The effect of CSR on financial
performance
Karen Guit
10542949
June 2017
MSc Finance
Track: Corporate Finance
Supervisor: Jeroen Ligterink
This document is written by Student Karen Guit who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This thesis researches the effect of CSR on financial performance for the years 2003 till 2013 for 3,000 companies in the United States, based on the data availability of the MSCI database. The results show that CSR has a positive impact on market performance, it leads to a higher Tobin’s Q. However, it does not show a consistent effect on operating performance, which is measured by ROA, ROE and sales growth. This implies that engaging in CSR activities lead to a competitive advantage during the financial crisis. Moreover, for the effect of CSR on financial performance it does not matter whether the firm is in a high intensity advertising industry. By using instrumental variable regressions and treatment effects regressions with full maximum likelihood estimation this thesis addresses possible
endogeneity issues and is likely to measure causality.
Table of contents
1. Introduction ... 5
2. Literature review ... 8
2.1 Definition of CSR ... 9
2.2 Implementation reasons of CSR ... 10
2.3 Endogeneity issues ... 13
2.5 CSR and financial performance ... 14
2.5 Financial crisis and CSR ... 16
3. Hypotheses ... 17
4. Methodology ... 19
4.1 Main regression ... 19
4.2 Empirical model and possible endogeneity issues ... 22
5. Data and descriptive statistics ... 26
5.1 Data sources ... 26
5.2 Descriptive statistics ... 27
6. Results ... 30
6.1 Main results ... 30
6.2 Results in relation to the crisis ... 35
6.3 Results in relation to industry advertising ... 37
7. Robustness checks ... 39
8. Discussion and conclusion ... 42
Literature list ... 46
Appendix ... 53
1. Introduction
In the last years many firms in the United States (US) increased their investment in corporate social responsibility (CSR) (Deng, Kang & Low, 2013). This is done voluntarily as part of their strategy and vision or due to the pressure from activist shareholders (Deng et al., 2013). The activist shareholders use their equity stake to put public pressure on the management of the firm. Furthermore, in the annual reports the importance of CSR is shown. Many firms devote large sections of their annual reports to their CSR activities or publish annual CSR reports about their CSR achievements and activities (Deng et al., 2013). Despite the growing importance of CSR, there is still much debate about the question why and whether managers should invest in CSR activity. Particularly, there is much debate about the effect of CSR on firm performance due to the mixed evidence on this relation.
There are two opposite views on CSR in relation to firm performance; the stakeholder
value maximization and shareholder expense view (Deng et al., 2013). According to the stakeholder value maximization view, CSR activities have a positive influence on shareholder wealth. Dimson, Karakaş and Li (2015) support this. They find that successful engagements are followed by positive abnormal returns. The engagements address environmental, social and governance concerns. On the other hand, the shareholder expense view suggests that engaging in social responsible activities is at the expense of shareholders. This is in line with Renneboog, Ter Horst and Zhang (2008), who find that investors are paying a price for ethics. They compare social responsible mutual funds with conventional funds and find that overall investors pay a price for CSR.
This thesis contributes to this ongoing debate by researching the following question: does corporate social responsibility increase firm performance? It focuses on 3000 firms based in the US from 2003 till 2013. This paper uses the MSCI ESG KLD STATS (MSCI) database to construct the CSR variables. The different measurements of CSR increase the robustness of this research. The data range is chosen based on the availability of the data of this database. Several papers focus on firms based in the US due to the data availability of the CSR variable (among others Cornett, Erhemjamts and Tehranian (2016), Dimson et al. (2015), Dowell, Hart and Yeung (2000), Guenster, Derwall and Koedijk (2011) and Jiao (2010)). Most of these papers use the MSCI database or formerly the KLD and GMI database to construct the CSR variable.
This research focuses on the long-‐term firm performance. Existing research about this subject can be divided in two separate groups, research about short-‐term and long-‐term firm performance. Event study methodology is a method that can research short-‐term firm performance when firms engage in either responsible or irresponsible actions (for example, Hannon & Milkovich (1996) and Flammer (2015)). Previous work of CSR in relation to long-‐ term firm performance can be classified in two different categories; operating performance and market performance.
