Corporate Social Responsibility, Firm Value and External Corporate Governance

Transcription

1 Corporate Social Responsibility, Firm Value and External Corporate Governance Bonnie Buchanan Howard Bosanko Professor of Economics and Finance Department of Finance Albers School of Business and Economics Seattle University th Avenue P.O. Box Seattle. WA Ph: (206) Fax: (206) Cathy Xuying Cao, CFA Department of Finance Albers School of Business and Economics Seattle University th Avenue P.O. Box Seattle. WA Ph: (206) Fax: (206) Chongyang Chen, CFA Department of Finance School of Business Pacific Lutheran University Seattle. WA Ph: (253) Keywords: Corporate social responsibility; ESG; financial crisis; institutional investors; Shareholder value; corporate governance. JEL classification: D22; G34; M14 This version: September 26, DO NOT QUOTE WITHOUT PERMISSION Page 1 of 45

2 Abstract This paper examines how Corporate Social Responsibility (CSR) affects firm value before, during and after the 2008 global financial crisis. We measure CSR practice using Bloomberg ESG scores. We treat the 2008 financial crisis as an exogenous shock to firms economic environment. We find that when the financial crisis occurs, firms with CSR policies experience a larger decrease in firm value than comparable firms. We show that such a big discount in firm value exists only among firms with ESG scores and with high external corporate governance. In addition, we show that firms with ESG scores do not have any advantage in recovery of firm value from financial crisis, regardless of their level of external governance. The key contribution of this paper is to show that the impact of CSR on firm value changes over time. Page 2 of 45

3 Being socially responsible is both a moral and business imperative. Any company that ignores its obligation to be a good corporate citizen risks undermining its reputation and jeopardizing its profitability over the long run Jay Hooley, CEO and Chairman, State Street Corporation October 24, 2014 Corporate social responsibility is an ill-defined- and incompletely defined concept Baron (2001) 1. Introduction In an increasingly financialized world, Corporate Socially Responsibility (CSR) has gained prominence due to the substantial increase of socially responsible investment (SRI) funds. For example, US-domiciled assets under management using SRI strategies expanded from $3.74 trillion in 2012 to $6.57 trillion in 2014 a 76 percent increase (US SIF, 2014) 2. Moreover, an increasing number of Fortune 1000 companies have issued CSR reports (Ghoul et. al, 2011; Prakash et al., 2016). More institutional investors have committed to the Principle for Responsible Management (PRME). Finally, there has been a surge in shareholder proposals targeting social and environmental issues (Buchanan et. al, 2012) and this trend has continued since 2006 according to a 2015 Proxymonitor Report. Despite the increased interest in CSR related issues, when it actually comes to taking action there is hesitancy. A 2013 triennial UNGC Survey reports that 37 percent of those surveyed said a lack of a clear link between ESG 3 and firm value deterred them from taking action, although 93 percent of respondents regarded ESG issues essential to the success of their 1 Companies have no choice but to be good corporate citizens, Wall Street Journal. October 24, This now accounts for more than one out of every six dollars under professional management in the United States. 3 ESG refers to Environmental, Social and Governance score. It jointly captures CSR actions and corporate governance actions (Starks, 2009). Page 3 of 45

4 business. Similarly, according to a 2015 CFA Institute survey report 4, when asked what if anything would cause them to begin considering ESG issues in their investment analysis, 48 percent of respondents stated they would do so if there was a proven link between ESG and firm financial performance. Is there a significant link between CSR and financial performance or, ultimately, firm value? How does Corporate Social Responsibility (CSR), jointly with external corporate governance, affect firm value? Existing studies remain inconclusive on the effect of corporate social responsibility (CSR) on firm value or financial performance 5. On the one hand, high CSR practices bring benefits to companies. First, according to the conflict resolution hypothesis (Harjoto and Jo (2012, 2011) and Jo and Harjoto (2012)), CSR activities can be an effective extension of corporate governance mechanisms to resolve conflicts between managers and other stakeholders. Thus one would expect a positive relationship between CSR and firm value. Secondly, a substantial theoretical literature suggests that CSR activities can improve firm reputation, increase firm competitive advantage, enhance profitability and, hence, firm value (see, among others, Turban and Greening (1997), Jensen (2002), Scherer et al (2006), Cespa and Cestone (2007), Jo and Harjoto (2011, 2012), Servaes and Tamayo (2013). On the other hand, implementing CSR practices can be costly, especially when facing adverse financial shocks. Firm CSR practice requires significant financial commitment, which can increase a firm s financial risk during bad times. Referred to as the overinvestment hypothesis (Harjoto and Jo, 2012, 2011), an overinvestment in CSR initiatives can be a waste of valuable resources and hence, value destroying to the firm (Barnea and Rubin (2010)). Thus the overinvestment hypothesis predicts a negative relation between CSR and firm value. In Excellent literature reviews may be found in Berman et al (1999), Margolis and Walsh (2003), Margolis et al. (2007), Renneboog et al (2008), Friede et al. (2015) and Servaes and Tamayo (2017). Page 4 of 45

