Corporate social responsibility and corporate performance : the effect on operating performance, firm value and excess returns

(1)

Corporate Social Responsibility

and Corporate Performance

The effect on Operating Performance, Firm Value

and Excess Returns

Mirte van Loenen

2/4/2015

University of Amsterdam Amsterdam Business School MSc Business Economics, Finance track Master Thesis Thesis Supervisor: J.E. Ligterink

(2)

1. Introduction

Corporate Social Responsibility (CSR) is becoming an increasingly important part of doing business around the world. Companies are allocating significant portions of their expense budgets to CSR. For example, Di Gulli and Kostovetsky (2014: 159) estimate that large U.S firms spent $28 billion on sustainability and $15 billion on corporate philanthropy in 2010. More than half of the Fortune 1000 companies in the US regularly issue CSR reports, and nearly 10% of US investments are screened to ensure that they meet CSR-related criteria (Galema et al., 2008). It is however the question whether this makes sense: do CSR expenses lead to a better performance?

The CSR-performance relationship is obviously already examined more frequently. The majority of these previous studies focusses on accounting-based measures of performance, notably profitability measures. The results of these studies are far from conclusive. Some of the studies find a positive relationship between CSR and operating performance (for example, Orlitsky et al., 2003 and Falck and Heblich, 2007), some find a negative relationship (for example, McWilliams and Siegel, 2000) and others find no significant relation between the two (for example, Di Gulli and Kostovetsky, 2014). A much smaller part of previous studies on the CSR-performance relationship is focusses on the effect of CSR on capital markets, for example by examining the impact of CSR on excess returns of assets (see, for example, Hamilton et al., 1993; Sauer 1997; Statman, 2000; Bauer et al., 2005 and Derwall et al, 2005). The results of these study are inconclusive as well. It is notable that previous studies always focus on just one small aspect of the CSR-performance relationship: they examine the effect of CSR on operating performance or the effect of CSR on excess returns or the effect of CSR on cost of equity, etc. The different aspects of the relationship are never integrated in just one study. This may harm the explanatory power of the studies and the overall understanding of the effects of CSR. For example, studies who found a positive effect between CSR and firm value state that this is the result of improved profitability. This is remarkable, since these studies do not test the effect of CSR on profitability themselves and as we saw before, there is absolutely no scientific agreement on this effect. This means that – because of the segmentation in studies on the CSR-performance relationship - results are explained without first thoroughly testing this explanation itself. This thesis seeks to advance our understanding of the CSR-performance relationship by examining whether CSR performance affects firms’ the broad concept ‘corporate performance’, thereby looking at the effect of CSR on operating performance, firm value and excess returns on assets. This is one of the first studies that tests the impact of CSR on various

(4)

aspects of corporate performance. This will lead to a more complete understanding of the CSR-performance relationship.

In addition, previous studies predominantly focused on positive aspects of CSR, overlooking that firms also undertake actions and initiatives that qualify as negative CSR. Moreover, studies in this area have not investigated how both positive and negative CSR affect corporate performance of firms. To get a complete understanding of the effects of CSR on corporate performance, however, it also need to be clear what the effect of 'negative CSR' actions and initiatives on corporate performance are. This thesis, therefore, will not only focus on positive aspects of CSR, but on CSR performance as a whole. These two motivations lead to the following research question:

“To what extent does CSR performance has an impact on operating performance, firm value and excess returns?”

As explained above, this study integrates different aspects of the CSR-performance relationship and focusses on both positive and negative ‘CSR actions’, thereby filling two existing gaps in the literature on CSR. However, this study can also contributes to the different aspects individually. Firstly, existing studies on the relation between CSR and firm value use very narrow definitions of CSR. The vast majority of prior research use a measure of environmental performance (for example, Konar and Cohen, 2001) or a binary indicator (for example, Harjoto and Jo, 2009) as their measure of CSR. Next to that, previous studies that study this relationship do not check for possible endogeneity and/or reversed causality issues, meaning the their results could be biased. Since we use a broad definition of CSR and check for possible endogeneity and/or reversed causality issues, this study is a methodological improvement on previous studies that examine the relationship between CSR and firm value.

Secondly, a significant part of previous studies on the relation between CSR and excess returns compare the financial performance of socially responsible mutual funds with the performance of conventional mutual funds (see, for example, Hamilton et al., 1993; Sauer 1997; Statman, 2000; Bauer et al., 2005; Kreander et al.. 2005) These studies have the drawback that the performance of mutual funds depends to a large extent on the skills of the mutual fund manager (Baks, 2003). Since these skills cannot be easily measured or observed, they are not included in the analysis. Therefore, it is likely that endogeneity issues occurred in these studies. Other studies examine the effect of CSR on excess returns by examining socially screened stock portfolios. In a significant part of these studies, the definition of socially responsible companies is again solely based on their environmental score (see, for example, Derwall et al., 2005). Therefore, by using a broad definition of CSR and by examining portfolios instead of mutual funds, this study is an addition on already existing studies on the relation between CSR and excess returns.

(5)

With the help of the various measures, both for positive actions as CSR-concerns, we could test the different relationships between CSR and corporate performance. We tested the relation between CSR and operating performance and CSR and firm value by using a cross-sectional time-series regression with robust standard errors and industry- and yearly fixed effects. Since endogeneity and reversed causality issues can occur when testing this relationship, we conducted several robustness checks including an IV-regression. The relation between CSR and is tested by using the four-factor model of Carhart (1997).

In summary, the results of this study show that CSR only slightly influences corporate performance. CSR has no effect on firm value and excess returns on assets, but it has an effect on operating performance. The effect on operating performance is surprising: CSR has a positive effect on profitability (measured by the net profit margin and return on equity), but a negative effect on sales growth. This indicates that the positive effect of CSR on the net profit margin and the return on equity is probably the result of reduced costs. This is a new insight that has not been mentioned in previous studies. In addition, we found that CSR-concerns do not seriously affect corporate performance as well: we only found a negative effect on firm value, but a positive effect on sales growth. Since this is one of the first studies that examine the effect of CSR-concerns on corporate performance, this is a new insight as well.

After this introduction, the next chapter gives an overview of the relevant academic literature on CSR and the CSR-performance relationship. Subsequently, the third chapter provides an overview of the data in this study and explains the method that is used to answer the research question. In chapter 5, the results of our analysis are given and chapter 6 describes the robustness checks that are used to test for endogeneity and reversed causality issues. Finally, this thesis will be ended with a conclusion in chapter 7.

(6)

2. Literature Review

In this chapter, an extensive review of the relevant literature is given in order to obtain an in-depth understanding of the existing knowledge on relation between CSR and corporate performance. This existing knowledge contribute to the rest of the thesis in two ways. Firstly, it makes sure that we are aware of the underling mechanisms, whereby results of the empirical research can be better understood. Next to that, it makes sure that it will be noticed when found results that are not in line with the results of previous research. In this chapter, we start by defining the concept of CSR. Subsequently, reasons why companies implement CSR-policies are considered. Thirdly, the possible relationship between CSR and corporate performance is explored. The final section examines different methods which are used in previous studies that tried to establish a relationship between the CSR and corporate performance.