This thesis uses Tobin’s Q as a measure of market performance (Cornett et al. (2016), Dowell, Hart & Yeung (2000), Fisman, Heal and Nair (2006), Guenster et al. (2011) and Jiao (2010)). Return on assets (ROA), return on equity (ROE) and sales growth are the operational measures in this research (Aupperle, Carroll and Hatfield (1985), Fisman et al. (2006), Lev, Petrovits and Radhakrishnan (2010) and McGuire, Sundgren and Schneeweis (1988)). Different operational ratios measure different segments of a firm’s overall operational performance. So, the use of different measures generates a more complete effect of CSR on operational performance. For example, Fisman et al. (2006) and Lev et al. (2010) use only one operational measure. Therefore, their results could give a biased view of the effect of CSR on operational performance because they measure only a specific segment of the performance.
To control for possible endogeneity issues, this research uses IV regressions. The instruments in this research are: a dummy indicating whether the headquarter of the company is located in one of the top 25 greenest cities of the US, a dummy indicating whether the location of the company is in a Democratic state versus a Republican state and the average CSR score per industry. These variables are computed with several data sources, the process of some of these data sources is done manually. A lot of existing researches neglect possible endogeneity issues (for example, Aupperle et al. (1985), Dowell et al. (2000), Fisman et al. (2006), Hong and Kacperczyk (2009) and McGuire et al. (1988)). Therefore, their results could be biased. There are some researchers (for example, Cornett et al. (2016), Deng et al. (2013) Jiao (2010) and Jo and Harjoto (2011)) that try to solve the endogeneity problems with instrumental variable (IV) regressions, they focus mainly on the reversed causality and omitted variable bias.
performance tend to engage in CSR activities for reasons unrelated to taking responsibility for the company’s effects on environmental and social wellbeing and that the control variables do not capture this information. The decision to engage in CSR is modeled as a function of instruments that are shown to be important factors whether to engage in CSR, so CSR is a binary treatment variable in this model. The instruments are uncorrelated with financial performance. It addresses the self-‐selection bias where firms select their engagement in CSR activities because of some expected benefit in financial performance or other unaccounted reasons. Research suggests that this is not taken into account in previous literature that studies the effect of CSR on long-‐term financial performance. Therefore, this is a contribution to previous literature.
Moreover, this research has a contribution in comparison to existing papers due to the focus on the recent financial crisis. Papers that focus on CSR and a crisis, especially research about the effect of CSR on financial performance in context of a crisis, are scarce. The effect of CSR on financial performance could be different during the crisis. Investing in CSR might be a threat for the survival of the company. Nonetheless, it could also create a competitive advantage during the crisis by increasing the reputation of the firm and gaining trust of the community (Sitnikoc & Bocean, 2017). The research of Lome, Heggesseth and Moen (2016) suggests that the latter holds. They research this effect for research and development (R&D) investments. Furthermore, Cornett et al. (2016) provide an analysis of banks’ social performance and its relation to financial performance in the context of the recent financial crisis. However, this research focuses only on banks. So, this thesis clearly distinguishes itself by adding this financial crisis component.
Additionally, focusing on the effect of different industries enlarges the relevance of this research. Fisman et al. (2006) state that CSR has a different effect on financial performance for high intensity advertising industries. The signal value of CSR expenditures could be greater in high intensity advertising industries (Fisman et al., 2006). The transparency and therefore customer awareness of CSR is greater in such industries. This argumentation is backed by the research of Servaes and Tamayo (2013). They state that the implementation of CSR leads to an increase of customers and sales. However, for firms with low customer awareness the relation between firm value and CSR is either negative or insignificant. The research of Fisman et al. (2006) is about this relationship, but does not take
into account possible endogeneity problems. Therefore, this thesis can distinguish itself by being the first that researches this and address the possible endogeneity issues.
The next chapter presents a discussion of the existing literature. It focuses on the
definition of CSR, the implementation reasons of CSR, the endogeneity issues, the link between financial performance and CSR and the financial crisis and CSR. In the third chapter there is a description of the hypotheses. The fourth chapter presents the methodology; it shows the main regressions, empirical model and possible endogeneity issues. The fifth chapter is about the data sources and descriptive statistics. Chapter six presents the results. The next chapter shows several robustness checks. The discussion and conclusion are in chapter eight.
2. Literature review
This chapter describes the existing literature in the field of research of this thesis. It starts with an introduction followed by the definition of CSR. Thereafter, there is a description about the implementation reasons of CSR. A section that describes the possible endogeneity issues in this field of research follows. The fourth section discusses papers that analyze the effect of CSR on financial performance. Furthermore, there is a discussion about the existing literature in relation to the financial crisis and CSR in the fifth part.