5 addition, CSR investment can amplify the conflicts between shareholders and thus induce potential agency costs. For example, Barnea and Rubin (2010) suggest that a firm s affiliated insiders such as managers and board members tend to over-invest in CSR for their private benefits such as improving their reputation as good global citizens, especially when the insiders bear little of the cost of doing so. It is still an empirical question whether the benefits from CSR overweigh its costs around a financial crisis. Prior empirical studies present mixed evidence on the effect of CSR on firm value. For example, Guenster et al. (2010) provide evidence suggesting that corporate environmental performance is positively related to firm value. Similarly, Jiao (2010) shows that corporate social performance is associated with a positive valuation effect. In contrast, Brammer et al. (2006) find that firms with higher social performance scores realize lower returns. However, Hamilton et al. (1993) find that the excess returns of socially responsible mutual funds do not differ statistically from those of conventional mutual funds, and Nelling and Webb (2009) find no evidence that CSR activities affect financial performance. The lack of consensus about the link between CSR and firm value can be attributed to the difficulty of dealing with the endogenous nature of firm performance and CSR activities. We propose that the conflicting findings are due to the implicit assumption that the firm s decision to carry out CSR activities is exogenous and random. In fact, as shown in Hong, Kubik and Scheinkman (2012), firms do good only when they do well in the sense of having financially slack resources. Specifically, a profitable firm is more likely to have resources to implement a corporate social responsibility program; meanwhile, corporate social responsibility programs can increase a firm s product awareness and investor recognition, which can affect firm value and the cost of equity. Hong, Kubik and Scheinkman (2012) model a firm s optimal choices of capital Page 5 of 45

6 and goodness subject to financial constraints. They predict that less financially constrained firms spend more on goodness. Failing to account for the endogenous choice of taking CSR activities can contaminate the pricing effect with the selection effect of CSR activities, and thus lead to inconsistent empirical estimation. This study offers a novel empirical approach that allows us to overcome this challenge and shed considerable light on the impact of CSR activities on the firm value. Specifically, we employ the difference-in-difference (DID) methodology to test for differences in firm value to the level of CSR. DID estimators usually complement natural or quasi-experiments created by sharp policy changes or exogenous shocks (Roberts and Whited, 2012). Motivated by existing studies (Lemmon and Lins (2003), Bharath, Jayaraman, and Nagar (2013), Lins, Volpin, and Wagner (2013), and Duchin, Ozbas and Sensoy (2010)), we use the financial crisis to disentangle the potential endogeneity issues that arise because the variable of interest, CSR activities, and firm value may all be jointly determined. During a financial crisis, firms face limited resources and uncertainty about the duration of scarce financial liquidity, which magnifies both the benefits and costs of maintaining high CSR activities. In this paper we focus on CSR information that is available to the public (i.e., regulators, investors, customers, industry competitors, and etc.). Specifically, we use Bloomberg s ESG scores to capture the observable CSR activities. Bloomberg constructs ESG scores based on various public sources such as company CSR reports, annual filings, firm websites, media coverage and interview, and surveys. We ask how CSR activities affect the change in firm values around the 2008 financial crisis. In addition, we examine one particular external corporate governance mechanism: institutional ownership. CSR practices trigger investors concerns about agency problems. Such Page 6 of 45

7 agency problems tend to become more serious during an economy-wide financial distress than during a business boom (Rajan and Zingales (1998), and Johnson et al. (2000)). External corporate governance can help mitigate agency problems (Shleifer and Vishny (1997), Gompers, Ishii and Metrick (2003) by improving the efficient usage of firm resources and decreasing managers discretion and expropriation of rents. Therefore, the presence of efficient external governance can protect shareholders against value-destroying over-investment in CSR activities. However, prior studies show that excess governance controls can have a negative impact on firm value (See, among others, Claessens et al., 2002; La Porta et al., 2002; Lemmon and Lins, 2003). In this paper, we empirically examine the differential impact of CSR practice on firm value across firms with different institutional ownership composition before, during, and after the financial crisis. We find that although CSR activities are positively related to firm value (Tobin s Q) before the financial crisis, firms with ESG scores experience a larger decrease in firm value than other comparable firms during financial crisis. In addition, CSR firms with different level of influential institutional ownership perform differently. We find that firm value of CSR firms with high influential ownership decreases more than that of other firms with the same level of ownership; while among firms with low influential ownership, the decrease in magnitude of firm value of CSR firms is not significantly different from that of other firms. The findings are consistent with the notion that during a financial crisis, influential institutions tend to actively engage in CSR firms management, which can decrease the level of employee perceived managerial integrity and lead to lower firm performance and firm value. Finally, we find that all firms experience improvements in their firm value after the financial crisis. However, our evidence suggests that firm value of CSR firms does not recover as Page 7 of 45

8 quickly as that of other firms after the crisis, regardless of their financial flexibility and influential institutional ownership. Our study is related to several studies that examine the relation between CSR and firm value (Margolis and Walsh, 2003; Harjoto and Jo, 2011; Jo and Harjoto, 2012, 2011; Servaes and Tamayo, 2013; Flammer, 2015). Different from prior studies, we focus on the effect of CSR on firm value around the financial crisis to mitigate endogeneity concerns. Our evidence suggests that the previously mixed results documented in the literature can be due to the different sample periods used in the tests. Our test design not only allows us to study the effect of CSR on firm value at a particular point in time, but also helps us examine the variation in such effect over economic cycles. Our results indicate that firms with high CSR experience a larger decrease in firm value than other comparable firms when a financial crisis occurs. Moreover, we show that CSR practices do not have significant impact on the recovery of firm value right after financial crisis. Related to Lins, Servaes, and Tamayo (2017), we examine the CSR effect around the financial crisis. However, our study is different from Lins, Servaes, and Tamayo (2017) in the following aspects. First, Lins, Servaes, and Tamayo (2017) examine how CSR intensity affects stock returns during the 2008 financial crisis. We ask a different research question: how does the interaction between CSR activities and external corporate governance affects firm value during financial crisis. Secondly, Lins, Servaes, and Tamayo (2017) find that firms with higher CSR intensity have higher stock returns during the crisis. In contrast, we focus on firm value, proxied by Tobin s Q as in Bharath, Jayaraman, and Nagar (2013), instead of stock returns. We show that although the CSR activity is positively related to firm value (Tobin s Q) before financial crisis, the firm values of the CSR Firms decrease more than those of other firms during financial crisis. Page 8 of 45

9 Thirdly, Lins, Servaes, and Tamayo (2017) find that internal corporate governance is not significantly related to the abnormal returns during the financial crisis. In addition, they show that internal corporate governance does not affect the impact of CSR on stock returns. Different from Lins, Servaes, and Tamayo (2017), our variable of interest is the external governance or the interaction between external governance and CSR activities. We find that the effect of CSR intensity on firm value varies with the different degree of external governance, proxied by the level of institutional ownership. Our paper also adds new evidence to the literature by showing that there are different CSR effects on firm value among firms with different composition of external corporate governance. Harjoto and Jo (2011) investigate how the internal and external corporate governance affect the value of firms engaging in CSR activities. Our paper is different from Harjoto and Jo (2011) in the following ways. First, different from Harjoto and Jo (2011), we use ESG score as a proxy for a firms CSR intensity. Our measure not only reflects environmental and social categories but also incorporates the internal governance issues. Secondly, instead of a cross-sectional comparison of the level of firm value between firms with and without CSR activities, we investigate the changes in firm values for the firms with CSR activities before, during, and after the financial crisis, controlling for the changes in firm values of other matching firms over the same period of time. Different from Harjoto and Jo (2011), we show that CSR engagement negatively influences firm value measured by Tobin s q. Lastly, Harjoto and Jo (2011) examine the sole effect of external corporate governance and find that blockholders ownership, and institutional ownership has a relatively weaker effect on firm value. In contrast, we focus on the joint effect between external corporate governance and CSR engagement on firm value. We show that the relation between CSR activities and firm values depends on the strength Page 9 of 45

10 of external corporate governance. Our evidence suggests that compared with CSR Firms with strong external corporate governance, CSR Firms with weak external corporate governance experience a persistently larger discount in firm values after the onset of the financial crisis but have quicker recovery in firm value after the crisis. Finally, our paper questions the contribution of CSR on firm value around the financial crisis. Are there other factors we should consider when investing in CSR firms, factors such as external corporate governance and market conditions? This potentially has a number of implications for SRI and asset pricing as well. The rest of the paper is organized as follows. Section 2 details the theoretical motivation and empirical methodology. Section 3 presents discusses data sources and sample characteristics. Section 4 reports the results from the empirical tests. Section 5 concludes. The Appendix provides definitions for the variables used in our tests. 2. Motivation and Empirical Methodology We study how Corporate Social Responsibility (CSR), jointly with external corporate governance, affects firm value before, during and after the 2008 global financial crisis. Existing studies remain inconclusive about the effect of corporate social responsibility (CSR) on both firm value and financial performance. According to the conflict resolution hypothesis (Harjoto and Jo (2011) and Jo and Harjoto (2012, 2011)), one would expect a positive relationship between CSR and firm value. Freeman s (1984) stakeholder theory states that firms can use CSR to mitigate conflicts between managers and non-investing stakeholders. Thus the CSR activities will enhance firm value through reduced conflicts of interests. Page 10 of 45

11 By contrast, the overinvestment hypothesis suggests a negative relation between CSR and firm value (Barnea and Rubin (2010)). Barnea and Rubin (2010) propose that overly invested CSR practices caused by agency problem can destroy firm value. Additionally, Di Giuli and Kostovetsky (2014) show a negative relation between future stock price returns and CSR metrics. Krüger (2015) finds that there are substantial costs associated with social irresponsibility. A major obstacle limiting empirical progress on the topic is the difficulty in dealing with the endogenous nature of firm performance and CSR activities. We argue that the conflicting findings are due to the implicit assumption that the firm s decision to adopt corporate social responsibility practices is exogenous and random. To address this question, we apply the difference-in-difference (DID) methodology to test for differences in firm value for the level of CSR. DID estimators usually complement natural or quasi-experiments created by sharp policy changes or exogenous shocks (Roberts and Whited, 2012). An advantage of DID estimation arises from its potential to circumvent many of the endogeneity problems that typically appears when making comparisons between heterogeneous individual firms (Meyer, 1995). Tobin (1958) first acknowledges the endogeneity problem and shows that if this is not taken into consideration in the estimation procedure, an ordinary least-square estimation (OLS) will produce biased parameter estimates. Motivated by prior literature, we use the 2008 financial crisis to mitigate concerns about the potential endogeneity between CSR activities and firm performance. We investigate changes in firm value for the CSR firms before, during, and after the financial crisis, controlling for the changes in firm values of other matching firms over the same period of time. We use comparable Page 11 of 45

12 firms without any CSR activities as a control group to help difference out possible confounding factors and isolate the effect of CSR practices on firm value. As discussed above, CSR practices trigger investor concerns about agency problems. Such agency problems tend to become more serious during economy-wide financial distress rather than during a business boom (Rajan and Zingales (1998), and Johnson et al. (2000)). External corporate governance can help mitigate agency problems (Shleifer and Vishny 1997, Gompers et al. 2003) by improving the efficient usage of firm resources and decreasing managers discretion and expropriation of rents. Therefore, the presence of efficient external governance can protect shareholders against value-destroying over-investment in CSR activities. Institutional investors tend to be the dominant investors in financial markets and play an important role in corporate governance (Gillan and Starks (2003), Starks (2009) and Gillan et al (2010)). Research shows that institutional investors affect firm management and firm values via two complementary channels: 1) to engage with management actively; and 2) to influence managers decisions through an exit threat (Admati and Pfleiderer 2009, Edmans 2009, Edmans and Manso 2011, and McCahery, Sautner, and Starks 2014). During the financial crisis, stock liquidity becomes thin and thus blockholders power of an exit threat becomes weaker and institutional engagement intensity increases (Edmans and Manso 2011, and McCahery, Sautner, and Starks 2014). In other words, around a financial crisis, institutional investors monitor firms mainly through their involvement in corporate management. However, prior studies show that excess governance controls can have negative impact on firm value (See, among others, Claessens et al., 2002; La Porta et al., 2002; Lemmon and Lins, 2003). In addition, Guiso, Sapienza, and Zingales (2014) show that high institutional investor intervention decreases management integrity. Low levels of perceived management Page 12 of 45

13 integrity are positively correlated with bad outcomes in terms of lower productivity and lower firm value (Tobin s Q). Therefore, there are mixed predictions as to how external corporate governance affects the relation between CSR and firm value. In this paper, we empirically examine the differential impact of CSR practices on firm value across firms with different institutional ownership composition before, during, and after the financial crisis. 3. Data and Variables Definition A. CSR Data Our study focuses on the relation between CSR and firm value before, during and after 2008 financial crisis. To be included in our sample, a firm has to be listed on the Russell 1000 index in The Russell 1000 Index includes approximately the largest 1,000 companies in the U.S. based on market capitalization. 6 We limit our sample to Russell 1000 index members for the following reasons. First, the limited sample makes our data collection doable since we hand collect firms ESG scores from Bloomberg. Second, the index members are representative in each industry. And third, the index firms have low variation in firm size. Thus the tests using this sample are less likely driven by size effects or membership effects. CSR data are collected from the Bloomberg ESG dataset before, during, and after the 2008 financial crisis. 7 The Bloomberg ESG disclosure scores are each based on the areas of environment, social, and governance. The Environmental score is measured along dimensions including (but not limited to): energy consumption, water use, methane emissions, environmental 6 The Russell 1000 includes 1,000 stocks, representing 90% of the total market capitalization of the Russell 3000 Index. Its top sectors include technology, financial services, health care, energy and consumer related firms. 7 Chatterji, Levine, and Toffel (2009) suggest that KLD is lack of transparency about firms environmental performance. In addition, Chatterji, Durand, and Levine (2015) suggest that CSR ratings such as KLD, FTSE4Good, and DJSI have low validity. (Take it to the formal text and add with the discussion on Griffin s paper we need highlight the weakness of other CSR datasets). Page 13 of 45

14 fines. The Social score is measured along the following dimensions: number of employees; percent of employees unionized; training policy; human rights policy; anti-bribery ethics policy; UN Global Compact Signatory etc. Governance is measured along the dimensions including: size of the board; percent of independent directors; number of board meetings; board meeting attendance; board meeting attendance etc. Bloomberg s ESG scores are gathered from public sources such as company CSR reports, annual filings, corporate websites, questionnaires, media coverage, and public disclosure data such as carbon disclosure project (CDP) data (Bloomberg (2012)). Bloomberg assigns a score in each area and then constructs an integrated ESG score for the company, which is then adapted to the geographical and industry sector 8. The resulting score ranges from 0 for firms that do not disclose ESG data to 100 for firms that disclose all ESG data recognized by Bloomberg. Therefore, the Bloomberg ESG scores reflect all publicly available information on CSR practices. However, Park and Ravenel (2013) show that ESG data is often qualitative rather than quantitative. For example, Bloomberg ESG data suffers inconsistencies across industries and across years. Therefore, the difference in the magnitude of ESG scores does not have rich quantitative implications. Should we mention at this point this is why we use a binary variable? Griffin and Mahon (1997) show there are inconsistencies amongst various CSR databases. They find that the Fortune Reputation survey and KLD databases are biased in that the databases can indicate erroneous decisions based on impressions of what the firm has done (and not on what the firm has actually done). Additionally, Griffin and Mahon (1997) find the Toxic 8 According to Bloomberg online ESG manual, Each data point is weighted in terms of importance, with data such as Greenhouse Gas Emissions carrying greater weight than other disclosures. The score is also tailored to different industry sectors. In this way, each company is only evaluated in terms of the data that is relevant to its industry sector. Although the scoring is based on the Global Responsive Initiative (GRI), Bloomberg s exact weighting methodology is proprietary. Page 14 of 45

15 Release Inventory (which emphasizes the environmental dimension) and corporate philanthropy database are based on more hard quantitative data. Given the inconsistencies described, we find that the Bloomberg ESG database provides a relatively robust measure given its emphasis on actual publicly available information. (Perceptual, purely self-reported. Note that Griffin compared 4 different CSR datasets) Due to the qualitative feature of ESG dataset, we classify firms into two categories based on firm ESG disclosure score. We define the dummy for CSR Firm as one if a firm has a positive ESG disclosure score and zero if a firm with a zero ESG disclosure score. Such treatment mitigates the concern of measurement error in the raw ESG scores. We combine data on CSR with the data from the Center for Research in Security Prices (CSRP) monthly and daily files and from Compustat Annual Industrial file (COMPUSTAT). To compare the firm values of CSR Firms with other firms, we apply the propensity matching approach to find comparable firms for each CSR Firm in year Specifically, we match each firm that has a recorded ESG score with other Russell 1000 index members that have no ESG score along two different dimensions: firm size and industry, where we define industry according to French s 17 industry definitions 9. We require matching firms not to have ESG disclosure score during any of the sample years of our study (from 2006 to 2010). B. Key variables definition We study how corporate social responsibility (CSR) affects firm value before, during and after the 2008 global financial crisis. We use Tobin s Q to proxy for firm value. Tobin s Q reflects market opinion about the firm s future cash flow and risk. 10 We define Q as the ratio of 9 The detailed definition is available at 10 Please note, all valuation is forward-looking. Page 15 of 45

16 market value of assets divided by the book value of assets, both computed at the end of each fiscal year. Since it is the dependent variable, according to econometric theory, our tests do not suffer any estimation bias due to estimation errors in Tobin s Q. Motivated by Laeven and Levine (2008), we control for pre-identified factors that affect firm value, factors such as firm size, sales growth, capital expenditures, fixed asset to book asset, leverage, R&D activities, and profitability. Specifically, we define size as the log of book asset. Sales growth rate is measured as the ratio of previous year s sales to current year s sales minus one. Capital expenditures to book asset ratio is calculated as the ratio of capital expenditures to the book value of total assets. We measure fixed asset to book asset as the ratio of book value of property, plant, and equipment to the book value of total assets. Book leverage is calculated as the ratio of total debt to total assets. We define R&D activity as R&D to book asset ratio, which equals to research and development expense divided by total book asset. We set R&D expense to zero when research and development expense is missing. Finally, we measure profitability as the ratio of operating income before depreciation to the book value of total assets. We further examine the differential effects of external corporate governance on the relation between CSR and firm value. We use three measures to proxy for external corporate governance: Top 5 Institutional Ownership, Block holder Ownership, and Long-term Institutional Ownership. Top 5 Institutional Ownership refers to the percentage of shares in a firm held by its top five institutional investors with highest amount of holdings. Block holder Ownership refers to the percentage of shares in a firm held by its block holders, the institutions with at least 5% ownership. Long-term Institutional Ownership refers to the percentage of shares in a firm held by its long-term institutional investors. Following Yan and Zhang (2009), we first classify institutional investors into short-term and long-term investors according to portfolio Page 16 of 45

17 turnover rates (or churn rates) over the past four quarters. The one with a churn rate ranked in the top (bottom) tercile is defined as short-term (long-term) investors. The data on institutional ownership is from Thomson Financial CDA/Spectrum 13f institutional holdings data. A full description of the variables used is presented in Appendix One. Table 1 provides the descriptive statistics for CSR Firms and their corresponding matching firms in 2007, just before the financial crisis. At a minimum, firm characteristics between the CSR Firm Group and the matched firms should not be statistically different so as not to undermine the effectiveness of the matching process. Table 1 show that the CSR Firms exhibit similar characteristics from those of their matching firms, with the exception of book leverage. For example, there is no significant statistical difference between CSR Firms and their peers in Tobin s Q, book asset, sales growth rate, capital expenditure, fixed assets, R&D expense, and three measures of institutional ownership. The results indicate that the matching process is efficient. In our later tests, we still control for the heterogeneity in firm fundamentals in order to cleanly identify the causality between CSR activity and firm value. [INSERT TABLE ONE ABOUT HERE] 4. Empirical Analysis We now apply multivariate regression models to examine the joint effect of corporate social responsibility (CSR) and external corporate governance on firm value around financial crisis. We define Crisis as an indicator variable to indicate the 2008 financial crisis. In particular, Crisis is set to 1 for fiscal-year ends in December of We assign a value of one to the variable CSR Firms if the sample firm has positive ESG disclosure score, and zero otherwise. Page 17 of 45

18 We compare the changes in firm values between CSR Firms and firms without CSR activities over the period from 2006 and Firm Value Change as a result of Financial Crisis We first test whether firm CSR plays a role in the changes of firm value during the financial crisis. We employ the difference-in-difference (DID) methodology to test for differences in firm value to the CSR activities. Specifically, we investigate the changes in firm values, measured as Tobin s Q, for the CSR Firms before and during financial crisis, controlling for the changes in firm values of other matching firms over the same period of time. We use comparable firms without CSR as a control group to help difference out possible confounding factors and isolate the effect of CSR practice on firm value. We estimate the following regression model: Tobin s Q i = α + β 1 (Dummy CSR Firm i ) + β 2 (Dummy Crisis ) + β 3 (Dummy CSR Firm i Dummy Crisis ) + β 4 Controls + e i, where Dummy Crisis equals one for year 2008, the year of financial crisis, and zero for year 2006, the year before the onset of the financial crisis. In the regression models, the first difference, captured by the estimate of a dummy variable Dummy Crisis, reflects the change in firm value between one year before and the year of financial crisis (i.e., year 2006 vs. year 2008). The second difference is the differences in the changes of firm values between the CSR Firms and other firms, reflected in the estimate of the dummy variable Dummy CSR Firm. To capture the difference-in-difference of the CSR Firms value, the key variable of interest is the interaction of two dummy variables: a Dummy for the CSR Firm and a Dummy for Page 18 of 45

19 the Crisis. The purpose is to examine how the CSR activity affects the changes in firm value by controlling for other possible factors that can affect the changes in firm value over time. We use a variety of control variables suggested by the Tobin s Q literature (see, among others, Laeven and Levine (2008)). Specifically, in first regression we control for firm size, sales growth rate, capital expenditure and leverage. In the second model we include R&D intensity and profitability as additional controls to mitigate concerns of omitted variables. Table 2 presents the estimation results over the period 2006 through The coefficient estimates on the indicator of CSR Firm is negative and statistically significant for both model specifications. For example, in Model 2, the coefficient of the dummy of being CSR Firm is with t value of The results in Table 2 suggest that on average, CSR Firms exhibit lower firm value than other comparable firms before the financial crisis. In addition, the coefficients on the dummy variable Crisis are negative and significant for both models. The evidence suggests that firms, on average, experience loss in firm values when the financial crisis arrives. [INSERT TABLE 2 ABOUT HERE] Interestingly, results in Table 2 indicate that there is no significant difference in the change of firm value between the CSR Firms and their matched peers when the financial crisis occurs. The coefficient estimates of the interaction term between the dummies for CSR Firm and the crisis are all positive, but not statistically significant across all the regression models. For example, in the second estimation model (as shown in Column 2), the coefficient of the interaction term between the dummy for the CSR Firm and the dummy for Crisis is with t- value of Page 19 of 45

20 The signs of the coefficients of the control variables in the regressions of Table 2 are consistent with the existing empirical evidence on Tobin s Q. In particular, note that firm Tobin s Q is negatively related to book asset value and asset tangibility while it is positively related to the sales growth rate, R&D activities, and firm profitability. In sum, the evidence in Table 2 suggests that the CSR Firms have lower firm values than other firms before financial crisis. In addition, the change in the firm values of CSR Firms is not significantly different from that of their matched peers when the (unexpected) financial crisis hits. The evidence suggests the benefits of employing CSR programs do not necessarily outweigh the costs when an unexpected crisis happens. We then whether external corporate governance can potentially affect the relation between CSR and firm value. 4.2 The impact of CSR activity on firm value in financial crisis for groups of firms with different institutional ownership composition In this section, we examine the differential impact of CSR practices on firm value across firms with different institutional ownership composition when unexpected financial crisis comes. We use three measures to measure the ownership of influential institutional investors: block ownership, the aggregate ownership of top five institutional investors, and long-term investors. Studies on block governance suggest that blockholders and large institutional investors have an impact on firm value through their intervention in management decisions and management control (Admati and Pfleiderer 2009, Edmans 2009, Edmans and Manso 2011). We sort our sample into two groups according to the ownership of influential institutional investors in year 2006: High- (above median) and Low- (below median) influential ownership. We keep the composition of groups unchanged for both year 2006 and We re-estimate the Page 20 of 45

21 same regression model as in Section 3.1 for the high- and low- influential ownership group respectively. Table 3 presents the estimation results over the period of year 2006 and year 2008 for high- and low-blockholder ownership group, high- and low-top five institutional ownership group, and high- and low-long-tern institutional ownership group, respectively. Consistent with the findings in Table 2, the coefficients on the dummy variable of the Crisis are all negative and significant across all ownership groups. Interestingly, the negative relation between CSR and firm value exists only among firms with low external influential ownership. For example, the coefficients on the dummy variable of the CSR Firm are all statistically negative among firms with low block ownership (Column 1), low top five institutional ownership (Column 3), and low long-term institutional ownership (Column 5). For example, among firms with low long-term institutional ownership, the coefficient for the dummy of being CSR Firm is with a t-value of However, the coefficients on the dummy variable of the CSR Firm are statistically insignificant among firms with high top five institutional ownership (Column 4), and high long-term institutional ownership (Column 6). For example, among firms with high long-term institutional ownership, the coefficient for the dummy of being CSR Firms is with a t-value of [INSERT TABLE 3 ABOUT HERE] The results show that among firms with a weak external corporate governance, the CSR Firms have lower firm values than their comparable peers before financial crisis. The evidence suggests that the costs of implementing CSR activities outweigh the benefits of doing so. The findings are consistent with the notion that strong external corporate governance help mitigate agency problems and prevent over-investment. Page 21 of 45

22 More importantly, Table 3 documents insignificant impact of CSR practice on firm value when unexpected financial crisis comes regardless of different influential ownership levels. The evidence implies that the CSR activities does not lead to either more or less loss in firm values. In other words, the external influential ownership has an insignificant impact on the relation between the CSR activities and the change in firm values at the onset of the financial crisis. 4.3 Firm value during financial crisis We next investigate the level of firm value for the CSR Firms and their matching firms during the crisis in We ask whether and how the CSR activities affect firm value during the crisis. In table 4, we run cross-sectional regression of firm value (proxied by Tobin s Q) on an indicator of the CSR Firm for the sample period of year We use the same control variables as in Table 2. We show that there is a significant and negative impact of the CSR on firm value during the financial crisis. As shown in Table 4, the coefficient estimates on the indicator of the CSR Firm are negative and statistically significant for both model specifications, even after we control for firm R&D activity and profitability. For example, in Model 2, the coefficient of the interaction term between the dummy for CSR Firm and the dummy for Crisis is with t- value of Consistent with our previous findings, the evidence suggests that on average, adopting CSR may not be a value-maximization investment. Although firms obtain social and customers recognition of good citizenship, investors view CSR practices value-destroying, especially during the financial crisis. [INSERT TABLE 4 ABOUT HERE] Page 22 of 45

23 4.4 External corporate governance and firm value during the Crisis We now study how firm external corporate governance affects the relation between the CSR practices and firm values during the financial crisis. As in Table 3, we use three measures to measure the ownership of influential institutional investors: block ownership, the aggregate ownership of top five institutional investors, and long-term investors. We present the results in Table 5. We show that the negative relation between the CSR and firm value exists only among firms with low external influential ownership. For example, the coefficients on the dummy variable of the CSR Firm are all statistically negative among firms with low block ownership (Column 1), low top five institutional ownership (Column 3), and low long-term institutional ownership (Column 5). For example, among firms with low long-term institutional ownership, the coefficient for the dummy of being CSR Firm is with a t-value of However, the coefficients on the dummy variable of the CSR Firm are statistically insignificantly among firms with high block ownership (Column 2), high top five institutional ownership (Column 4), and high long-term institutional ownership (Column 6). For example, among firms with high long-term institutional ownership, the coefficient for the dummy of being CSR Firm is with a t-value of [INSERT TABLE 5 ABOUT HERE] Results in Table 5 suggest that the relation between the CSR activities and firm values depends on the strength of external corporate governance. Among firms with weak external monitoring mechanism, investors will discount the value of firms that have CSR practices during the financial crisis. Although CSR activities brings firms many benefits such as improving productivity and innovation among employees (Turban and Greening, 1997) and boosting firms Page 23 of 45

24 product market reputation (Schuler and Cording, 2006; Smith, 2008), our evidence indicates that during the financial crisis, investors are concerned about CSR-related agency problems, especially for firms with weak external governance. 4.5 Recovery of Firm value after financial crisis In this section, we study the recovery of firm value after the financial crisis. It is still an open question whether the CSR Firms recover faster or slower than their comparable peers. On the one hand, the CSR practices help firms recovery by improving productivities, enhancing product market competition advantages. On the other hand, CSR practices may slow down the recovery of firm value. It costs firms to maintain high level of CSR practice. Firms with high costs of CSR programs bear high financial burden that can negatively affect their after-crisis recovery. In addition, the potential CSR-related agency costs will hurt firm value. We run a regression of firm value on the dummy of the CSR Firm for year 2008 and year We ask whether and how the CSR practices affect the changes in firm values after the financial crisis. To do so, we include the interaction term between the indicators of the dummy for being CSR Firm and the dummy for after the financial crisis period. We includes a set of control variables identified by literature on Tobin s Q, such as firm size, sales growth rate, capital expenditure, and leverage. In the second model, we include R&D and profitability as additional controls. Table 6 presents the estimation results. Consistent with the results in Table 4, we document a negative relation between high CSR practices and the level of firm value during the financial crisis. For example, in Model 2, the coefficient of the dummy of being CSR Firm is with t value of Page 24 of 45

25 Moreover, evidence in Table 6 suggests that on average firms improve their firm values after financial crisis. The coefficient for the indicator of period after crisis is positive and statistically significant. For example, in Model 2, the coefficient of the dummy of After Crisis is with t value of More importantly, the results indicate that the CSR Firms do not have as quick recovery as other firms right after 2008 financial crisis. The interaction terms between the CSR Firm dummy and After-Crisis dummy are insignificant for all model specifications. As shown in Model 2, the coefficient of the interaction term is with t value of In sum, the results in Table 6 show that all firms experience improvements in their firm values. However, evidence suggests that although the CSR Firms, on average, have lower firm values than their comparable peers in 2008, there is no significant difference in the recovery of firm values between the CSR Firms and other firms right after the financial crisis. 4.6 Recovery of Firm value after financial crisis for groups of firms with different institutional ownership composition Finally, we examine the impact of institutional ownership on the relation between CSR and firm value after financial crisis. Table 7 presents the estimation results over the period of year 2008 and year 2010 for high- and low-blockholder ownership group, high- and low-top five institutional ownership group, and high- and low-long-tern institutional ownership group, respectively. Consistent with the findings in Table 6, the coefficients on the dummy variable of After Crisis are all positive and significant across all influential institutional ownership groups, suggesting that firm value improves after financial crisis. Page 25 of 45

26 In addition, consistent with the results in Table 5, results in Table 7 show that during financial crisis, CSR practice has negative effect on firm value only when firms have low influential institutional ownership. Specifically, the coefficients on the dummy variable of the CSR Firm are negative and significant for low-ownership groups but insignificant for highownership groups. Interestingly, results in Table 7 show that among firms with weak external governance firms with CSR programs increase their firm values more than other comparable firms after financial crisis. As shown in Table 7, the coefficient estimates of the interaction term between the dummy for the CSR Firm and the dummy for After Crisis are positive and significant for low-ownership groups. For example, in Column 5, the coefficient for the interaction term is with a t-value of [INSERT TABLE 7 ABOUT HERE] One possible explanation is that CSR Firms with weak external corporate governance experience a persistently large discount in firm values since the onset of the financial crisis, even to the extent of undervaluing the firm values of CSR Firms. After the financial crisis, with the improvement in product and financial market conditions, the agency costs and financial costs of implementing CSR practices become lessen. Therefore, among firms with weak external corporate governance, CSR Firms tend to recover the loss in their firm values quicker than other firms right after the financial crisis. In contrast, for high-ownership groups, the coefficient estimates of the interaction terms between the dummy for the CSR Firm and the dummy for After Crisis are insignificant. The evidence suggests that among firms with high external influential ownership, the CSR Firms do not have as rapid recovery in firm value as other firms after the crisis. Page 26 of 45

27 5. Conclusion and Directions for Future Research In this paper, we assess how CSR activity affects firm value. To disentangle the confounding interactions between CSR practice and firm value, we employing an unexpected exogenous shock, 2008 financial crisis as a natural experiment. We find that although CSR activity is positively related to firm value (Tobin s Q) before financial crisis, the firm values of CSR Firms decrease more than those of other firms during financial crisis. Moreover, the CSR Firms with different level of influential institutional ownership perform differently. We find that the firm values of the CSR Firms with high influential ownership decrease more than other firms with same level of ownership; while among firms with low influential ownership, the magnitude of decrease in firm values of the CSR Firms are not significantly different from those of other firms. The findings are consistent with the notion that during financial crisis, influential institutions tend to actively engage in the CSR Firms management, which can decrease the level of employee perceived managerial integrity and lead to lower firm performance and firm value. Furthermore, we find that all firms experience improvements in their firm values after financial crisis. However, our evidence suggests that the CSR Firms do not have a more rapid recovery in firm value than other firms after the crisis, regardless of their influential institutional ownership. Our study has investment implications to socially responsible investments (SRI). According to our findings, investments in CSR firms with lower influential ownership can improve SRI performance. Our study thus contributes to ongoing research that gauges benefits Page 27 of 45