2.1 Defining CSR

Corporate Social Responsibility has become a more and more important part of firms operation over the past decade. This means that companies are not only trying to reach their financial goals, but also taking responsibility towards society, the environment or their stakeholders in general (Renneboog et al., 2008: 1730). Firms are investing ever more resources in public goods provision, and many companies reduce negative externalities below levels required by law. More than half of Fortune Global 250 firms now provide regular public statements exclusively discussing CSR, and 3approximately 10 percent of S&P 100 companies report in detail on CSR activities (Kotler and Lee, 2004; Baskin and Gordon, 2005). Also, more and more investors are concerned about non- financial dimensions of corporate performance, such as the impact on the environment, social relations, and corporate governance. Socially responsible investing (SRI) attracts a lot of investor attention. According to the Social Investment Forum1_{, approximately}

10% of all US investments are managed according to screening processes related to SRI; U.S. and European markets have over 2 trillion USD and 300 billion EURO in certified socially responsible assets.

To understand this movement towards more social responsible firms, we first have to define what CSR means. However, this is not that easy as it seems. As Baron (2001: 10) argued, “Corporate social responsibility is an ill- and incompletely deﬁned concept”; there are many different definitions of CSR. According to Kitzmueller and Shimshack (2012: 53), attempts to define CSR reveal two basic conceptual features. Firstly, CSR has to manifest itself in some observable and measurable behavior or output. Firms that only say to take responsibility 1

www.social-invest.org, last visited on 23-01-2015

(7)

towards the environment and their stakeholders are not considered as ‘good CSR-firms’. To be taken seriously, firms have to prove what they do to take responsibility and what the outcomes of these actions are. This is why a lot of firms provide public statements discussing CSR. Secondly, CSR has to exceed levels set by obligatory regulation or standards enforced by law. Probably the most broad and general definition is proposed by the World Business Council for Sustainable Development (WBCSD). They define CSR as “the commitment of a business to contribute to sustainable economic development, working with employees, their families, the local community and society at large to improve their quality of life” (WBCSD, 2004).

In this definition, some different dimensions of CSR can be found. These dimensions are also mentioned in other definitions of CSR. In his analysis of 37 definitions of CSR, Dahlsrud (2006) concludes that there are five of these CSR-dimensions. The first dimension of CSR is the environmental dimension. This means that companies should have some environmental stewardship when they want to be socially responsible. The second dimension is the social dimension, meaning that socially responsible firms should contribute to a better society and integrate social concerns in their business operations. The economic dimension is the third dimension. This dimension focuses on the socio-economic or financial aspects of CSR. The interaction with and treatment of stakeholders is also important for a social responsible firm, so the fourth dimension is the stakeholder dimension. Finally it is of interest that a company is voluntary socially responsible. The voluntariness dimension is therefore the fifth dimension of CSR (Dahlsrud, 2006: 4).

2.2 Reasons to implement CSR-policies

In spite of the growing importance of CSR, the question why companies invest in CSR is subject to much debate - especially given the mixed evidence on the relations between CSR and firm performance. Generally, there are two opposing views on CSR: the stakeholder value maximization view and the shareholder expense view. Supporters of the stakeholder value maximization view argue that CSR activities have a positive effect on shareholder wealth because focusing on the interests of other stakeholders increases their willingness to support a firm's operation (Deng et al., 2013: 88). According to this view, CSR has a positive effect on firm performance and on shareholder wealth. In contrast to this view, the shareholder expense view suggests that managers engage in socially responsible activities to help other stakeholders at the expense of shareholders wealth (Deng et al., 2013: 89). In this paragraph, we are going to examine what the financial and strategic motivations behind CSR could be. Different studies on the effect of CSR on stakeholders and vice versa are discussed. Some of these studies find positive effects of CSR on performance and therefore support the stakeholder value maximization view, while other studies are more in support of the shareholder expense view.

(8)

One of the most important stakeholders for a company are its (possible) customers. Different studies show that well designed CSR policies of companies may lead to an improvement of the image of a company, and thereby to an increase of customers and sales. Bhattacharya and Sen (2001) show that a good CSR policy of a company may lead to a competitive advantage over competing companies and that it helps to gain customers from the market leader. However, they mention that CSR is not all there is: company- specific factors as product quality and individual-specific factors as customers beliefs about CSR determine magnitude of the effect of CSR. Servaes and Tamayo (2009) also find that CSR policies may lead to an increase of customers and sales, but they emphasize that customer awareness is key in this finding; when customer awareness is low, CSR policies will not lead to an improvement in performance. Finally, CSR policies do not only attract customers, irresponsible corporate action even lead to a reduction in customers. Grappi et al. (2013: 1814-1816) find that consumers respond angry to irresponsible corporate actions. This anger and disgust results in a negative word of mouth, and will therefore negatively influence the sales of a company. Good CSR policies may prevent irresponsible corporate actions and therefore may prevent a loss in sales. These positive effects of CSR on customers’ attitudes are in line with the stakeholder value maximization view.

A second important group of stakeholders for a company are its employees. Brekke and Nyborg (2005) argue that companies who implement good CSR policies attract (intrinsic) motivated workers, who can increase the productiveness of a company and therefore positively influence firm performance. However, other studies find more negative effects of CSR policies with regard to this group of stakeholders. For example, Surroca and Tribó (2008: 749-750) find that (an improvement in) CSR policies is related to managerial entrenchment practices: entrenched managers, who work to ensure their own job security even though they are no longer competent, use CSR to collude with stakeholders in order to reduce a firm’s attractiveness to potential raiders. Managerial entrenchment practices are, according to Surroca and Tribó (2008), thus positively related to improvements in CSR which, in turn, negatively affect firms’ financial performance. Cronqvist et al. (2009: 309-310) add to this finding that entrenched managers pay their workers more, which also negatively affects firm performance. These empirical results show that both the stakeholder value maximization view and the shareholder expense view can be supported when examining the effect of CSR on employees.

In addition to employees, insiders of firms (managers, board members and large blockholders) are an important group of stakeholders as well. Different studies show that this stakeholder group is very determinative in the choice of a firm to implement CSR policies. Barnea and Rubin (2010) show that CSR results from agency costs. They argue that insiders may seek to over-invest in CSR strategies for their private benefit, namely to improve their

(9)

reputations as good citizens and responsible managers. Di Guili and Kostovetsky (2014) show that firms score higher on CSR when they have Democratic rather than Republican shareholders, CEOs and directors and when they are headquartered in Democratic rather than Republican states. The ideas and values of insiders may thus lead to a strong CSR policy. Since ideas and values do not immediately lead to an increase in financial performance, these results support the stakeholder expense view.