According to the classical finance theorem, corporations have a shareholder-‐value approach. This means that value-‐maximizing shareholders should control firms; other stakeholders are protected by contracts and regulations (Bénebou & Tirole, 2009). However, society and lawmakers demand for individual and corporate social responsibility as a response to market and redistributive failures have recently become more prevalent. There is an increase in demand of assets under management in the US that fall into the social responsible investing category (Social Investment Forum, 2007), which shows the increase in demand for CSR. Furthermore, the proliferation of fair-‐trade products, carbon offsets and newly created positions such as Corporate Sustainability Officer in many large companies show the same trend (Bénebou & Tirole, 2009).
According to Bénebou & Tirole (2009) there are several reasons to account for this
trend. Firstly, social responsibility is likely to be a normal good. For example, the demand of products that are socially responsible does not depend on income of the consumer.
accessible. Thirdly, the extension of CSR reaches to more laxly regulated countries due to the existence of multinationals. Lastly, the long-‐run cost of atmospheric pollution increases significantly, or at least the public’s awareness of it.
In reality, corporations mostly aim to focus on objectives beyond profit maximization.
They engage in activities that improve other stakeholder’s welfare, so besides the shareholder’s welfare (Deng et al., 2013). The social responsibility of corporations leads to an increased debate, examination, media attention and academic research (Sahlin-‐ Andersson, 2006). To better understand that shift, the next part elaborates on the definition of CSR.
2.1 Definition of CSR
According to Carroll (1999) CSR contains the economic, legal, ethical and philanthropic responsibilities of companies. This is a wide-‐ranging explanation. Kitzmueller and Shimshack (2012) define CSR from an economic perspective, looking at the higher welfare of CSR compared to other public good provision channels. CSR manifests itself in some observable and measurable behavior or output. Furthermore, it exceeds levels set by obligatory regulation or standards enforced by law.
Darhlsrud (2006) state that there are five CSR dimensions. First of all, the environmental dimension in which companies should have some environmental conservancy to become socially responsible. Secondly, according to the social dimension there must be a relationship between business and society. Thirdly, the economic dimension focuses on the socio-‐economic or financial aspects. Fourthly, according to the stakeholder dimension a firm must focus on the interaction with and treatment of stakeholders. Lastly, the voluntariness dimension includes actions that are not prescribed by law.
In this thesis there are seven CSR dimensions, which are in line with Darhlsrud (2006) and link to the MSCI database. The seven dimensions are: community, environment, diversity, employee relations, human rights, product and corporate governance. In this thesis CSR focuses on the positive and negative environmental and social performance indicators of the MSCI database.
The computation of CSR can be done in different ways. The amount of charitable donations could be the base of the CSR variable, Fombrun and Shanley (1990), Kedia and Kuntz (1981), Levy and Shatto (1980), Seifert, Morris and Bartkus (2003) and Seifert, Morris
and Bartkus (2004) use this. Also observer perceptions can be the base of CSR, however this could be biased by the subjectivity of the observer. Examples are Graves and Waddock (2000), Griffin and Mahon (1997) and Luo and Bhattacharya (2006). There are many other bases possible to measure CSR, such as corporate policies, revealed misdeeds, transparency, screened mutual funds, self-‐reports (Margolis, Elfenbein & Walsh, 2007). Third-‐party audit is the base of the construction of the CSR variable in this research, just as in Cornett et al., (2016), Deng et al., (2013), Jiao (2010), Jo and Harjoto (2011), and Krüger (2015). This is an objective measure and is easily accessible. Section 4.1 shows the computation of the CSR variable.
2.2 Implementation reasons of CSR
The next few paragraphs discuss the decision whether and why companies should implement and invest in CSR. To structure this discussion, the reasoning relates to all the stakeholders of a firm: customers, employees, managers and large blockholders, investors and society. Thereafter, there is a description about the reasoning that relates to the stakeholder value maximization and shareholder expense view.
This paragraph focuses on the reasoning relating to the customers. The research of Sen and Bhattacharva (2001) is about consumer reactions to CSR. They find that CSR leads to a competitive advantage and that a company could gain customers from the market leader. However, it is important to note that company specific factors determine the magnitude of the effect of CSR. According to Servaes and Tamayo (2013), implementation of CSR leads to an increase of customers and sales. However, for firms with low customer awareness, the relation between firm value and CSR is either negative or insignificant. Furthermore, Grappi, Romani and Bagozzi (2013) investigate consumer’s response to irresponsible corporate actions. They find that consumers react angry on those actions, which has a negative impact on sales. With the implementation of CSR activities a company could prevent the loss in sales. With CSR a company might be able to prevent this lower tail outcome, which increases firm value. Furthermore, according to Luo and Bhattacharya (2006), CSR affects market value partially through the mediator of customer satisfaction.
Another motivation to engage and invest in CSR is that it has a positive impact on
social responsible firm and exert higher effort when productivity cannot be observed. Furthermore, morally motivated workers demand lower wages (Brekke & Nyborg, 2004). Vinerean, Cetina and Dumitrescu (2013) support this; they show that CSR initiatives lead to attraction and retention of good employees. These papers only focus on CSR in relation to employee satisfaction and attraction and retention of employees. However, motivated employees work harder and this could lead to an improvement of financial performance. Additionally, lower wages lead to a reduction in costs, which has a positive impact on financial performance.
The third set of motivations to engage and invest in CSR has a link with managers and
large blockholders. Entrenched managers may collude with non-‐shareholder stakeholders in order to reinforce their entrenchment strategy (Surroca & Tribó, 2008). These entrenchment strategies and the implementation of socially responsible actions have a negative effect on financial performance. Additionally, insiders may seek to overinvest in CSR for their private benefits, because it improves their reputations as good global citizens and they bear little of the costs of doing so (Barnea & Rubin, 2010). This could potentially reduce the financial performance of a firm. Additionally, Di Giuli and Kostovetsky (2014) state that firms with Democratic rather than Republican founders, CEOs and directors are more likely to score high on CSR. Moreover, firms that are headquartered in a Democratic rather than a Republican-‐leaning state score higher on CSR (Giuli & Kostovetsky, 2014). The latter has no potential impact on financial performance. Therefore, it can be used as an instrument for the IV and TE regression.
This paragraph shows the reasoning linked to investors. CSR mitigates to stock price
crash risk (Kim, Li & Li, 2014). The cost of equity is lower for firms that have better CSR scores (El Ghoul, Guedhami, Kwok & Mishra, 2011). Furthermore, according to El Ghoul et al. (2011), participation in tobacco and nuclear power increases firm’s cost of equity. Tobacco and nuclear power can be seen as ‘sin’ industries, so industries that are not social responsible. Therefore, their research supports the arguments that socially responsible firms have higher valuation and lower risk. Additionally, Goss and Roberts (2011) study the link between CSR and bank debt. Firms that have social responsibility concerns pay between seven and eighteen basis points more than firms that have social responsibility strengths. This suggests that it decreases financial performance, because the cost of debt is higher.
The last set of reasoning in this literature review relates to society. Allen, Carletti and Marquez (2007) establish that societies with stakeholder-‐oriented firms (more socially responsible firms) have higher prices, lower output and can have greater firm value. If a firm has lower output it can be at the expense of the shareholder. However, they use a simple model for the product market, so their results must be interpreted with care. Furthermore, according to Baron (2001) CSR can be the result of pressures from society. Liang & Renneboog (2017) support his. They state that the reason why a firm is socially responsible relates to regulations, institutional arrangements, and societal preferences and do not link it to financial performance.
Deng et al. (2013) provide two opposite views on CSR in relation to firm
performance: the stakeholder value maximization and shareholder expense view. The stakeholder value maximization view suggests 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. According to the shareholder expense view managers engage in socially responsible activities to help other stakeholders at the expense of shareholders. For example, when managers adopt pollution control standards that are too stringent compared with those that are implemented by their competitors. This could lead to a competitive disadvantage by forcing the firm to spend too many resources on nonproductive CSR projects, which leads to less profits and reduced shareholder wealth (Deng et al., 2013).
In the next paragraphs the previous literature of this section is linked to the stakeholder maximization and shareholder expense view. A graphical overview can be found in Appendix A. There is support for both views. To start with the reasoning linked to the customer stakeholder. CSR leads to a competitive advantage, an increase in customers, an increase in sales and it prevents loss in sales. The implementation and investment argumentations related to employees support also the stakeholder value maximization view. Studies suggest that CSR could be a screening device, causes workers to demand lower wages, leads to workers exerting higher effort and attracts and retains good employees.
