find that the number of a manager’s social activities that are captured by BoardEx is significantly and positively correlated with the duration of BoardEx’s coverage of the manager. As a robustness check, we include the duration of BoardEx’s coverage of the manager as an additional control in all our regression analyses. Our results do not change qualitatively, and our inferences remain the same.

15. We acknowledge that the involvement with charitable activities reflects not only CEOs’ interest in charitable activities but also the opportunities available to them. Our approach will classify CEOs as non-prosocial if they are willing to be involved in charities but do not have any opportunities to do so, adding noise to our measure. However, we control for CEOs’ individual characteristics (e.g., gender) and firm characteristics (e.g., firm size, industry) that might be correlated with charity opportunities. Additionally, this concern would bias against finding our results.

16. The IRS lists all tax-exempt organizations in the Exempt Organizations Business Master File Extract, which can be downloaded at https://www.irs.gov/charities-non-profits/exempt-organizations-business-master-file-extract-eo-bmf. An organization with subsection code 03 and classification code 1 in this file is classified as a “Charitable Organization.” An organization can use up to four classification codes, and thus the classification code in the IRS’s file has four digits. In our main analyses, we consider an organization charitable if it uses only classification code 1 (i.e., classification code 1000). As a robustness test, we also consider an organization charitable if any digit of its classification code is 1 (e.g., classification code 7100), and our results continue to hold. For more information on IRS classifications, see https://www.irs.gov/pub/irs-soi/eo_info.pdf.

17. The non-charitable organizations in our sample include a) tax-exempt entities without a charitable mission (e.g., CPA, veterans associations, business councils) and b) entities that are not classified as nonprofit and thus are not tax-exempt (e.g., golf clubs, wineries). Organizations of type a) are listed in the IRS’s Exempt Organizations Business Master File Extract as tax-exempt but are not classified as charitable by the IRS. Organizations of type b) are not listed in the extract.

18. For example, Penner et al. (2005) state that “these [prosocial] tendencies are relatively stable across a person’s life” (p. 375). Moreover, Batson and Powell (2003), when reviewing prosocial literature, state that dispositional factors, which tend to be stable, can predict higher-cost, nonspontaneous, long-term prosocial behavior more accurately than other types of prosocial behavior. As we have discussed, the positions that CEOs commonly hold in charitable organizations suggest that CEOs’ prosocial behavior that we examine likely has a higher cost and longer-term involvement.

19. We thank Zhenhua Chen for sharing his data on CEOs’ use of personal pronouns in earnings conference calls.

20. Specifically, we estimate a regression as follows: \({Charity}_{i}={{\beta }_{0}+{\beta }_{1}I}_{j,t}+{{\beta }_{2}We}_{j,t}+{{\beta }_{3}They}_{j,t}+\varepsilon\). I is the percentage of first-person-singular pronouns (i.e., I, me, my, mine, and myself) spoken by the CEO during conference calls; We is the percentage of first-person-plural pronouns (i.e., we, us, our, and ourselves), and They is the percentage of third-person pronouns (e.g., she, he, they, etc.). When using conference calls with positive earnings surprises, we find that the coefficient on I is significantly negative, and the coefficients on We and They are significantly positive. When we use conference calls with negative surprises, the coefficient on They is negative and significant, and the coefficients on I and We are insignificant.

21. We use ExecuComp to obtain executive compensation and turn to BoardEx when compensation information is not available on ExecuComp.

22. We validate this assumption using subordinates who are not listed as executives in the subsequent two years in ExecuComp, but whose employment history can be found in BoardEx. We find that 70 percent of these subordinates leave the firm in the year they disappear from the top executives list in ExecuComp, consistent with our assumption.

23. Starting from 1991, KLD rated approximately 650 firms every year, including all firms in the S&P 500 and Domini 400 Social SM Index. During 2001 to 2002, KLD expanded its coverage to the largest 1,000 U.S. companies by market capitalization. Since 2003, it has covered the largest 3,000 U.S. firms based on market capitalization.