Next to customers, employees and insiders, investors are of great importance for any company. CSR is becoming more and more important to investors and several studies have shown the impact of CSR on this field. Firstly, El Ghoul et al. (2011: 2388) show that firms with better CSR scores obtain cheaper equity financing. They have two explanations for this finding. Firstly, they argue that effective CSR policies may result in a reduction in agency and information asymmetry problems. Next to that, they argue that CSR affects the riskiness of a firm in a positive way (El Ghoul et al, 2011: 2389). In addition, they find that stocks of firms in ‘sin’ industries like tobacco and nuclear power have higher expected returns as they are less preferred by norm-constrained investors. However, they are more likely to face greater litigation risk and therefore have a significantly higher cost of equity capital (El Ghoul et al, 2011: 2390). The findings of El Ghoul et al. are also confirmed by Kim et al. (2014: 1), who show that a firms' CSR performance is negatively associated with future stock price crash risk, resulting to cheaper equity financing of good CSR-firms. Besides equity financing, a firm can also lend money from banks. Goss and Roberts (2011) examine the link between corporate social responsibility (CSR) and bank debt. They find that find that firms with social responsibility concerns pay between 7 and 18 basis points more than firms that are more responsible. However, they do not find evidence that lenders reward firms for their efforts to be leaders in corporate social responsibility.

Finally, the last stakeholder group that will be discussed is ‘the society’, consisting of people who are not directly linked to the firm because of their investments or their jobs, but do feel the consequences of the firms’ operations. Firstly, CSR policies can be the result of external pressures from people within the society. Baron (2001) models CSR as firms’ rational responses to ‘private politics’ where lobbyists put pressure on firms to adopt more stringent environmental standards. CSR policies can thus be the result of pressures from society, but society can also benefit from good CSR policies. Heal (2005) states that, by anticipating and minimizing the potential conflicts between corporations and society, CSR plays a role in reducing the costs of conflicts. However, other studies find a more negative effect of incorporating society as an important stakeholder. Allen et al. (2007) discuss the advantages and disadvantages of stakeholder-oriented firms in relation to shareholder-oriented firms. They

(10)

show that societies with stakeholder oriented firms have higher prices and lower output due to reduced competition in product markets.

There are two opposing views on CSR: the stakeholder value maximization view and the shareholder expense view. In this part of the literature review, we examined reasons to implement CSR-policies. We found support for both the stakeholder value maximization view and the shareholder expense view. We found that firms who invest in CSR may increase their customer-base, attract intrinsic motivated workers, have access to cheaper equity-financing and have less potential conflicts with societal stakeholders. However, CSR also can result in managerial entrenchment and reduced competition. Next to that, external pressures from the society and ideas and values of insiders may be reasons why some firms implement CSR-policies. Since these reasons are not financial or strategic considerations, it may harm shareholder wealth.

2.3 The relationship between CSR and Corporate Performance

In the previous section we found that CSR activities can have a positive effect on shareholder wealth (stakeholder maximization view), but can destroy shareholder wealth as well (shareholder expense view). In this thesis, we analyze the effect of CSR on corporate performance. Corporate performance is operationalized in three ways, namely in operating performance, firm value and excess returns on assets. This paragraph explores the possible relationship between CSR and these three components of corporate performance. This is done by outlining the possible theoretical relationship(s) for every part, by using the literature in the previous section. Next, we review previous empirical studies that have examined the relationship.

The first relation that is examined is the link between CSR and operating performance. Operating performance measures look at how well a company turns its assets into revenues (thereby increasing shareholder value). There are numerous ways to measure operating performance. In this thesis, it is measured by the net profit margin, the return on equity and the one-year sales growth. CSR can influence operating performance in two possible ways. Firstly, it may have an effect on the revenues of a firm. In the previous section we saw that CSR policies may lead to an improvement of the image of a firm, and thereby to an increase of customers and sales (Bhattacharya and Sen, 2001; Grappi et al., 2013; Servaes and Tamayo, 2009). Secondly, CSR has effect on the costs of firms. Conducting CSR-policies means that a firm has to invest in the environment, communities, etc. This will increase the costs of a firm. Next to that, in the literature review we saw that CSR may lead to managerial entrenchment practices (Cronqvist et al., 2009; Surroca and Tribó, 2008), leading to higher costs as well. However, CSR may also reduce costs of firms: studies show that companies who implement good CSR policies attract

(11)

(intrinsic) motivated and thus very productive employees (Brekke and Nyborg, 2005) and can attract cheaper financing (El Ghoul et al., 2011; Goss and Roberts, 2011; Kim et al., 2014). Concluding, implementing CSR-policies will always increase the costs of a firm since firms have to invest in the environment, communities, etc. When the revenue of a firm increases due to the investments in CSR and this revenue-increase is bigger than the total investment in CSR, there exists a positive link between CSR and operating performance. The same goes when the cost-decreases of a firm due to the investments in CSR are bigger than the total investment of CSR. However, when the total investment in CSR is bigger than the revenue-increase or cost-decline, there exists a negative relationship between CSR and operating performance.

The empirical relationship between CSR and operating performance is analyzed a lot, but the results are still unclear and inconclusive. Some studies show a positive link between CSR and the operating performance of firms. Falck and Heblich (2007: 250-251) show that CSR can lead to better performance in two ways. First, when CSR is used to react on ‘social trends’ it can result in greater commitment of stakeholders, resulting in a more efficient working environment. Next to that, CSR can be used to strengthen the image of companies. According to Falck and Heiblich (2007: 253) if companies treats society well, society will return the favor. The positive link of CSR on a firms corporate performance is also shown by Orlitsky et al. (2003) and Margolis et al (2007). The studies of these researches are both meta-analysis of all the different studies that examine the relationship between CSR and financial performance of firms. Although they see a positive relationship between the two, they also conclude that this relationship is very small and not very convincing. The results of Orlitsky et al. (2003) are especially interesting since they take into account the potential reversed causality by controlling for the good management theory and the slack resources theory (both explained in the next section). Other studies on CSR and financial performance of firms find a neutral relation. An example of such a study is the study of McWilliams and Siegel (2000: 605). According to them, the studies that show a positive link between CSR and financial performance are biased because they do not include R&D expenses in their regressions, while R&D expenses and investments in CSR are correlated and R&D expenses often lead to a better performance in the future. When R&D expenses are included, there no significant relationship between CSR and financial performance of firms can be found. Yet another group of studies – for example Di Gulli and Kostovetsky (2014) - find that there exists a negative relationship between CSR and a firms financial performance, because according to them the costs of investing in CSR can never offset the benefits of it. It is thus still very unclear what the effect of CSR on operating performance is.

The second relation that is examined in this thesis, is the relation between CSR and firm value. Firm value reflects the market value of a firm and is often measured by Tobins’ Q, calculated as the market value of a company divided by the replacement value of the firm's

(12)

assets. The firm value, or the Tobins’ Q, increases when the demand for shares of a firm is high and the price of the shares rises. This may happen for multiple reasons. Firstly, when investors expect a firm to be very profitable in the upcoming years, they would be willing to pay more for a share of that firm and vice versa. When CSR has an effect on operational performance (the relationship we examined above), it will therefore indirectly impact the firm value. Next to this possible relationship, socially responsible firms are very popular on the stock market for two other reasons. Firstly, about 10% of investors are only willing to invest in socially responsible firms because they value the effort of firms to invest in the society and the environment (Renneboog et al., 2008). This increases the demand and therefore the price of shares from socially responsible firms, and in this way it can increase the market value of these firms. However, since this is not a ‘financially-motivated’ reason to buy shares of a certain firm, shares of socially responsible firms may become overvalued. This will reduce the expected return of these shares. This makes the shares less attractive, which may result in a decrease in the price of the share, reducing the firm value again to its old level. Secondly, we saw before that firms with CSR policies are less risky than firms that don’t have CSR policies (Godfrey et al., 2009; Oikonomou et al., 2012; Kim et al., 2014). CSR activities of firms can therefore provide an ‘insurance-like’ benefit to investors. This may cause a ‘premium’ on the prices of good CSR firms, increasing the firm value. El Ghoul et al. (2011) show that – because of a reduction in risk - the cost of equity is lower for good CSR-firms. This leads to a higher firm valuation as well. In conclusion, CSR may impact firm value in different ways. Firstly, when CSR has an effect on operating performance it probably has an indirect effect on firm value as well. Secondly, the high demand for shares from socially responsible firms may increase the share prices and therefore firm value. However, this will also reduce the expected return of the shares, which may result in a decline of the share price again. Thirdly, CSR-firms are less risky, what possibly results in a premium on their shares.

Compared to the amount of studies on the effect of CSR on operating performance, there are just a few studies that analyze the relationship between CSR and firm value. The majority of prior research on this relation use a measure of environmental performance as their measure of CSR. For example, Konar and Cohen (2001) examine the extent to which a firm's environmental reputation is valued by investors. They find that bad environmental performance is negatively correlated with market value of S&P 500 firms. Consistently, Dowell et al. (2000) find that U.S.-based multinational enterprises adopting a single, stringent global environmental standard have much higher market values than firms defaulting to less stringent, or poorly enforced host country standards. Wahba (2008) studied the influence of engaging in environmental responsibility on corporate market value in the Egyptian context. His findings demonstrate a positive relation between environmental responsibility and firm market value as well. Another

(13)

problem of these studies, besides using a very narrow definition of CSR, is that these studies do not account for standard corporate governance attributes when investigating the valuation effect of corporate social responsibility. Two studies that use a more broad measure of CSR are Fisman et al. (2006) and Harjoto and Jo (2009). Fisman et al. (2006) use data on CSR from the Kinder, Lydenberg, and Domini's (KLD) Socrates database and find in general a negative relation between CSR and firm value. However, they show that the presence of outside blockholders with board representation and a stronger competition on the product markets both lead to a more positive relationship between CSR and profitability. Harjoto and Jo (2009) use corporate governance data from IRRC/RiskMetrics and CSR data from the KLD Socrates database and show that CSR is positively associated with corporate governance and that both corporate governance and CSR have a positive valuation effect. They argue that firms use corporate governance mechanisms along with CSR to reduce the conflicts-of-interest between shareholders and non-investing stakeholders. A common problem in these two studies is that the measure of CSR is binary and only indicates whether a firm invests in CSR or not. Moreover, the studies do not account for potential endogeneity and/or reversed causality issues. Finally, all studies above explain the effect of CSR on firm value by pointing at the effect of CSR on operating performance: studies who found a positive effect between CSR and firm value state that this is the result of improved profitability and vice versa. This is remarkable, since these studies do not test the effect of CSR on operating performance themselves and as we saw before, there is absolutely no scientific agreement on this effect. Explanations about the popularity of stocks of socially responsible firms and the riskiness of those stocks are disregarded in previous studies. In conclusion, the results on previous studies on the relation between CSR and firm value are inconclusive and methodological improvements are necessary.

The third relation that is examined in this thesis, is the relation between CSR and excess returns. Theoretically speaking, this relationship can be interpreted in different ways. If CSR matters for corporate performance and this relationship is fully incorporated by the market, then a stock price should adjust to any relevant change in the ﬁrm’s CSR policies. In that case, the excess return of good CSR-firms is equal to the excess return of bad CSR-firms. However, if CSR matters but is not incorporated immediately into stock prices, then realized returns on the stock of good CSR-firms would be systematically higher than from equivalent securities (Gompers et al., 2001: 121). Next to that, excess demand for socially responsible stocks and a shortage of demand for irresponsible stocks can lead to overpricing of the first and underpricing of the latter. This would lead to low to negative excess returns of stocks of socially responsible firms and a return premium for irresponsible stocks (Merton, 1987). In addition there is the link via risk. Previous research (Mishra and Modi, 2012) shows that shows that CSR has a significant

(14)

effect on the idiosyncratic risk of firms, with positive CSR reducing risk and negative CSR increasing it. If the risk of a firm decreases, then the cost of equity for that firm decreases.

The relationship between CSR and excess returns is just like the relationship between CSR and firm value not extensively studied yet. A significant part of previous studies on the relation between CSR and excess returns compare the financial performance of socially responsible mutual funds with the performance of conventional mutual funds. The results of these studies (see, for example, Hamilton et al., 1993; Sauer 1997; Statman, 2000; Bauer et al., 2005; Bello, 2005; Geczy et al., 2005; Kreander et al.. 2005; Renneboog et al., 2008) are univocal: socially responsible mutual funds do not earn statistically significant excess returns and the performance of such mutual funds is not statistically different from the performance of conventional mutual funds. However, these studies have the drawback that the performance of mutual funds depends to a large extent on the skills of the mutual fund manager (Baks, 2003). Since these skills cannot be easily measured or observed, they are not included in the analysis. Therefore, it is likely that endogeneity issues occurred in these studies. Other studies examine the effect of CSR on excess returns by examining socially screened stock portfolios. In a significant part of these studies, the definition of socially responsible companies is solely based on their environmental score. Yamashita et al. (1999), for example, examined with an event-study if the announcement of a better environmental score positively impacts short-term stock returns and found that the US capital markets have only weakly rewarded environmentally conscientious companies. Derwall et al. (2005) constructed two equity portfolios that differed in eco-efficiency. The high-ranked portfolio provided substantially higher average returns than its low-ranked counterpart over the 1995-2003 period. These results suggest that environmental responsibility positively impacts returns. However, the Social Investment Forum report that socially responsible investors typically consider a multitude of criteria. There are only a few studies that examine socially screened stock portfolios with multiple criteria. Becchetti and Ciciretti (2009) analyzed the performance of a large sample of socially responsible stocks relative to a control sample of equivalent size for 14 years. They found that individual socially responsible stocks have on average significantly lower returns but are significantly less risky as well. Consistently, Brammer et al. (2005) found that CSR is negatively related to stock return and that considerable abnormal returns are available from holding a portfolio of the socially least desirable stocks. In addition, Hong and Kacperczyk (2007) report higher expected returns for stocks that are usually excluded from a portfolio because of negative ethical issues. For example, they exclude companies involved in alcohol and tobacco. They find that these so-called sin stocks are underpriced, have higher book-to-market values, and higher excess returns than other stocks. They also find that sin stocks have less analyst coverage and are less held by pension funds. The studies of Becchetti and Ciciretti (2009),

(15)

Brammer et al. (2005) and Hong and Kacperczyk (2007) use respectively linear regression and the CAPM to estimate excess returns, while in the ‘mutual-fund’ studies the Carhart model is often used. Concluding, there are just a few studies that analyze socially screened stock portfolios with a broad definition of CSR. These studies may be improved by using another model like the Carhart (1997) model.

2.4 Measuring the effects of CSR

There are essentially three types of empirical studies to investigate the relationship between CSR and financial performance of firms. The first type uses the so-called event study methodology to assess the short-run financial impact when firms engage in either socially responsible or irresponsible acts (see, for example, Clinebell and Clinebell, 1994; Hannon and Milkovich, 1996; Posnikoff, 1997; Teoh, Welch and Wazzan, 1999; Worrell, Davidson and Sharma, 1991; Wright and Ferris, 1997). A second set of studies examines the relationship between CSR and measures of long term firm performance, using accounting or financial measures of profitability (see, for example, Aupperle, Carroll, and Hatfield, 1985; McGuire, Sundgren and Schneeweis, 1988 and Waddock and Graves, 1997). The third steam are those studies that focus on market performance of companies, by looking at the firm value and excess returns (see, for example, Brammer et al., 2005; Fisman et al., 2006; Hong andKacperczyk, 2007; Becchetti and Ciciretti, 2009; Harjoto and Jo, 2009).

Even though the timescope of these short- and long-run studies differs, the variables they use are more or less the same. Firstly, both studies need to use a proxy for the Corporate Social Performance of firms as an independent variable. In many cases in the past, subjective indicators are used, such as a survey of business students (Heinze, 1976), or business faculty members (Moskowitz, 1972). In other cases, CSR-reports or annual reports of firms are analysed. However, these reports are subjective and might under- or over-report on different topics. In 1988, Kinder Lydenberg Domini (KLD) started with providing a CSR rating system, where numerous companies are rated on multiple attributes considered relevant to CSR. KLD uses a combination of surveys: financial statements, articles on companies in the popular press, academic journals (especially law journals), and government reports in order to assess CSR along different dimensions. The KLD ratings are considered as the most objective and reliable CSR-measure available, and therefore this rating is used by almost all recent studies on CSR (For example, El Ghoul et al., 2011; Kim et al., 2014; McWilliams and Siegel, 2000; Servaes and Tamayo, 2013 and Tsoutsoura, 2004).

As an dependent variable, both studies need to use a measure of profitability. Some researchers put forth accounting measures (McWilliams ans Siegel; Waddock and Graves 1997), while others use market measures or both. Accounting measures are ‘backward –looking’ – they measure past performance – and are more than market measures subject to managerial 15

(16)

manipulation and differences in accounting procedures. Market measures are ‘forward-looking’ – they measure investors’ expectations of future growth – and focus on market performance (Tsoutsoura, 2004: 10).

Next to a dependent and an independent variable, both type of studies need control variables. Firstly, firm size is an important control variable, since larger firms seem to adopt the CSR principles more often (Tsoutsoura, 2014: 12). Next to that, risk and and industry have been suggested in previous articles (Ullman 1985, McWilliams and Siegel 2000) to be factors that affect both a firm’s performance and CSR. Finally, earlier research shows that R&D expenses and investments in CSR are correlated and R&D expenses often lead to a better firm performance in the future (McWilliams and Siegel, 2000; Graves and Waddock, 1994). R&D expenses are therefore another important control variable.

The results of event-studies who used (a type of) the above method have been mixed; Wright and Ferris (1997) found a negative relationship; Posnikoff (1997) reported a positive relationship; and Teoh et al. (1999) found no relationship between CSR and financial performance. Also the results from long run studies have also been mixed. Aupperle et al. (1985) found no relationship between CSP and profitability, McGuire et al.(1988) found that prior performance was more closely related to CSP than was subsequent performance, and Waddock and Graves (1997) found significant positive relationships between an index of CSP and performance measures such as ROA in the following year.

The big variety of results of studies that investigate the effect of CSR on firm performance point out that measuring the effects of CSR on performance is quite difficult. This has a lot to do with possible endogeneity and the direction of causality. Although this can be caused in numerous ways, in the CSR-performance relationships two causes are very common. The first cause is named ‘the theory of slack resources’. The idea of this theory is that better financial performance of firms potentially results in the availability of slack resources that provide the opportunities for firms to invest in social performance domains as employee relations and community relations. If this is the case, better financial performance would be a predictor of better CSR policies instead of the other way around (Waddock and Graves, 1997: 306). The second cause is called ‘the good management theory’ and this theory states that there is a high correlation between good management and CSR, because attention to CSR domains improves the relationship between the manager and the stakeholders. These good relationships between managers and stakeholders may result in better overall performance of the firm, meaning that not CSR but good management is the explanation of (an improvement in) firm performance (Waddock and Graves, 1997: 307-308).

Although it is never possible to completely eliminate these endogeneity effects, in recent studies researches were more aware of these problems and tried to find ways to reduce these

(17)

endogeneity effects. Firstly, a lot of studies use the so called ‘event study’, originally made to assess the effect of an unanticipated event on stock prices. For example, Becchetti et al. (2012: 1631) used an event study to examine the market reaction to corporate entry and exit from the Domini 400 Social Index. It is often argued that – because events are unanticipated and thus endogenous – an event study will reduce endogeneity effects. In this case however, it is possible that a ‘reversed version’ of the theory of slack resources may have influenced the results of this study: when firms are performing badly, they have less resources to invest in CSR, causing an exit on the Domini 400 Social Index. In this case, not the exit of the Domini Index causes a negative effect on abnormal returns, but the fact that the firm is performing badly. For McWilliams et al. (1999), this is one of the reasons to state that event study methodologies are inappropriate to assess the effect of strategic decisions like CSR on firm performance. Another reason for rejecting event study methodologies is that event study findings are in this case sensitive to even small changes in research design (McWilliams at al., 1999: 341).

Another way to control for the effects of endogeneity, are extensive robustness checks. For example, El Ghoul et al. (2011) are aware of the fact that in their study on the effect of corporate social responsibility (CSR) on the cost of equity capital, there are some possible endogeneity problems; for example, it could be that capital choice and CSR are not independent of each other. Also Kim et al. (2014) are aware that, when studying the effect of CSR performance on stock price crash risk, endogeneity can arise due to unobservable heterogeneity when unobservable firm specific factors affect both CSR and crash risk. Both El Ghoul et al. (2011) and Kim et al. (2014) control for these possible endogeneity effects by for example regressing the dependent variable on various CSR proxies. These kinds of robustness checks are thus a possible way to check for and prevent endogeneity effects.

(18)

3. Hypotheses

This study will start with exploring the relationship between CSR and operating performance, measured by the net profit margin, the return on equity and the one-year sales growth. As stated in the literature review, CSR has the possibility to improve the operating performance of firms, because it may positively influence the image and thereby increase the customer-base of firms, it attracts intrinsic motivated workers, it results in cheaper financing possibilities and less potential conflicts with societal stakeholders. However, CSR also can result in managerial entrenchment and reduced competition. In general, CSR is only beneficial if its’ benefits are bigger than its’ costs.

The results of empirical studies on the relationship between CSR and operating performance are very divided and inconclusive. However, big meta-analyses of Orlitsky et al. (2003) and Margolis et al. (2007) show that CSR has a (small) positive effect on operating performance. However, these studies have focused on positive aspects of CSR, and not on actions and initiatives that qualify as CSR-concerns. We assume that CSR-concerns will result in a smaller customer-base, resulting in a decrease in operating performance. Our first hypotheses are therefore:

H1a: CSR has a positive impact on the operating performance of firms

H1b: CSR-concerns have a negative impact on the operating performance of firms

Secondly, in this thesis the relationship between CSR and firm value will be examined. CSR may influence firm value in three ways. Firstly, when CSR has an effect on operating performance it probably has an indirect effect on firm value as well. Secondly, the high demand for shares from socially responsible firms may increase the share prices and therefore firm value. Thirdly, CSR-firms are less risky, what possibly results in a premium on their shares.

There are just a few empirical studies that analyze the relationship between CSR and firm value. The results of these studies are again divided and inconclusive. Next to that, previous studies use very narrow definitions of CSR, do not control for endogeneity and/or reversed causality issues and are do not take CSR-concerns into account. Therefore, our hypothesis is not based on previous empirical results but on reasoning. Since we expect CSR to have a positive impact on operating performance, it will probably also positively influence firm value and vice versa. The fact that CSR-firms are less risky will probably positively impact firm value as well. Our hypotheses are therefore:

H2a: CSR has a positive impact on firm value

(19)

Thirdly, the relation between CSR and excess returns will be analyzed. When CSR matters for operating performance and this relationship is fully incorporated by the market, the excess return of ‘good’ CSR-firms will be equal to the excess return of ‘bad’ CSR-firms. However, if CSR matters but is not incorporated immediately into stock prices, then realized returns on the stock of good CSR-firms would be systematically higher than from equivalent securities. Next to that, excess demand for socially responsible stocks and a shortage of demand for irresponsible stocks can lead to overpricing of the first and underpricing of the latter - independent of the performance of ‘good’ CSR-firms. This would lead to low to negative excess returns of stocks of socially responsible firms and a return premium for irresponsible stocks.

Most previous empirical studies on the relation between CSR and excess returns analyze the performance of social responsible mutual funds. They all find no effect of CSR on mutual fund performance. However, these studies might be biased since mutual fund performance depends to a large extent on the skills of the mutual fund manager. There are just a few studies that examine the effect of CSR on excess returns by examining socially screened stock portfolios. Most of these studies use a very narrow definition of CSR. The studies that use a broad definition find a negative relation between CSR and excess returns, and a positive relation between the so-called ‘sin’ stocks and excess returns. Although we are going to use a different method than these previous studies, we assume their results will hold. Therefore our final hypotheses are:

H2a: CSR has a negative impact on excess returns

H2b: CSR-concerns have a positive impact on excess returns

(20)

4. Data and Method

This thesis examines the effect of CSR on (different measures of) corporate performance in 2000 to 2012. To measure CSR, we make use of the KLD Stats Database. The construction of the CSR variable is explained extensively in the next section, however for now it is important to know that this database had data available for only a limited group of companies: the largest 500 US companies (S&P 500) and the KLD 400 Social Index (Domini Index). The latter is an index, created by KLD themselves, of the 400 most socially responsible firms. Obviously, using this index would mean that you test the effect of CSR on firms that all have a very high CSR-score. This would lead to biased results. That is why in this thesis data on US S&P 500 listed firms are used. At the moment, the KLD database is expanded to the biggest 3000 US firms. However, this expansion started after 2000. Since our timeframe is 2000-2012, we cannot use these data.

In the literature review we saw that a big part of studies on CSR use the so-called ‘event study methodology’. This method is used to assess the short-run financial impact when firms engage in either socially responsible or irresponsible acts. Although it is often argued that an event study will reduce endogeneity effects, the literature review told us that is not the case when studying the effects of CSR. Besides, an event study is only suitable when assessing the effect of an unanticipated event on stock prices. Since we do not only analyze the effect of CSR on stock prices but also on performance in general, the event study methodology is not suitable for this thesis. Therefore, we examine the relationship between CSR and measures in the long run, using both ‘backward-looking’ accounting measures as ‘forward-looking’ market measures to measure corporate performance. Our method is based on the method that Gompers et al. (2003) use in their article “Corporate governance and equity prices”, however we modified it to make it suitable for an analysis with CSR as independent variable.

It is very likely that our method, as explained in this chapter, suffers from endogeneity and/or reversed causality issues. Both the ‘theory of slack resources’ as ‘the good management theory’ might be applicable on the relationships we are going to analyze. We will not pay attention to these endogeneity- and reversed causality problems in this chapter, but it we will control for these issues in our robustness checks.

In order to conduct our analysis, we merged four datasets: the KLD Stats Database, data on Fama/French factors from the website of Kenneth R. French2_{, Compustat data, which}

provides accounting data and CSRP, which provides information on stock returns. Corporate performance is operated in three ways, namely in returns, firm value and operating

2_{http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html, last visited on 22-12-2014}

(21)

performance. Returns are measured as the excess returns on an asset, firm value is measured with the Tobins’ Q and operating performance is measured as the net profit margin, the return on equity and one-year sales growth. This section starts with the construction of the CSR-variables. Afterwards, the method to test the effect of CSR on operating performance is explained, followed by the explanation of the method of the other two parts (firm value and excess returns) of this thesis. To maintain overview on what data are used in what part, the construction and use of all variables (except the CSR-variable) is described in part to which they belong.

4.1 Construction of CSR-variables

The KLD Stats Database (part of MSCI) started with providing a CSR rating system in 1988, and is up until now the most extensive database available for evaluating a firm's CSR activity. It rates numerous companies on multiple attributes considered relevant to CSR, using a combination of surveys: financial statements, articles on companies in the popular press, academic journals (especially law journals), and government reports in order to assess CSR along different dimensions. The KLD ratings are considered as the most objective and reliable CSR-measure available, and therefore this rating is used by almost all recent studies on CSR (For example, El Ghoul et al., 2011; Kim et al., 2014; McWilliams and Siegel, 2000; Servaes and Tamayo, 2013 and Tsoutsoura, 2004).

KLD STATS organizes the various CSR-related items into two major categories: qualitative issue areas and controversial business issues. The qualitative issues are ratings built on a point-by-point assessment of companies on nearly 60 categories along a seven dimensions: community activities, diversity, employee relations, environmental record, products, human rights and corporate governance. For each dimension, KLD assigns a binary (0/1) rating to a set of concerns (for example, ‘support for controversial regimes’) and strengths (for example, ‘community engagement’). The controversial business issues include alcohol, gambling, tobacco, firearms, the military and nuclear power. For each controversial business issue, KLD assigns a binary (0/1) rating for whether a firm is involved in (at least one of) a set of concerns. An extensive description of the qualitative issue areas and the controversial business issues can be found on the website of WRDS3_.

Because qualitative issue areas and controversial business issues are inherently different, we examine them separately. We calculate a score for each qualitative issue area equal by subtracting the number of concerns from the number of strengths. To obtain the final variable CSR-Score, we sum the qualitative issue areas’ scores to obtain the total scores of the

3 http://wrds-web.wharton.upenn.edu/wrds/support/Data/_001Manuals%20and%20Overviews/_070KLD/index.cfm, last visited on 10-1-2015 21

(22)

qualitative dimensions. In estimating CSR-Score, we exclude the score on ‘corporate governance’, because it is not in accordance with the definition of CSR as given in the literature review. The other dimensions are very much in line with the five general dimensions of CSR mentioned in the literature review, and therefore are very suitable to use for this research. Next to that, most other studies exclude the corporate governance dimension as well (For example, El Ghoul et al., 2011; Kim et al., 2014; Servaes and Tamayo, 2013). Therefore, excluding corporate governance makes our results more comparable to results of previous studies. The pearson correlation coefficients for CSR-Score and the Corporate Governance Score are showed in table 1.

Table 1 - Pearson correlation coefficients for CSR_Score and Corporate Governance Score

CSR_Score Corporate Governance Score CSR-Score 1.0000

Corporate Governance Score 0.2186 1.0000

In this thesis, we also analyses the effects of actions and initiatives that qualify as CSR-concerns. The controversial business issues in the KLD database are a useful way to do this. We capture a firm’s involvement in controversial business issues with a dummy variable that takes the value of 1 if a firm is involved in any of the six controversial business areas (CSR_Contr). However, these data are only available until 2010. Summary statistics for CSR_Score and

CSR_Contr are shown in table 2.

Table 2 – Summary statistics of CSR_Score and CSR_Contr

2000 2004 2008 2012 2000-2012 CSR-Score - Minimum -8 -8 -8 -3 -9 - Maximum 11 10 13 16 18 - Mean 0.69 0.35 0.52 3.44 1.12 - Median 1 0 0 3 1 - Standard deviation 2.75 2.92 3.41 3.15 3.52 CSR_Contr

- Percentage of firms involved in any of the

controversial business areas (CSR_Contr=1) 24.04% 14.32% 15.37% - 16,54%

After the CSR-Score variable is calculated, two portfolios are constructed: one with the 20% firms that have the highest CSR-score and one with the 20% firms that have the lowest score. To construct this portfolio, we determine what the 20% best and 20% worst CSR-firms are in every single year. The reason we have not chosen to select the 20% best and 20%

(23)

worst companies over all years (2000-2012) is that the summary statistics show that CSR-score of companies is increasing over the years. This high score in the last few years would therefore give a distorted view and associated endogeneity issues when the portfolio is constructed over all the years instead of every year apart. Summary statistics for the constructed portfolio are shown in table 3.

Table 3 – Summary statistics of the portfolios

2000 2004 2008 2012 2000-2012 Best 20% - Minimum 4 4 4 7 3 - Maximum 11 10 13 16 18 - Mean 5.04 5.43 6.12 8.46 6.59 - Median 4 5 5 8 6 - Standard deviation 1.62 1.50 2.47 1.89 2.72 Worst 20% - Minimum -8 -8 -8 -3 -9 - Maximum -1 -2 -2 1 1 - Mean -2.35 -3.19 -3.37 0.03 -2.67 - Median -2 -3 -3 0 -2 - Standard deviation 1.52 1.35 1.56 1.03 1.84

4.2 The effect of CSR on operating performance

Firstly, we are going to test the effect of CSR on operating performance using three operational measures: net profit margin (income divided by sales), the return on equity (income divided by book equity), and one-year sales growth. The data for these measures are derived from Compustat. To examine the effect of CSR on operating performance, we conduct a cross-sectional time-series regression with robust standard errors and industry- and yearly fixed effects:

𝑃𝑃𝑖𝑖𝑖𝑖 = 𝑎𝑎𝑖𝑖 + 𝑏𝑏𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶𝑖𝑖𝑖𝑖+ 𝑐𝑐𝑖𝑖𝐶𝐶𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖𝑖𝑖+ 𝑑𝑑𝑖𝑖𝐿𝐿𝑆𝑆𝐿𝐿𝑆𝑆𝐿𝐿𝑎𝑎𝐿𝐿𝑆𝑆𝑖𝑖𝑖𝑖+ 𝑆𝑆𝑖𝑖𝐶𝐶&𝐷𝐷𝑖𝑖𝑖𝑖+ 𝑓𝑓𝑖𝑖𝐼𝐼𝐼𝐼𝑑𝑑𝐼𝐼𝐼𝐼𝐼𝐼𝐿𝐿𝐼𝐼𝑖𝑖+ 𝑆𝑆𝑖𝑖𝑌𝑌𝑆𝑆𝑎𝑎𝐿𝐿𝑖𝑖+ 𝜀𝜀𝑖𝑖𝑖𝑖

P represents one of the three operational measures. CSR represents one of the CSR-variables we

constructed; we conduct this regression on the continue CSR-score variable, on the dummy for the 20% best CSR-scores, on the scores of the separate qualitative CSR issues and on the controversial issues. Hereby we get a complete picture of the possible ways that CSR effects firm

(24)

value. Size, Leverage and R&D are control variables. Size is captures size effect and is measured by the total market capitalization. This measure is used by almost all studies that examine the effect of CSR on operating performance, since bigger firms often invest more in CSR. Secondly leverage - measured by dividing the total leverage by the total assets - is added as a control variable, since firms who have a lot of leverage do not invest that much in CSR. McWilliams and Siegel (2000: 604) argue that R&D expenses should be included as a control variable as well, since investments in R&D results in improvements in long-run economic performance. All these variables are derived from Compustat and they are logs, except for CSR and R&D. Table 4 shows the descriptive data for the regression variables.

Table 4. Descriptive data for regression variables

Mean Min Q1 Median Q3 Max St. Dev

Panel A. Descriptive statistics for control variables

Size 9.35 5.22 5.89 9.26 9.99 13.13 1.14

Leverage -1.70 -13.26 -1.98 -1.45 -1.07 0.44 1.05

R&D -16.81 -5052 -15.12 0 0 0 170.85

CSRScore Size Leverage R&D Panel B. Pearson correlation coefficients between regression variables CSRScore 1.0000

Size 0.2136 1.000

Leverage -0.0731 -0.0853 1.000

R&D -0.0437 -0.1266 0.0001 1.000

Beside the above mentioned control variables, industry- and yearly fixed effects are included in the regression. The industry fixed effects are included to control for industry-wide differences in CSR, as shown by El Ghoul et al. (2013). These effects are based according to the 12 industry group affiliations by Fama and French4_{. We chose for the 12 industry group and not}

the 48 industry group since our database consists of (just) 500 firms. When using a lot of industry groups, each group consists of only a few firms, which makes it harder to control for industry-wide differences.

As already explained in the introduction of this chapter, it is possible that this method suffers from endogeneity and/or reversed causality issues. We will pay attention to these endogeneity- and reversed causality problems in our robustness checks.

4.3 The effect of CSR on firm value

In the second part of our analysis, we study whether variation in firm-specific CSR-scores is associated with differences in firm value. We use Tobin’s Q as a valuation measure for firm

4_{For the portfolios, see}_{http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html}

24

(25)

value, since this is the most used measure for firm value since Demsetz and Lehn (1985) and Morck, Shleifer, and Vishny (1988). To examine the effect of CSR on Tobin’s Q, we conduct a cross-sectional time-series regression with robust standard errors and industry- and yearly fixed effects:

𝑄𝑄_{𝑖𝑖𝑖𝑖} = 𝑎𝑎_𝑖𝑖+ 𝑏𝑏_𝑖𝑖𝐶𝐶𝐶𝐶𝐶𝐶_{𝑖𝑖𝑖𝑖}+ 𝑐𝑐_𝑖𝑖𝐶𝐶𝑆𝑆𝑆𝑆𝑆𝑆_{𝑖𝑖𝑖𝑖}+ 𝑑𝑑_𝑖𝑖𝐵𝐵𝐼𝐼𝐵𝐵_{𝑖𝑖𝑖𝑖}+ 𝑆𝑆_𝑖𝑖𝐿𝐿𝑆𝑆𝐿𝐿𝑆𝑆𝐿𝐿𝑎𝑎𝐿𝐿𝑆𝑆_{𝑖𝑖𝑖𝑖}+ 𝑓𝑓_𝑖𝑖𝐼𝐼𝐼𝐼𝑑𝑑𝐼𝐼𝐼𝐼𝐼𝐼𝐿𝐿𝐼𝐼_𝑖𝑖+ 𝑆𝑆_𝑖𝑖𝑌𝑌𝑆𝑆𝑎𝑎𝐿𝐿_𝑖𝑖+ 𝜀𝜀_{𝑖𝑖𝑖𝑖} For the computation of Q, we follow Kaplan and Zingales’ (1997) method: we divided the book value of assets plus the market value of common stock less the sum of the book value of common stock and balance sheet deferred taxes by the book value of assets. CSR represents one of the variables we constructed. Again, we conduct this regression on the continue CSR-score variable, on the dummy for the 20% best CSR-CSR-scores, on the CSR-scores of the separate qualitative CSR issues and on the controversial issues to get a complete picture. Size, BtM and

Leverage are control variables that are used by numerous other studies that examine effects on

firm value (see, for example, Eisenberg et al., 1998; Lins, 2003; Lemmon and Lins, 2001). Size captures the size effect and is again measured by the total market capitalization. BtM represents the Book-to-Market ratio and is computed by dividing the book value of common equity with the market value of common equity. Leverage captures the risk of a company and is computed by dividing the total leverage with total assets. All variables – except CSR – are logs and are derived from Compustat. Table 5 shows the descriptive data for the regression variables.

Table 5. Descriptive data for regression variables

Mean Min Q1 Median Q3 Max St. Dev

Panel A. Descriptive statistics for control variables

Size 9.35 5.22 5.89 9.26 9.99 13.13 1.14

BtM -0.97 -3.59 -1.41 -0.97 -0.51 0.62 0.69

Leverage -1.70 -13.26 -1.98 -1.45 -1.07 0.44 1.05

CSRScore Size BtM Leverage Panel B. Pearson correlation coefficients between regression variables CSRScore 1.0000

Size 0.2123 1.000

BtM -0.1462 -0.3209 1.000

Leverage -0.0925 -0.0867 0.0355 1.000

Also in this regression, industry- and yearly fixed effects are included. The industry fixed effects are included to control for industry-wide differences in CSR, and are based according to the 12 industry group affiliations by Fama and French. We will pay attention to possible 25

(26)

endogeneity and/or reversed causality issues in the robustness checks.

4.4 The effect of CSR on returns

If CSR matters for firm performance and this relationship is fully incorporated by the market, then a stock price should quickly adjust to any relevant change in the firm’s CSR-policies. However, if CSR matters but is not incorporated immediately into stock prices, then realized returns on the stock would differ systematically from equivalent securities (Gompers et al., 2001: 121). To examine the relationship between CSR and subsequent returns, we conduct the following regression:

𝐶𝐶_𝑖𝑖 = 𝑎𝑎 + 𝛽𝛽₁∗ 𝐶𝐶𝐵𝐵𝐶𝐶𝑅𝑅_𝑖𝑖+ 𝛽𝛽₂∗ 𝐶𝐶𝐵𝐵𝐵𝐵_𝑖𝑖+ 𝛽𝛽₃∗ 𝐻𝐻𝐵𝐵𝐿𝐿_𝑖𝑖+ 𝛽𝛽₄∗ 𝐵𝐵𝑀𝑀𝑀𝑀𝑆𝑆𝐼𝐼𝐼𝐼𝐼𝐼𝑀𝑀_𝑖𝑖+ 𝜀𝜀_𝑖𝑖

This is the so-called four-factor model of Carhart (1997), also used by Gompers et al.(2001) in their study on the relationship between corporate governance and equity prices. Rt is the excess return to assets in time t, calculated as the return on an asset minus the risk free rate. RMRFt, SMBt and HMLt are the so-called Fama-French factors, where RMRFt is the month t value-weighted market return minus the risk-free rate. The terms SMBt (small minus big) and

HMLt (high minus low) are designed to capture size and book-to-market effects, since Fama and French (1993: 5-6) found that small cap stocks tend to outperform large cap stocks and value stocks outperform growth stocks. Carhart (1997) added the momentum factor to the three-factor model of Fama and French, because he proved that stock prices tend to continue rising if it is going up and vice versa. Data on (excess) return of assets is derived from the CRSP-database. Data on the Fama and French factors and the Momentum factor are derived from the website of Kenneth R. French5_.

We will conduct this regression on the two portfolio’s: one with the 20% best CSR-firms and one with the 20% worst-CSR CSR-firms. The intercept coefficient α is interpreted as the abnormal return in excess of what could have been achieved by passive investment in CSR. We will calculate the return difference between the 20% firms with the best CSR-score and the 20% firms with the worst CSR-score. The alpha in this estimation is the abnormal return on a zero-investment strategy that buys the portfolio with the 20% best CSR-firms and sells short portfolio with the 20% worst CSR-firms. In addition, we will also conduct the regression on the separate qualitative issues and on the controversial issues.

5_{http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html, last visited on 22-12-2014} 26

Corporate social responsibility and corporate performance : the effect on operating performance, firm value and excess returns