The implementation reasons related to the managers and large blockholders are
backed by the shareholder expense view. CSR leads to collusion with non-‐shareholder stakeholders and overinvestment for private benefits. Furthermore, Democratic rather than
linked to financial performance and therefore is not categorized in one of the two views. The implementation and investment argumentations related to investors support the stakeholder value maximization. CSR leads to mitigation to stock price crash risk, lower cost of equity, higher valuation, lower risk and lower cost of debt.
The argumentation linked to society supports the stakeholder value maximization
and shareholder expense view. On the one hand, CSR leads to higher prices and could lead to greater firm value. On the other hand, CSR could cause lower output. Furthermore, society pressures, regulations, institutional arrangements and social preferences cause the implementation and investment in CSR, however this does not lead to a change in financial performance and therefore is not linked to one of the two views.
2.3 Endogeneity issues
Endogeneity issues are important in this field of research, because neglected endogeneity issues causes potential biased results. Several papers neglect these issues (e.g. Aupperle et al. (1985), Dowell et al. (2000), Fisman et al. (2006), Hong and Kacperczyk (2009) and McGuire et al. (1988)). Therefore, this section introduces possible endogeneity before discussing empirical studies that study the effect of CSR on financial performance. According to Jiao (2010) there are three types of endogeneity problems: omitted variable bias, reversed causality bias and causality that runs in both directions. An instrumental variable approach could help to solve these issues. Furthermore, there could also be a potential self-‐ selection bias. Research suggests that this is not taken into account in previous literature that studies the effect of CSR on long-‐term financial performance. This sections presents the endogeneity issues one by one.
Firstly, omitted variable problem has an influence on the results. Exogenous
characteristics that vary across entities in observable and unobservable ways determine both CSR and financial performance (Jiao, 2010). CSR could influence factors that impact financial performance, for instance managerial decisions. If the unobservable firm characteristics also affect managerial decisions and financial performance, CSR becomes an endogenous variable for financial performance (Jiao, 2010). Secondly, causality that runs in both directions influences the results. The recursive effects could disentangle the performance effects of CSR from the impacts of firm performance on CSR with an ordinary least squares (OLS) regression (Jiao, 2010), so also for a FE regression.
Thirdly, the reversed causality issue is one of the endogeneity issues. The relationship between CSR and financial performance is refers to ‘doing well by doing good’. However, there are also studies that consider the inverse, ‘doing good by doing well’ (Liang & Renneboog, 2017). They examine whether it is only well-‐performing firms that can afford to invest in CSR. For instance, Hong, Kubik and Scheinkman (2012) show that financial constraints are an important driver of corporate goodness. They conclude that firms are more likely to do good when they do well. Several researches try to solve the previous endogeneity problems by using instruments for CSR (e.g. Cornett et al. (2016), Deng et al. (2013) Jiao (2010) and Jo and Harjoto (2011)).
Lastly, the self-‐selection bias could cause biased results. It is possible that firms with
high value tend to engage in CSR activities for reasons unrelated to the wellbeing of stakeholders and that the control variables do not capture this information. For example, if a typical firm engages in CSR activities based on some expected benefit in financial performance, a normal OLS regression will not correctly measure the effect of CSR. Allayannis, Lel and Miller (2012) address this problem by introducing a TE model with full maximum likelihood estimation. They study a different topic than this thesis, but they study a causal relationship on financial performance. With the TE model the explanatory variable is binary, in this thesis it represent either engaging or not engaging in CSR activities. This binary variable is modeled as a function of instruments that shows to be important factors to engage in CSR.
2.5 CSR and financial performance
Researches that study the effect of CSR on financial performance can be classified in two groups; studies that focus on short-‐term financial performance and on long-‐term financial performance. Event study methodology mostly researches short-‐term firm performance when firms engage in either responsible or irresponsible actions (for example, Hannon and Milkovich (1996) and Flammer (2015)). Hannon and Milkovich (1996) and Flammer (2015) find that CSR engagement has a positive announcement effect on share prices. In this research the focus is on long-‐term financial performance. Long-‐term performance is interesting to research, because long-‐term performance ensures continuity of the firm. Furthermore, to also research short-‐term performance is beyond the scope of this research.
Therefore, this section of the literature review focuses on studies that research long-‐term financial performance.
Studies that focus on long-‐term financial performance can again be classified in two groups: studies that focus on operating performance and market performance. ROE, ROA and sales growth are common measures of operating performance (for example, Aupperle et al. (1985), Cornett et al. (2016), Deng et al. (2013), Fisman et al. (2006) and McGuire et al. (1988)). Tobin’s Q is a common measure for market performance (for example, Cornett et al. (2016), Dowell et al. (2000), Fisman et al. (2006) Jiao (2010) and Jo and Harjoto (2011)).
Margolis et al. (2007) and Orlitzky, Schmidt and Rynes (2003) conduct a meta-‐ analysis. Both researches suggest that overall CSR is likely to pay off. However, there are some conflicting results, therefore the next paragraphs highlights the main empirical studies on this subject. The following paragraphs focus on papers that research operating performance. Thereafter, it focuses on papers that study market performance. A graphical overview of the next paragraphs can be found in Appendix B.
The next paragraphs show researches that focus on operational performance. Deng et al. (2013) research whether CSR creates value for acquiring firms. High CSR acquirers have a larger increase in post-‐merger long-‐term operating performance than low CSR acquires. Also when looking at environmental, social and governance ratings CSR for banks, there is a reward for being socially responsible, because they are more likely to have a higher ROE (Cornett et al., 2016). Both studies account for endogeneity issues with an IV regression. This value creation is in line with several other papers, such as Berman, Wicks, Kotha and Jones (1999), Griffin and Mahon (1997), Mahoney and Roberts (2004), Ruf, Muralidhar, Brown, Janney and Paul (2001) and Waddock and Graves (1997). Similar as in this thesis, they use KLD ratings to construct the CSR variable. They find that CSR has a positive influence on operating performance, such as ROA and ROE. These papers support the stakeholder value maximization view described at section 2.2.
However, there is also literature that contradicts previous literature. Blackburn, Doran and Shrader (1994), Dooley and Lerner (1994), O’Neill, Saunders and McCarthy (1989) and Turban and Greening (1996) find that CSR initiatives have a negative influence on ROA. Cowen, Ferreri and Parker (1987) find a similar result for ROE. These articles support the shareholder expense view described at section 2.2. Aupperle et al. (1985) even suggest that
there is no relationship between CSR and profitability. However, all these studies do not take into account endogeneity problems.
In the next two paragraphs there is a discussion of papers that focus on market
performance. Overall, there is a positive relation between CSR and market performance. In this paper Tobin’s Q is used as a measurement of market performance, therefore these paragraphs focus on papers that also use Tobin’s Q. Dowell et al. (2000) and Hong and Kacperczyk (2009) find a positive relation, nonetheless these paper do not account for endogeneity issues. However, Cornett et al. (2016) and Jiao (2010) also find a positive relation, they use an IV regression to address possible endogeneity issues. This supports the stakeholder value maximization view. Cornett et al. (2016) use political environment, percentage deposits low income, percentage of female and minority directors and headquarter (HQ) in green city as instruments for CSR. Jiao (2010) use an indicator for positive earnings in the previous year and monitoring from activist public pension funds in the previous year as instruments.
However, there is some literature that contradicts previous statements. These
studies do not account for endogeneity issues and support the shareholder expense view. According to Hillman and Keim (2001) corporate resources for social issues that do not relate to primary stakeholders may not create value for shareholders. Additionally, Fisman et al. (2006) find that CSR and Tobin’s Q do not have a positive relationship in certain industries. Only industries with high intensity advertising and high competition show a positive relationship, due to the signal value of CSR expenditures. The transparency and therefore customer awareness is greater for such industries. According to Servaes and Tamayo (2013) this leads to an increase in customers and sales, which has a positive effect on financial performance. This thesis also researches this component, more on this can be found in the methodology.
2.5 Financial crisis and CSR
This section of the literature review emphasizes on the relationship between CSR and the financial crisis. The first paragraph deals with the CSR performance in relation to the crisis. The second and third paragraph is about the relation between CSR and financial performance in the context of the crisis.
Giannarakis and Theotokas (2011) find that CSR performance increased before and during the financial crisis, except for the period 2009-‐2010. They state that companies increase their CSR levels to regain the lost trust in businesses. Additionally, Jacob (2012) shows that the recent financial crisis has a clear impact on CSR initiatives; this is due to the pressure that they had in order to survive. The main consequences of the crisis were massive layoffs and expenditure cuts on community involvement programs (Jacob, 2012). Moreover, Cornett et al. (2016) see that the recent financial crisis has an influence on the CSR initiatives of the largest banks. The largest banks see both a steep increase in CSR strengths and a steep drop in CSR concerns after the financial crisis.
The effect of CSR on financial performance could be different during the financial
crisis. On the one hand, during the financial crisis there is a strong pressure to reduce the costs and therefore the market value of a company that engages in CSR decreases. On the other hand, engaging in CSR activities could lead to a competitive advantage during the financial crisis, by increasing the firm’s reputation and gaining the trust of the local community (Sitnikoc & Bocean, 2017).
Cornett et al. (2016) analyze banks’ CSR performance in relation to financial
performance in a context of the recent financial crisis. They find that banks get in general a reward for being socially responsible. ROE relates positively and significantly to CSR scores. Furthermore, this holds after alternative definitions of CSR engagement, financial performance and size cutoffs. Lome et al. (2016) research the link between R&D expenses and growth in turbulent times. They state that cutting down potentially profitable R&D investments, when competitors do not, could have serious effects on performance during the recession and recovery period. So, keeping high R&D expenses during the financial crisis could benefit financial performance. A similar relation could exist for CSR initiatives, just as R&D expenses CSR is also a long-‐term investment.
3. Hypotheses
This part emphasizes on the hypotheses of this research and the contribution of this research in relation to other papers. Solving the possible endogeneity problems, including the self-‐selection bias, is an important contribution in relation to previous papers (for more information about this, see chapter four).
The first hypothesis is about the effect of CSR on operating performance. Below the formal formulation can be found. ROA, ROE and sales growth measure operating performance. There are only a few papers that combine measures of operating performance, most of the papers use only one of them or a combination of two (such as, Aupperle et al. (1985), Berman et al. (1999), Cornett et al. (2016), Deng et al. (2013), Griffin and Mahon (1997), Mahoney and Roberts (2004), McGuire et al. (1988), Ruf et al. (2001) and Waddock and Graves (1997)).
𝐻!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑛𝑜 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
𝐻!!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑎 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝐻!!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑎 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑛𝑔 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
The second hypothesis investigates the effect of CSR on market performance. Below
the formal formulation can be found. The expectation is that for this and the previous hypothesis hypothesis one A is holds. Margolis et al. (2007) and Orlitzky et al. (2003) conduct a meta-‐analysis. Both researches suggest that overall CSR is likely to pay off.
𝐻!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑛𝑜 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
𝐻!!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑎 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑎𝑟𝑘𝑒𝑡 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
𝐻!!: 𝐶𝑆𝑅 ℎ𝑎𝑠 𝑎 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑛 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
The third hypothesis explores the different effect of CSR on financial performance in context of the recent financial crisis. There is not much literature that researches the effect of CSR on financial performance in context of the financial crisis. Cornett et al. (2016) research this relationship, however they only investigate banks. The formal formulation of the hypothesis is below. The assumption is that hypothesis one A holds based on the research of Lome et al. (2016). They research the effect of R&D expenses on financial performance on the financial crisis. The importance of R&D activities increases during a financial crisis, the effect on financial performance for firms who devote a large amount of resources to R&D during a financial crisis is even stronger than in a period of normal growth. Just as R&D expenses CSR is a long-‐term investment and could pay off during a financial crisis. Moreover, CSR could create a competitive advantage during the crisis by increasing
𝐻!: 𝑇ℎ𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑐𝑟𝑖𝑠𝑖𝑠 𝑑𝑜𝑒𝑠 𝑛𝑜𝑡 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒 𝑡ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
𝐻!!: 𝑇ℎ𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑐𝑟𝑖𝑠𝑖𝑠 𝑝𝑜𝑠𝑖𝑡𝑖𝑣𝑒𝑙𝑦 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒𝑠 𝑡ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝐻!!: 𝑇ℎ𝑒 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑐𝑟𝑖𝑠𝑖𝑠 𝑛𝑒𝑔𝑎𝑡𝑖𝑣𝑒𝑙𝑦 𝑖𝑛𝑓𝑙𝑢𝑒𝑛𝑐𝑒𝑠 𝑡ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒
The last hypothesis focuses on the different effect of CSR on financial performance
for high advertising industries. The addition of this hypothesis is based on the research of Fisman et al. (2006). They find that CSR has a different effect on financial performance for high intensity advertising industries. This thesis distinguishes itself in comparison to other papers by adding this hypothesis. Besides the paper of Fisman et al. (2006) existing literature suggests that this effect is never researched. This thesis contributes in comparison to Fisman et al. (2006) because they do not try to solve possible endogeneity issues and this paper does. Furthermore, Fisman et al. (2006) research companies between 1991 and 2003 and this thesis researches companies between 2003 and 2013. The formal formulation is below. Hypothesis one A is expected to be true, because the transparency and therefore customer awareness of CSR is greater in high intensity advertising industries, which could benefit financial performance. 𝐻!: 𝑇ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑖𝑠 𝑡ℎ𝑒 𝑠𝑎𝑚𝑒 𝑓𝑜𝑟 ℎ𝑖𝑔ℎ 𝑖𝑛𝑒𝑛𝑠𝑖𝑡𝑦 𝑎𝑑𝑣𝑒𝑟𝑡𝑖𝑠𝑖𝑛𝑔 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠 𝐻!!: 𝑇ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑖𝑠 𝑔𝑟𝑒𝑎𝑡𝑒𝑟 𝑓𝑜𝑟 ℎ𝑖𝑔ℎ 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 𝑎𝑑𝑣𝑒𝑟𝑡𝑖𝑠𝑖𝑛𝑔 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠 𝐻!!: 𝑇ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 𝑜𝑓 𝐶𝑆𝑅 𝑜𝑛 𝑓𝑖𝑛𝑎𝑛𝑐𝑖𝑎𝑙 𝑝𝑒𝑟𝑓𝑜𝑟𝑚𝑎𝑛𝑐𝑒 𝑖𝑠 𝑙𝑒𝑠𝑠𝑒𝑟 𝑓𝑜𝑟 ℎ𝑖𝑔ℎ 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 𝑎𝑑𝑣𝑒𝑟𝑡𝑖𝑠𝑖𝑛𝑔 𝑖𝑛𝑑𝑢𝑠𝑡𝑟𝑖𝑒𝑠 4. Methodology
This chapter describes the methodology of the research. There are three different regression methods: fixed effects, instrumental variable and treatment effects regression. The last two address possible endogeneity problems. The first section of this chapter is about the main regression and the second section is about the econometric methods.
4.1 Main regression
To compute the main regression, in order to research the effect of CSR on financial performance a CSR proxy, measures of financial performance and control variables are necessary. This section starts with the CSR proxy. Thereafter, there is a description of the different measures of financial performance, the control variables and the instruments.
The MSCI database is used to construct the CSR variable. Several recent papers use this database to quantify corporate social responsibility (For instance, Cornett et al., (2016), Deng et al., (2013), Jiao (2010), Jo and Harjoto (2011), and Krüger (2015)). The MSCI database classifies events into 7 issue areas: community, corporate governance, diversity, employee relations, environment, human rights and product. For each issue area it defines the number strengths and/or concerns, so the MSCI database utilizes a number representation of the ESG ratings. This is on a yearly basis.
To illustrate the intuition behind the number representation an example follows. For the human rights area, it receives strength(s) as a company establishes relations with indigenous peoples near its proposed or current operations that respect the sovereignty, land, culture, human rights and intellectual property (MSCI ESG Research Inc., 2015). The amount of strengths is one if the company establishes some relations and it could be two if the company has many relationships in this area. However, it could also be zero if the company has no relations in this field. For concerns it is the other way around.
According to Hong, Kubik and Scheinkman (2012), the corporate governance area differs from the other areas, i.e. community, diversity, employee relations, environment, human rights and product. Furthermore, improving corporate governance does not necessarily require monetary investments (Krüger, 2015). Additionally, by deleting the corporate governance issue the CSR variable focuses on the firm’s primary non-‐shareholding stakeholders (Krüger, 2015). Therefore, the corporate governance issue is excluded in the standard CSR variable. Nonetheless, it is included in the robustness checks for CSR. The construction of the CSR variable assigns equal importance to all categories. All concerns minus all strengths represent the CSR variable (Cornett et al., 2016).
The MSCI database contains data points for approximately 3,000 US companies. These are the top 3,000 US companies by market capitalization, as of December of each year (MSCI ESG Research Inc., 2015). The starting point is 2003, because the database expands from 1,000 to 3,000 companies in that year. The data of the strengths and concerns are available until 2013; therefore this research is based on the years 2003 till 2013.
The measurement of financial performance is split between operating and market performance. The use of ROE, ROA and sales growth is common for operating performance measures (For example, Aupperle et al. (1985), Cornett et al. (2016), Deng et al. (2013),