24. KLD also provides assessment data on the dimension of corporate governance. However, as Davidson et al. (2019) argue, corporate governance is about the mechanisms that allow shareholders to reward and exert control on managers, whereas CSR deals with social objectives and stakeholders other than shareholders. Therefore, we leave this category out of total KLD scores following Davidson et al. (2019). KLD also assesses firms in the areas of human rights and firearms since 2002. Since these two dimensions are unavailable before 2002, we exclude them in constructing total KLD scores, also following prior literature. In addition, KLD evaluates only negative indicators in exclusionary screen categories, including alcohol, gambling, military contracting, nuclear power, and tobacco. We do not consider these exclusionary categories in calculating KLD scores since they do not pertain to CEOs’ discretionary decisions.

25. Specifically, their characteristic-based ERP uses the average of two variants: ERP from Lewellen (2015) and ERP from Lyle and Wang (2015) and Chattopadhyay et al. (2022). We obtain these two characteristic-based ERPs from https://leesowang2021.github.io/data/. The ERP from Lewellen (2015) is calculated based on a firm’s market capitalization, book-to-market ratio, and cumulative stock return from 12 months to two months prior to the forecast date and is denoted as JLR in the dataset. The ERP from Lyle and Wang (2015) and Chattopadhyay et al. (2022) is calculated based on book-to-market ratio, return on equity, and the mean daily squared returns in the prior month and is denoted as LPV in the dataset. Our CoC variable is the average of these two estimates. Please see Lee et al. (2021) for more details. As a robustness test, we also use these two estimates separately as our cost of capital measures, and our results continue to hold. We thank Charles Lee, Eric So and Charles Wang for making their cost of capital estimates publicly available.

26. For brevity, we do not include all control variables and keep only firm size (LogAT), market-to-book ratio (MTB), and leverage (Leverage) in the correlation table. We conduct tests for multicollinearity for all our regressions and find that no variance inflation factor is greater than 10.

27. Following prior studies (e.g., Cornelli et al. 2013; Guo and Masulis 2015), we report a linear probability model instead of a non-linear logit or probit model because it is easier to implement fixed effects and interpret coefficients. As a robustness check, we also estimate this regression with logit specification and find qualitatively similar results.

28. As a robustness test, we use the number of charitable organizations and the number of non-charitable organizations in which a CEO is involved to construct Charity and NonCharity, respectively. Under this alternative definition, Charity measures the extent of a CEO’s prosocial tendency on a spectrum. Thus, a high value of Charity indicates a highly prosocial CEO, and a low value indicates a less prosocial CEO. Apart from customer satisfaction, all of our results continue to hold.

29. In robustness tests, we control for CEO narcissism and managerial ability for subsamples where these data are available. Following prior studies (e.g., Olsen and Stekelberg 2016; Judd et al. 2017), we use a CEO’s relative cash pay, noncash pay, and the prominence of his or her photograph in the annual report to measure her narcissism. We thank Kari Olsen for sharing his data on CEO narcissism. To proxy for managerial ability, we use the measure developed by Demerjian et al. (2012), which is available at https://peterdemerjian.weebly.com/managerialability.html. We conduct these robustness tests for all our analyses, and our results hold.

30. As a robustness check, for firms with more than one CEO turnover, we keep only the first CEO turnover in the turnover sample, and our results are qualitatively the same.

31. One possible channel through which a CEO influences customer satisfaction is employees. A CEO can improve customer satisfaction indirectly through higher employee retention rates and by having more satisfied employees who serve customers better. We explore this possibility in additional tests. First, we add a firm’s employee-friendly policies, measured by Employee_KLD, as an additional control to Eq. (4). We find that Employee_KLD does not load significantly, and Charity continues to be significantly positive in explaining customer satisfaction. Next, we add an interaction term between Employee_KLD and Charity in the regression. The interaction term does not load significantly, while Charity continues to be positively associated with customer satisfaction. This result suggests that a prosocial CEO improves customer satisfaction through ways other than employee-friendly policies.

32. When we exclude Employee_KLD from Total_KLD, we continue to find significantly positive coefficients on Charity and Post × CharityImprove, indicating that prosocial CEOs are more likely not only to implement employee-friendly policies but also to engage in other CSR activities.

33. As a robustness check, we measure CSR using ratings of firms’ environmental, social and governance (ESG) performance provided by Sustainalytics from 2009 to 2018. We replace KLD score in Eq. (4) with a) a firm’s total ESG score, which is an aggregate of social, environment, and governance scores, or b) a firm’s social score. We continue to find a positive and significant correlation between prosocial CEOs and firm ESG scores.

34. When we regress Total_KLD on ROA or CAR with industry fixed effects and year fixed effects, the coefficient on ROA or CAR is significantly positive, suggesting a positive correlation between a firm’s past performance and its CSR rating. Thus, the negative coefficients on ROA and CAR in Table 3, Panel B may be due to the correlations between these two variables and other control variables such as LogAT and CFO.

35. In untabulated analyses, we also explore the possibility that prosocial CEOs affect firm value through their impact on traditional corporate policies, such as capital expenditure, acquisitions, and R&D expenditure. However, we find that prosocial CEOs are not significantly different than other CEOs when examining these policies, suggesting that the effect of CEOs’ prosocial tendencies does not extend to these outcomes. One possible reason is that these traditional metrics may relate more to CEOs’ ability rather than their prosocial tendencies. Therefore, we do not examine whether prosocial CEOs’ effect on firm value operates through these traditional corporate policies.

36. All control variables in Eq. (5) are included in the regression of proxies for firm risk on Charity, and all control variables in Eq. (3) (Eq. (4)) are included in the regressions of Employee_KLD and Total_KLD (Cus_Satis) on Charity.

37. In the fixed effect research design, CEOs must switch firms for their fixed effects to be estimated. Our sample for customer satisfaction has only 12 CEOs who switch firms. Thus, we did not estimate CEO fixed effects on customer satisfaction.

38. To identify organizations that relate more directly to improving the well-being of others, each of the four authors went over the IRS classification list independently. We use either the intersection or the union of organizations identified by each author to define charitable organizations. Our results still hold.

39. We acknowledge that an individual’s involvement in a charitable organization varies among different organizations, i.e., some organizations require more time commitment than others. These differences may reflect the extent of one’s prosocial tendency. The only way to examine these differences is to know the specific work an individual does at the organization and the number of hours he or she spends there. However, we do not have such data. Our measure Charity, the indicator variable in our main analyses or the continuous variable in the robustness test in footnote 28, treats all the charitable organizations the same. That is, we consider a CEO involved in a charity that requires substantial time commitment as having the same prosocial tendency as a CEO involved in a charity that may require less commitment. This approach adds noise to our measure.

40. We find similar results when we assume that the new CEO joined the firm three years before the actual turnover date.

We would like to thank two anonymous reviewers, Shane Johnson, Byungki Kim, Charles Lee, Hai Lu, Sara Malik, Dawn Matsumoto, Sarah McVay, Donald Moser, Shawn Thomas, Xing Su, Liandong Zhang, and workshop and conference participants at Emory University, Florida International University, George Washington University, Syracuse University, the University of Calgary, the University of Delaware, the University of Nebraska-Lincoln, the University of Pittsburgh, the University of Washington, the 2020 USTC-UW Conference on Fintech and Management Innovation, the 2021 Hawai’i Accounting Research Conference, the 2021 Chinese Accounting Professors’ Association of North America Annual Research Conference, the 2021 MIT Asia Conference in Accounting, and the 2021 American Accounting Association Annual Meeting for their valuable comments and suggestions. Ge would like to thank the Moss Adams Professorship at the University of Washington for financial support. Any errors or omissions are our responsibility.

Authors and Affiliations

1. University of Pittsburgh, Pittsburgh, PA, 15260, USA

   Mei Feng

2. University of Washington, Seattle, WA, 98195, USA

   Weili Ge

3. California State University, Fullerton, Fullerton, CA, 92834, USA

   Zhejia Ling

4. Singapore Management University, Singapore, 178900, Singapore

   Wei Ting Loh

Corresponding author

Correspondence to Weili Ge.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendix 1

Appendix 2

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions