The effect of corporate social
responsibility on financial performance
Kyra Berendsen 10362142 Supervisor: Drs. P.V.Trietsch BSc Economics & Business,
2
Statement of originality
This document is written by Kyra Berendsen 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
3
Abstract
This thesis tests the relationship between corporate social responsibility (CSR) and corporate financial performance (CFP) of firms included in the MSCI database using a random effects panel data model. Results from current literature are mixed mainly because of the different methods employed and misspecification. Also, the increased popularity of CSR presented an unique opportunity to re-test this relationship using recent data. The hypothesis is that there exists a positive relationship between CSR and CFP. CSR was measured by the assessment of the MSCI database and CFP by both accounting measures as well as a market measure. Based on existing literature risk, size, and industry were added as control variables. A second regression with reduced sample also included R&D intensity as a control variable. For CSR as well as the control variables a time lag was used to take into account the possibility of reverse causality. Results show a non-significant relationship for all but one regression, rejecting the hypothesis.
4
Table of contents
1 Introduction 5
2 Theoretical background and literature review 7
2.1 Defining CSR 7
2.2 Theories explaining firm behavior 7
2.3 Measuring CSR 9 2.4 Measuring CFP 10 2.5 The CSR-CFP relationship 11 2.6 Hypothesis development 15 3 Methodology 16 3.1 Measurement CSR 16 3.2 Measurement CFP 16 3.3 Control variables 17 3.4 Empirical models 19
3.5 Sample and data collection 19
4 Results 21 4.1 Descriptive statistics 21 4.2 Assumptions 22 4.3 Regressions 22 5 Conclusion 26 5.1 Summary 26
5.2 Limitations and recommendations 26
6 References 28
5
1 Introduction
The increased popularity of corporate social responsibility (CSR) is clearly visible: firms reporting on their CSR activities grew from 9 percent in 2008 to 51 percent in 2013 (KPMG, 2013), 67 percent of employees prefer to work for socially responsible companies, and 55 percent of consumers are willing to pay extra for socially responsible products (Nielsen, 2014). The dominant view has always been that engaging in CSR has a negative effect on corporate financial performance (CFP), which measures the ability of a firm to generate revenues, (Carroll, 2008) but why then is CSR so popular in recent years?
Previous research shows mixed results. Margolis and Elfenbein (2007) conducted a meta-analysis of all studies between 1972 and 2007 researching the CSR-CFP relationship. Overall they found a positive relationship but they also found it hard to generalize results as most
research used different measurements for the variables included in the studies. Another common problem was that some studies did not account for endogeneity (Surroca et all, 2010). Both Garcia-Castro et all (2010) and Schreck (2011) did account for endogeneity and concluded that there is a non-significant relationship. Oberndorfer et all (2013) conducted an event study after inclusion in a sustainability index using abnormal returns to measure financial performance and found a negative relationship. To be able to get a better understanding of the true CSR-CFP relationship it is important that all variables are clearly defined and included in the regression model and that the cause of this relationship is explained.
This thesis will study the effect of CSR on CFP. The aim is to answer the following question: what is the relationship between CSR and CFP? This thesis is relevant for two reasons. The increased popularity of CSR under firms may result in different outcomes when the CSR-CFP relationship is tested than in prior research when CSR was not as popular. This thesis also takes into account the possibility of reverse causality by employing the time lag approach.
To be able to answer the main research question five sub-questions have been formulated. 1. What is CSR and how can it be measured?
2. What is CFP and how can it be measured? 3. What causes a positive (negative) relationship? 4. What causes the possibility of reverse causality? 5. What are important control variables?
6
are divided into two sets. The first set of regressions will test the relationship of CSR with Tobin’s Q, ROA, ROE, and ROS using the three most used control variables in prior research: size, risk, and industry. The second set of regressions will add R&D intensity as a control variable, but has a reduced sample because of the lack of available information on R&D expenditures.
CSR data will be gathered from the MSCI (formerly KLD) database. Financial data will be retrieved from Compustat. Data for the CSR scores and the control variables will be from 2009 until 2013. The dependent variables consists of data from the years 2010 until 2014. It was decided to use a 5 year sample because of the increased popularity in recent years. 2013 was chosen as the final year because the MSCI database changed their scoring system in 2014 . The first set of regressions consists of 7566 observations. For the second set of regression the sample consisted of 4535 observations.
This thesis is organized as follows. Chapter 2 presents the theoretical background and the empirical literature results, concluding with the development of the hypothesis of this thesis. The methodology will be described in chapter 3. In chapter 4 the results are presented. Finally, chapter 5 will conclude this thesis.
7
2 Theoretical background and literature review
This chapter will discuss the theoretical background of CSR followed by a literature review. To be able to discuss the background CSR must first be defined. This is followed by a section of theories that explain that a firm needs to interact with the outside world. Then the different measurements for CSR and CFP are described. Finally an overview of the current knowledge of the CSR-CFP relationship is given.
2.1 Defining CSR
While there is still no common definition of CSR, there are some common aspects to most
characterizations. The firm does more than just the legal requirements in order to benefit not only the financial owners of the firm but also other stakeholders and society as a whole (Idowu et all, 2015). The European Commission (2008) defines CSR as “a concept whereby companies
integrate social and environmental concerns in their business operations and in their interaction with their stakeholders on a voluntary basis”. This definition will be used throughout this thesis.
The overall concept of CSR can be divided into seven different dimensions: environment, corporate governance, community, diversity, product, employee relations, and human rights. These categories represent the different stakeholders a firm has to deal with. Environment has to do with how much a firm tries to prevent environmental damage. Corporate governance refers to the way in which a firm is controlled and directed and important topic include management compensation and transparency. Community refers to the actions taken for the social group to which a firm belongs and include charitable giving and volunteering programs. Diversity refers to the variety in a firm on different levels such as woman and other minorities. The produced good and its qualities are part of the product category. How a firm deals with employees are part of the employee relations category. The last category, human rights, encompasses how a firm handles the fundamental rights of people.
2.2 Theories explaining the use of CSR
In this paragraph the theories that explain why firms engage in CSR are described. It is clear that companies would only engage in CSR if they can benefit from it and these theories explain how that could be possible. Here are discussed the legitimacy and stakeholder theory. Both theories are part of the system-oriented perspective, which states that the firm and society influence each
8
other bidirectional (Deegan, 2006). The most important aspect for firm survival according to these theories are the perceptions held by society (Deegan, 2006).
2.2.1 Legitimacy theory
According to legitimacy theory, a firm is bound to a social contract with society (Guthrie & Parker, 1989). Suchman (1995) states that “legitimacy is a generalised perception or
assumption that the actions of an entity are desirable, proper, or appropriate within some
socially constructed system of norms, values, beliefs, and definitions”. To be approved by society,
a firm’s actions must be seen as legitimate (Suchman, 1995).
When the believes of a society about how a firm should act and the perceived behaviour of the firm are in contrast with each other, a legitimacy gap exists (O’Donovan, 2002). Reasons for a legitimacy gap to arise may be: a change in firm behaviour, while expectations stay the same; expectations change, while firm behaviour stays the same; or both change, but in different directions (O’Donovan, 2002). Legitimacy gaps can be reduced by attempting to alter the
perceptions by society, or by changing firm behaviour. 2.2.2 Stakeholder theory
Stakeholder theory is related to legitimacy theory but while firms have a contract with society as a whole in legitimacy theory, firms have a contract with every stakeholder separately in
stakeholder theory (Guthrie & Parker, 1989). Strategic objectives of the firm can only be achieved if certain stakeholder demands are met (Roberts, 1995). The more important the group of stakeholders, the more important it is to meet these demands.
Freeman (1984) defines stakeholders as “any group or individual who can affect or is
affected by the achievement of the firm’s objectives”. There are many different stakeholder
groups, each with their own demands, and each with their own behaviour towards the firm. According to Wood and Jones (1995) these stakeholders fill in three distinct roles:
1. Stakeholders define expectations about good and bad firm behaviour 2. Stakeholders experience this behaviour
3. Stakeholders evaluate how much this behaviour has met the expectations 2.2.3 The different theories and their relation with CSR
The former two theories described state that a firm cannot put all their focus on the economic aspect of business. According to legitimacy theory a firm has a contract with society to which it
9
should comply to guarantee firm survival. Engaging in CSR is a way to meet the requirements of that contract and to be approved by society. Stakeholder theory refers to the contracts a firm has with each of their stakeholders. Each of these stakeholders have different demands but the different stakeholder groups are not equally important. The firm can change the level of CSR activities in accordance with the importance and demands of each group.
2.3 Measuring CSR
One of the obstacles faced in the literature is the measuring of CSR. CSR can be measured in a variety of ways, each with their own implications for research. This section presents an overview of the measurements used in literature, using the five categories presented by Soana (2011). The first measure is the content analysis. With a content analysis publicly available information, such as an annual report, are scanned for reported CSR activities. According to Soana, this can be as simple as a count of CSR related words. This method implies that reported CSR activities are a good proxy for actual CSR activities. Drawback of this method is that these documents may be biased as the firm composes these documents themselves (Waddock & Graves, 1997).
The second measure is questionnaire surveys. With this method, researchers sent out specific questionnaires about CSR activities to multiple company officials. Their responses will then be used to determine the corporate social performance of that firm. This method is similar to the first method in that they measure CSR by reported activities. Disadvantage of this method is that results may be biased as company officials may not always be honest and may have different perceptions than outsiders (Soana, 2011).
The third measure is reputational measures. Companies are ranked by social reputation where firms with higher positions have a better social reputation. One example of a social rating mentioned by Soana is the Corporate Reputational Index by Fortune Magazine. This method insinuates that a firm’s reputation is a good measure of CSR. Therefore these ratings may be a better measure of good overall management instead of CSR (Waddock & Graves, 1997). The fourth category is one-dimensional indicators. One CSR dimension is used to
measure overall CSR, such as environmental pollution. Carpentier and Suret (2013) for example examine the returns on the stock market after environmental accidents. Consequences for
research may be different than when multiple dimensions are used (Soana, 2011).
10
on multiple CSR dimensions, from which an overall CSR score is calculated. An example is the MSCI (formerly KLD) database, which measures different CSR dimensions such as employee relations, corporate governance, product, environment, community, human rights, and diversity. This database is the one mostly used in previous research (i.e. Waddock & Graves, 1997;
Schreck, 2011; Jiao, 2010; Garcia-Castro et all, 2010).
The MSCI database first started collecting the data of 650 firms in 1991. Since then the coverage was expanded and there are currently 3000 firms included in this database.
Professionals rate these firms on a binary scale using multiple sources of information. A score of 0 means a neutral score, and a score of 2 is a significant weakness or strength score. Firms are scored on multiple aspects, summarized in seven overall categories or CSR dimensions. The categories that MSCI employs are environment, diversity, product, corporate governance,
community, employee relations, and human rights (for a complete overview of the categories and its sub-scores see appendix). MSCI also has six controversial concern categories. Firms from the six controversial concern categories were excluded from this dataset, as it is believed that they may result in a bias in the regression, as they only have neutral or negative influences (Van der Laan et all, 2008).
According to Waddock and Graves (1997) the MSCI database is better than other CSR measures because of four reasons. The first reason is the wide range of companies MSCI includes in their database. The scoring on different CSR categories is the second reason. Third is that the research and scoring is being done by independent researchers. The final reason is that each company rated must pass the same criteria and the information needed to give a score is gathered not only from the firm itself but also from other, external sources.
2.4 Measuring CFP
Measuring corporate financial performance is more straightforward than measuring CSR.
Financial performance measures can be divided into two groups: accounting-based measures and market-based measures (Margolis & Elfenbein, 2007). Some researches use accounting measures (i.e. Waddock & Graves, 1997; Tsoutsoura, 2004), others use market measures (i.e. Fisher-Vanden & Thorburn, 2011; Schreck, 2011), and others use both (i.e. Garcia-Castro et all, 2010). The first group consists of the accounting-based measures, which is the group most used to measure financial performance (Margolis & Elfenbein, 2007), and for which the (slightly
11
positive) relationship is most clear (Orlitzky et all, 2003). Accounting based measures contain ratios such as ROE, ROA, and ROS (i.e. Margolis & Elfenbein, 2007). Return on assets is a profitability measure indicating the effectiveness of generating assets into profits. Return on equity indicates the effectiveness of the profit that is generated with the shareholder’s money. How much profit per sale a company generates is the return on sales measure.
One advantage of using accounting based measures is that these measures are easily available and that these measures are related to economic returns (Richard et all, 2009). Disadvantages of using accounting-based measures are for instance, that they only reflect historical performance and that it may be easier to manipulate (Gentry & Shen, 2010).
The second group are market-based measures, which contains for example Tobin’s Q and stock returns. Tobin’s Q is the ratio of a firm’s market value to the replacement value of its assets (Jiao, 2010). Compared to accounting-based measures, market measures are more forward
looking (Richard et all, 2009). According to the efficient market hypothesis the market value of a firm represents all relevant information including expectations such as future cash flows
(Tsoutsoura, 2004). Market measures also value intangibles more than accounting measures (Richard et all, 2009). Drawback of using market-based measures is that share prices are affected by more than just firms’ actions (Richard et all, 2009), but this can be controlled for by
benchmarking.
2.5 The CSR-CFP relationship
There is still debate about the CSR-CFP relationship in literature. The main reasons for this are problems of measuring the different variables and misspecification problems. As discussed in the two sections before both CSR as well as CFP can be measured in multiple ways. Because of this lack of consensus there are still two opposing groups defending both sides of the relationship. A fundamental question that needs to be asked in regard to this discussion is if the focus of the firm should be on shareholder value or social value (Renneboog et all, 2008).
2.5.1 Positive relationship
Proponents of CSR argue that firms would not engage in CSR if it only enhanced costs as that would not be sustainable for the firm (Tsoutsoura, 2004). This group puts emphasis on the stakeholder demands as stressed by stakeholder theory. Benefits from taking care of stakeholders are for example increasingly motivated employees and a higher customer demand (EU, 2008).
12
Firms will also earn a good moral reputation which can be valuable because it may affect the actions stakeholders take (Godfrey, 2005).
This was supported by Waddock and Graves (1997). They used accounting measures to study the CSR-CFP relationship and found a positive outcome. They argued that this positive outcome may arise because of the linking of social performance to the stakeholder interests. They also add that firms may not be committed to CSR at all but only want to appear committed to have all the reputational benefits.
The transaction cost economics perspective also emphasizes a positive relationship between CSR and CFP. This perspective argues that firms engage in CSR to signal their willingness to cooperate (Ruf et all, 2001). Firms cooperation might prevent future costs from lawsuits or large fines for polluting (Tsoutsoura, 2004). This was confirmed by the meta-analysis by Margolis and Elfenbein (2007). Using all studies between 1972 and 2007 they concluded that it was costly for a firm to be associated with scandals and overall they found a positive
relationship between CSR and CFP.
The resource based view also emphasizes a positive relationship. This view argues that firms can gain competitive advantages from their own personal, internal capabilities (Oberndorfer et all, 2013) and engaging in CSR is seen as a strategic investment (Ruf et all, 2001). According to Godfrey (2005) the resource based view argues that CSR can be seen as an insurance. He states that many of the firm’s resources are relation based and negative firm events may harm these relationships. The negative financial impacts of these relationships cannot be insured for and thus the good reputation from CSR acts as an insurance as CSR increases a firm’s reputation. Jiao (2010) researched the relationship between CSR and Tobin’s Q as proxy for financial performance. His study supported the resource based view as it was concluded that engaging in CSR creates intangible resources for the firm. He found a positive relationship between CSR and CFP and the most important dimensions were employee relations and environmental issues. 2.5.2 Non-significant relationship
A non-significant relationship indicates that there is no relationship between CSR and CFP. According to Tsoutsoura (2004) reason for this may be that there are a lot of intervening
variables that influence CSR as well as CFP. Surroca et all (2010) confirmed this. They included different intangibles as control variables and concluded that there was no significant relationship.
13
They argued that there was merely an indirect relationship which was mediated by the inclusion of the intangibles.
2.5.3 Negative relationship
The group hypothesizing a negative relationship are mostly focusing on defending
shareholder’s rights as shareholders are the legal owners of the firm. This group sees engaging in CSR as merely cost enhancing. Costs associated with engaging in CSR are for example the purchase of new equipment or the implementation of new management structures or quality controls (Tsoutsoura, 2004). The sole focus of the firm should be shareholder’s wealth. This is supported by Jensen’s (2002) shareholder-value theory. This theory states that if a company maximizes its own shareholder value, total welfare will be maximized as well.
Oberndorfer et all (2013) used an event study to investigate the effect on returns after a firm was announced to be included in a sustainability index (in their research they used the DJSI world and STOXX index) on the German stock market. Inclusion resulted in negative stock returns suggesting that shareholder wealth is reduced when firms achieved a CSR level high enough to be included in a sustainability stock index.
Another theory important for this group is the agency theory. Agency theory explains the information asymmetry between managers, who lead the firm, and the shareholders, who own the firm. This information asymmetry might eventually lead to an over-investment hypothesis. Managers are likely to over-invest in CSR to improve their own reputation and enhance their career opportunities (Jo & Harjoto, 2011).
Fisher-Vanden and Thorburn (2011) researched the effect of voluntary corporate
environmental initiatives on shareholder wealth, by examining the stock returns after firms joined two different environmental programs. They argue that one reason for a firm to join these
initiatives is the reduction in information-asymmetry in committing to being responsible. Inclusion resulted in negative results. These negative results were stronger for weak corporate governance firms, confirming the over-investment hypothesis.
14
Positive Negative Non-significant
Main focus group Stakeholders Shareholders Neither Arguments Reputational benefits,
transaction cost economics, resource based view
Increasing costs, shareholder value theory, agency theory
There exist a lot of intervening variables
Empirical evidence Waddock & Graves (1997), Margolis & Elfenbein (2007), Jiao (2010)
Oberndorfer et all (2013), Fisher-Vanden & Thorburn (2011)
Surroca et all (2010)
Table 1: overview of the discussions of the CSR-CFP relationship and the empirical results that support either side.
2.5.4 Reverse causality
Some recent research suggests a reverse causality relation between CSR and financial
performance (i.e. Margolis & Elfenbein, 2007; Schreck, 2011; Jo & Harjoto, 2011; Garcia-Castro et all, 2010; Jiao, 2010). This causality means that CSR not only affects financial performance, but also that financial performance affects CSR. When there exists a reverse causality
relationship, results will be biased if this causality is not accounted for.
Two theories might explain this reverse causality: slack resources and good management (Schreck, 2011). According to slack resources theory firms with higher financial performance might have more (financial) resources available to invest in CSR (Waddock & Graves, 1997). Firms with higher financial performance are thus more likely to engage in CSR instead of the other way around. Good management theory explains that both good social performance and good financial performance are achieved because of good management (Schreck, 2011). Because the firm is managed well, both the social performance and the financial performance will go up. There have been two solutions put forward to deal with reverse causality and to reduce the bias. The first is the lagged value approach, and the second is the instrumental variable approach (Schreck, 2011). Lagged value uses past CSR scores as an instrument, and instrumental variable approach uses another instrument. Bias is avoided by using instruments which have to meet two conditions: relevance and exogeneity (Schreck, 2011). Relevance means that the instrument is related to CSR and exogeneity means that the instrument is not related to financial performance. Prior studies used the instrumental variables approach to account for endogeneity but they found different results. Jo and Harjoto (2011) used firm age as an instrument and found a positive relationship between Tobin’s Q and CSR. However, Garcia-Castro et all (2010) found no
15
firm’s visibility as instruments. Industry characteristics were included because different industries have different CSR standards. Corporate governance conditions were included because some studies found that they have an effect on the social performance. Because more visible firms receive more pressure to engage in CSR visibility was also included but appeared to be a weak instrument after testing. The results of their study showed a non-significant relationship.
2.6 Hypothesis development
The primary goal of this thesis is to study the relationship between CSR and CFP and contribute to a better understanding of this relationship. The literature review shows that there is still debate about this relationship. This section will discuss this debate shortly and take into account
previous empirical results to eventually state the hypothesis of this thesis.
Two opposing groups are defending either side of the CSR-CFP relationship. The shareholder group sees CSR as merely cost enhancing and argue that the sole focus of the firm should be the shareholders. The stakeholder group argues that the benefits of CSR will be higher if the demands of stakeholders are taken into account.
To end this debate convincing results are needed from the empirical literature but unfortunately results are mixed. According to Margolis and Elfenbein (2007) the overall relationship seems positive, especially for accounting measures. These results may become insignificant when endogeneity is accurately accounted for. Most negative results are found when an event study is conducted using stock returns.
Based on the literature review the following hypothesis will be tested in this thesis:
16
3 Methodology
This chapter will discuss the methodology. First will be explained how CSR and CFP are measured. This will be followed by a description of the control variables. Then the empirical models themselves will be presented. This chapter concludes with the sample and data collection.
3.1 Measurement CSR
Summarizing the section in the literature review, there are five different ways in which CSR can be measured: content analysis, surveys, reputational measures, one-dimensional indicators, and ethical ratings. Because of the time limit and of the advantages an ethical rating will be used.
In this thesis the different dimensions scores will be calculated first. This is being done by adding all the different strengths and then subtracting all the different category weaknesses for all seven dimensions of CSR. This will then be used for the category scores in the different
regressions. The total CSR score is calculated by adding all the dimension scores and will be used to determine if the CSR-CFP relationship is positive, negative, or non-significant. No weighting scheme for the different CSR categories is used. This means that a strength in the environmental category is just as important as a strength in the diversity category. The reason for this is because of simplicity. Waddock & Graves (1997) for example do use a weighting scheme, but they send out questionnaires to different field professionals and to do that is behind the scope of this thesis.
𝐶𝑎𝑡𝑒𝑔𝑜𝑟𝑦 𝐶𝑆𝑅 𝑠𝑐𝑜𝑟𝑒 = ∑ 𝑐𝑎𝑡𝑒𝑔𝑜𝑟𝑦 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ𝑠 − 𝑤𝑒𝑎𝑘𝑛𝑒𝑠𝑠𝑒𝑠
𝑇𝑜𝑡𝑎𝑙 𝐶𝑆𝑅 𝑠𝑐𝑜𝑟𝑒 = ∑ 𝑐𝑎𝑡𝑒𝑔𝑜𝑟𝑦 𝑠𝑐𝑜𝑟𝑒𝑠 7
1
3.2 Measurement CFP
This thesis will focus on both accounting as well as market measures as a proxy for financial performance. ROA, ROE, and ROS will be used as accounting measures and Tobin’s Q will be used as a market measure. In doing so this thesis follows earlier research (i.e. Waddock &
17
Graves, 1997; Aras et all, 2010; Schreck, 2011; Garcia-Castro et all, 2010). The calculation of these measures is as follows:
𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑎𝑠𝑠𝑒𝑡𝑠 (𝑅𝑂𝐴) = 𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑒𝑞𝑢𝑖𝑡𝑦 (𝑅𝑂𝐸) = 𝑛𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑠ℎ𝑎𝑟𝑒ℎ𝑜𝑙𝑑𝑒𝑟′𝑠 𝑒𝑞𝑢𝑖𝑡𝑦 𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑠𝑎𝑙𝑒𝑠 (𝑅𝑂𝑆) = 𝑒𝑎𝑟𝑛𝑖𝑛𝑔𝑠 𝑏𝑒𝑓𝑜𝑟𝑒 𝑖𝑛𝑡𝑒𝑟𝑒𝑠𝑡 𝑎𝑛𝑑 𝑡𝑎𝑥 (𝐸𝐵𝐼𝑇) 𝑡𝑜𝑡𝑎𝑙 𝑠𝑎𝑙𝑒𝑠
Tobin’s Q can be measured in different ways. Sometimes a proxy of market value / total assets is used (i.e. Andersen & Dejoy, 2011), and sometimes a more extensive proxy is used (i.e. Schreck, 2011; Jiao, 2010). There exist a precise formula for Tobin’s Q, but this requires
extensive and difficult to obtain data (Schreck, 2011). Therefore, in this thesis, the more extensive proxy of Tobin’s Q will be used:
𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄
= 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠 − 𝑏𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒 𝑒𝑞𝑢𝑖𝑡𝑦 − 𝑑𝑒𝑓𝑒𝑟𝑟𝑒𝑑 𝑡𝑎𝑥𝑒𝑠 + 𝑚𝑎𝑟𝑘𝑒𝑡 𝑣𝑎𝑙𝑢𝑒 𝑐𝑜𝑚𝑚𝑜𝑛 𝑠𝑡𝑜𝑐𝑘𝑠 𝑡𝑜𝑡𝑎𝑙 𝑎𝑠𝑠𝑒𝑡𝑠
3.3 Control variables
Control variables are important to include in the regression as they may influence the CSR-CFP relationship. These variables are held constant to avoid omitted variable bias and to see the true relationship (Schreck, 2011). In previous research, the most used control variables were industry, firm size, and risk (Margolis & Elfenbein, 2007). With event studies, the firm was its own control variable, as event studies compare the effect on price of the same firm before and after (Margolis & Elfenbein, 2007).
SIZE is a relevant factor because larger firms may engage more in CSR than smaller firms (Waddock & Graves, 1997). This is because larger firms receive more attention and thus also pressure to engage in CSR. Larger firms may also have more resources to invest in CSR (Margolis & Elfenbein, 2007). Size can be measured in different ways. The most common ways are total assets, total sales and total number of employees (Andersen & Dejoy, 2011). Since all
18
three measures are highly correlated (Andersen & Dejoy, 2011) total sales will be used as a proxy for firm size.
INDUSTRY is another widely used control variable. This is a relevant control variable because the CSR practices may differ per industry (Margolis & Elfenbein, 2007). Some industries might be more polluting than others. According to Waddock & Graves (1997) the financial performance and R&D intensity also differ between industries. By controlling for industry these differences are accounted for. This thesis uses the first two-digit SIC codes to identify different industries (see appendix for a summary of the different industries used).
RISK is the third common control variable. According to Orlitzky and Benjamin (2001), risk is an important control variable because lower risk firms have more certain cash flows. Low risk firms face therefore less uncertainty regarding future CSR opportunities. Margolis and Elfenbein (2007) also say that riskier firms are less likely to engage in CSR activities. A common proxy of risk is the total debt to total assets ratio (i.e. Andersen & Dejoy, 2011; Waddock & Graves, 1997). This ratio will also be used in this thesis.
INTANGIBLES are also suggested as a control variable. While most studies only used the three main control variables (Margolis & Elfenbein, 2007), some literature suggest the use of intangibles as control variables as well (i.e. Andersen & Dejoy, 2011; Surroca et all, 2010). While Andersen and Dejoy only added R&D as a variable, Surroca et all also used human
resources, reputation and culture. The reason these variables may need to be included is related to stakeholder theory: the closer the relationship with stakeholder, the more these intangibles are developed, which eventually can lead to a competitive advantage (Surroca et all, 2010). However, according to Aras et all (2010) R&D is not significant related to financial performance and thus is not an important control variable.
Adding R&D expense and other intangibles to the regression as control variables may result in a few problems. Results are mixed regarding R&D expense and its importance as a control variable. Obtaining intangibles information is difficult as firms are not required to disclose this information in the financial reports. Using only firms that do report on R&D
expense may result in a bias (Jiao, 2010). According to Waddock & Graves (1997) R&D expense is already controlled for when industry is already a control variable. Because of these reasons it was decided to not include the intangibles in the main regression. Nonetheless another regression
19
with a reduced sample using R&D intensity as a control variable will be conducted to see if the results differ significantly. R&D intensity is measured as the ratio of R&D expense to total sales.
3.4 Empirical models
Using all the variables previously described results in the following models that will be tested: 𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝜀
𝑅𝑂𝐴𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝜀 𝑅𝑂𝐸𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝜀 𝑅𝑂𝑆𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝜀
As can be seen from this model, a time-lag is used to take into account the possibility of reverse causality. Another model will be tested to be able to include R&D intensity as a control variable. Because not all firms in the original sample reported their R&D expense, the sample was
reduced. This resulted in the following models:
𝑇𝑜𝑏𝑖𝑛′𝑠𝑄𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝛽5𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑡+ 𝜀 𝑅𝑂𝐴𝑡+1 = 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝛽5𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑡+ 𝜀 𝑅𝑂𝐸𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝛽5𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑡+ 𝜀 𝑅𝑂𝑆𝑡+1= 𝛽0+ 𝛽1𝐶𝑆𝑅𝑡+ 𝛽2𝑆𝑖𝑧𝑒𝑡+ 𝛽3𝑅𝑖𝑠𝑘𝑡+ 𝛽4𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑡+ 𝛽5𝑅&𝐷 𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦𝑡+ 𝜀 While one solution to the possibility of reverse causality is using a time-lag approach, the other is to use instrumental variables. Because this study already employs a time lag approach it was decided to not use an instrumental variables approach as well.
3.5 Sample and data collection
Data regarding the CSR scores will be gathered through the MSCI database on WRDS. The data range for the CSR scores will be the years 2009 through 2013. It was decided to use a data range of 5 years because the increasing popularity of CSR in recent years. 2009 was chosen as a starting point because this was the first year after the financial crisis.
After converting the CUSIP numbers through the WRDS website financial data was gathered through the Compustat database. For the control variables data for the same year as the
20
CSR scores were used. For the dependent variables data for the next year was used. If financial data other than R&D expense was missing the observations would be deleted from the sample. This resulted in 7566 unique observations for the main set of regressions and 4535 observations for the regressions with R&D intensity as a control variable.
21
4 Results
In this chapter the results of this study will be presented. First is explained what kind of data is used together with the descriptive statistics. This is followed by the different assumptions that need to be tested before a regression can be conducted. Then the results of the different models presented in the methodology will be given. This chapter closes with the limitations of this study.
4.1 Descriptive statistics
The two tables below summarize the data. The data set is not perfectly balanced. R&D intensity only has 4535 observations, all other variables have 7566 observations. The mean overall CSR score of this sample is negative. The lowest score of -11 belongs to Titan International, a manufacturing company. The highest score of 19 belongs to Danaher corporation, also a
manufacturing company. As regard the CSR dimensions, environment scores the weakest and the highest scores are for employee relations. The spread in ROS and size are mainly caused by some firms who suffer large losses. This was not caused by a data entry error.
Variable Mean Std. Dev. Min Max
CSR score -.279 2.825 -11 19 Environment .197 .839 -5 5 Community .107 .470 -2 4 Human Rights -.001 .234 -3 2 Employee Rel. .143 1.076 -4 8 Diversity -.348 1.400 -3 7 Product -.089 .539 -4 2 Corp. Gov. -.289 .679 -4 2
Table 2 Descriptive statistics CSR scores each with 7566 observations.
Variable Mean Std. Dev. Min Max
Tobin’s Q 1.947 1.474 .429 36.245 ROA .027 .146 -3.321 1.617 ROE .042 2.291 -127.061 70.385 ROS -1.557 60.022 -4015.286 1.698 Risk .218 .240 0 3.466 Size 5388.706 20659 -1543 474259 R&D intensity 2.033 57.005 0 2900.630 Table 3 Descriptive statistics with financial data. R&D intensity has 4535 observations, all others 7566 observations.
22
4.2 Assumptions
Before conducting the regressions it is necessary to satisfy some assumptions. CSR scores were gathered for a period of five years. Because most firms were scored for multiple years the sample consists of panel data. To regress panel data a fixed effects or a random effects model is needed. A random effects model is suspected because it is believed that differences across the firms influence the financial performance. Random effects were confirmed with a Hausman test. To approximate a normal distribution some variables were transformed into their natural logarithms. This were size, Tobin’s Q, ROA, ROE, and finally ROS. Multicollinearity was checked by looking at the correlation matrix and at the variance inflation factor (VIF). The Wooldridge test confirmed autocorrelation. By plotting the residuals heteroscedasticity was also assumed and accounted for.
Variable Tobin’s Q ROA ROE ROS CSR score Risk Size R&D intensity
Tobin’s Q 1.000 ROA -.104 1.000 ROE -.019 .191 1.000 ROS -0.055 .148 .002 1.000 CSR score -.008 .021 -.006 -.003 1.000 Risk -.069 -.095 .013 -.158 -.000 1.000 Size -.207 .350 .045 .155 .011 .140 1.000 R&D intensity .082 -.089 .001 -.644 .008 .160 -.169 1.000 Table 4 Correlation matrix
4.3 Regressions
The tables below present the results. A total of eight regressions were conducted to test the relationship between CSR and CFP. The first four regressions used the complete sample of 7566 observations. The second four regressions had a reduced sample of 4535 observations. The sample consists of CSR scores for the years 2009 until 2013. All results are robust to heteroscedasticity and autocorrelation and are clustered by company.
23
Variable Tobin’s Q ROA ROE ROS
CSR score -.002* .007* .004 .001 (.001) (.004) (.004) (003) Size -.037*** .024** .067*** .011 (.008) (.012) (.013) (.011) Risk -.033 -.677*** .506*** .096 (.046) (.147) (.113) (.071) Industry dummy 2 -.183 -.454 -.289 -.539* (.165) (.287) (.305) (.320) Industry dummy 3 -.132 -.608** -.364 .486 (.161) (.254) (.267) (.310) Industry dummy 4 .139 -.007 .041 -.091 (.160) (.247) (.263) (.306) Industry dummy 5 -.137 -.005 -.067 .324 (.163) (.256) (.273) (.314) Industry dummy 6 -.153 -.346 -.293 -.167 (.240) (.360) (.343) (.379) Industry dummy 7 .145 -.033 -.031 -.553* (.165) (.255) (.269) (.312) Industry dummy 8 .147 -.133 -.043 -.048 (.161) (.249) (.265) (.308) Industry dummy 9 -.073 -.369 -.188 .020 (.163) (.257) (.275) (.312) Industry dummy 10 -.017 -.141 -.115 -.744** (.167) (.258) (.270) (.321) Intercept .689 -2.937 -2.721 -2.376 R squared .086 .071 .038 .101
Table 5 *, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels respectively. Accounted for heteroscedasticity and autocorrelation using random effects within the regression. Tobin’s Q, ROE, ROA, RO S , and size are log transformations. Random effects model. The results are different from prior research. CSR only has a significant impact at the 10 percent level and only for the dependent variables of Tobin’s Q and ROA. There exists no significant relationship with ROE and ROS. Reason for the difference in results might be that previous research did not implement a time lag. Another reason might be that different
measurements of CSR were used, as already discussed before. The final reason might be that not all research used the same control variables. The signs of the coefficients however does follow earlier research. The accounting measures show a positive relationship which is the same as Ruf et all (2001), Tsoutsoura (2004) and Waddock and Graves (1997). The negative relationship for Tobin’s Q is in accordance with Garcia-Castro et all (2010). These results reject the hypothesis developed earlier that engaging in CSR leads to higher financial performance.
Furthermore it can be seen that size is significant for all measures except ROS. Risk is also an important control variable and is significant for both ROA as well as ROE but with different signs. An explanation for this may be the proxy that is used for risk which was the debt to asset ratio. Firms with higher debt do not need as much revenue from stocks resulting in a positive sign for the ROE coefficient as ROE is calculated by dividing net income by
shareholders equity. Most industry dummies are insignificant which is inconsistent with prior research (i.e. Waddock & Graves, 1997; Garcia-Castro et all, 2010).
24
Variable Tobin’s Q ROA ROE ROS
CSR score -.002* .011** .005 .001 (.001) (.005) (.005) (.004) Size -.035*** .073*** .077*** .052*** (.009) (.014) (.018) (.015) Risk -.036 -1.098*** .544*** -.121 (.056) .148 (.172) (.113) R&D intensity .000*** .000 -.000 .000*** (.000) (.000) (.000) (.000) Industry dummy 2 -.993*** -1.194*** -1.355*** -.556*** (.030) (.060) (.075) (.054) Industry dummy 3 -.696*** -.664*** -.831*** .068 (.045) (.100) (.123) (.112) Industry dummy 4 -.418*** -.261*** -.105** -.747*** (.029) (.041) (.050) (.046) Industry dummy 5 -.578*** -.401* -.544** -.437** (.084) (.238) (.249) (.163) Industry dummy 6 -.494** -1.285*** -1.018*** -.765** (.224) (.067) (.072) (.274) Industry dummy 7 -.448*** -.341*** -.156** -1.272*** (.047) (.071) (.075) (.070) Industry dummy 8 -.357*** -.380*** -.248*** -.672*** (.046) (.067) (.079) (.071) Industry dummy 9 -.687*** -.411* -.306 -.941*** (.088) (.240) (.269) (.201) Industry dummy 10 -.477*** -.431*** -.262** -1.393*** (.075) (.131) (.107) (.128) Intercept 1.276 -2.938 -2.619 -1.934 R squared .071 .084 .067 .108
Table 6 *, **, *** indicate significance at the 10 percent, 5 percent, and 1 percent levels respectively. Accounted for heteroscedasticity and autocorrelation. Reduced sample because of the added variable R&D intensity. Tobin’s Q, ROA , ROE, ROS, and size are log transformations. Random effects model.
The second set of regressions added R&D intensity as a control variable. The signs for the CSR score stay the same but the significance for both Tobin’s Q as well as ROA has increased. Size is now a significant control variable for all financial performance measures. The significance of risk has stayed the same but the coefficient for ROA is now higher. R&D intensity is
significant for Tobin’s Q and ROS although the coefficient is small in number. The positive sign is expected as well. Tobin’s Q is the market to book value. R&D expenditures are not included in the book value but they do increase market value. ROS is a measure of operating efficiency and R&D is mostly directed at increasing efficiency or profit thus increasing the ROS measure. Finally when R&D intensity is added as a control variable the industry dummies almost all increase in significance. This second set of regressions also mostly rejects the hypothesis that the relationship between CSR and CFP is positive. Only when financial performance is measured with ROA does engaging in CSR has a positive impact. Important to remember is that the
25
difference between the first four regressions and the last four regressions may be due to bias. Firms are not required to publish their R&D expenditures. It may well be the case that firms that do report on R&D expenditures may be different than the firms that do not report their R&D expenditures. Also the reduction in sample size of almost 1/3rd may also have a significant effect on the results.
26
5 Conclusion
This section will summarize this thesis followed by limitations and recommendations.
5.1 Summary
The goal of this thesis was to test the relationship between corporate social responsibility and financial performance. CSR was defined as “a concept whereby companies integrate social and
environmental concerns in their business operations and in their interaction with their
stakeholders on a voluntary basis”. The CSR-CFP relationship was tested before but there has
never been a definite conclusion as results were mixed. The reasons for this were the difficulties with respect to measurement and model specification. This study tried to address these issues and used recent data to test if the increased popularity of CSR has a different results than prior
research.
The hypothesis of this thesis was based on earlier research and stated that there would be a positive relationship between CSR and CFP. CSR was measured by using the MSCI database scores. This database was also used in earlier research and has some advantages over other measures of CSR: the wide range of companies included, the use of different CSR categories, independence, and the criteria and sources used. Both accounting as well as market measures were used to test the hypothesis. The three most important control variables used in previous research were added to the main regression: size, risk, and industry. Because of the lack of available information on R&D expense and other intangibles a second regression was conducted with a reduced sample. Results reject the hypothesis and indicate a non-significant relationship in all but one case. The only significant positive result at the 5 percent level was with R&D intensity as control variable and ROA as independent variable. However, as already addressed before the regressions with R&D intensity as a control variable may be biased.
5.2 Limitations and recommendations
Although the regressions were carefully conducted this study still contains some limitations regarding the research of the relationship between CSR and CFP. The first is that this study did not use a weighting scheme for the different CSR categories. Waddock and Graves (1997) did use such a scheme but to construct one would be beyond the scope of this thesis as Waddock and Graves send out questionnaires to different field professionals. Another shortcoming has to do
27
with the R&D variable. The sample reduced significantly in size and this would almost certainly results in a bias. Also, Surroca et all (2010) stressed the importance of other intangible resources to be included as control variables. Because data about these resources is not easily available it was decided to not include them in this study. The final limitation of this research is the
addressing of the possibility of endogeneity. Although a time lag was used to deal with this issue there was no use of other instruments and the instrumental variable approach. The reason for this was that there were no suitable instruments available.
With regard to future research there are some recommendations. First, the measurement of CSR can be improved by constructing an up to date weighting scheme. Second, the use of intangibles should be further examined and tested, preferably with a full data set. The final recommendation is to address the possibility of endogeneity by using an instrumental variables approach as well as a time lag approach.
28
6 References
Andersen, M. L., & Dejoy, J. S. (2011). Corporate social and financial performance: The role of size, industry, risk, R&D and advertising expenses as control variables. Business and
Society Review, 116(2), 237-256
Aras, G., Aybars, A., & Kutlu, O. (2010). Managing corporate performance: Investigating the relationship between corporate social responsibility and financial performance in emerging markets. International Journal of productivity and Performance
management, 59(3), 229-254
Carpentier, C., & Suret, J. M. (2013). Stock market and deterrence effect: A long-run analysis of major environmental and non-environmental disasters. Available at SSRN 2253272 Carroll, A. B. (2008). A history of corporate social responsibility: Concepts and practices. The
Oxford handbook of corporate social responsibility, 19-46
Deegan, C. (2006). Legitimacy theory. Methodological issues in accounting research: theories
and methods, 161-182
EU, (2008). European Competitiveness report, from
http://aei.pitt.edu/45440/1/Competitiveness_2008.pdf
Fisher-Vanden, K., & Thorburn, K. S. (2011). Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and management, 62(3), 430-
445
Freeman, R. E. (1984). Strategic management: A stakeholder perspective.Boston: Pitman
Garcia-Castro, R., Ariño, M. A., & Canela, M. A. (2010). Does social performance really lead to financial performance? Accounting for endogeneity.Journal of Business Ethics, 92(1),
107-126
Gentry, R. J., & Shen, W. (2010). The relationship between accounting and market measures of firm financial performance: How strong is it?. Journal of Managerial Issues, 514-530 Godfrey, P. C. (2005). The relationship between corporate philanthropy and shareholder wealth: A risk management perspective. Academy of management review, 30(4), 777-798
Guthrie, J., & Parker, L. D. (1989). Corporate social reporting: a rebuttal of legitimacy theory. Accounting and business research, 19(76), 343-352
Idowu, S. O., Fifka, M., Zu, L., & Schmidpeter, R. (2015). Dictionary of corporate social responsibility. Springer.
29
Jensen, M. (2002). Value maximization, stakeholder theory and the corporate objective function. Journal of applied corporate finance, 14(3), 8-21
Jo, H., & Harjoto, M. A. (2011). Corporate governance and firm value: The impact of corporate social responsibility. Journal of business ethics, 103(3), 351-383
Jiao, Y. (2010). Stakeholder welfare and firm value. Journal of Banking & Finance, 34(10),
2549-2561
KPMG, (2013). The KPMG survey of corporate responsibility reporting. From
https://www.kpmg.com/Global/en/IssuesAndInsights/ArticlesPublications/corporate- responsibility/Documents/corporate-responsibility-reporting-survey-2013-exec- summary.pdf
Laan van der, G., van Ees, H., & van Witteloostuijn, A. (2008). Corporate Social and Financial Performance: An Extended Stakeholder Theory, and Empirical Test with Accounting Measures. Journal of Business Ethics, 79, 299-310
Margolis, J.D., & Elfenbein, H.A. (2007). Does it pay to be good? A meta-
analysis and redirection of research on the relationship between corporate social and financial performance. Working paper, Ross school of Business, University of Michigan, 1-79
Nielsen, (2014). Global corporate social responsibility report. From
http://www.nielsen.com/content/dam/nielsenglobal/apac/docs/reports/2014/Nielsen- Global-Corporate-Social-Responsibility-Report-June-2014.pdf
Nollet, J., Filis, G., & Mitrokostas, E. (2016). Corporate social responsibility and financial performance: A non-linear and disaggregated approach. Economic Modelling, 52, 400-
407
Oberndorfer, U., Schmidt, P., Wagner, M., & Ziegler, A., (2013). Does the stock market value the inclusion in a sustainability stock index? An event study analysis for German firms. Journal of Environmental Economics and Management, 66(1), 497-509 O'Donovan, G. (2002). Environmental disclosures in the annual report: Extending the applicability and predictive power of legitimacy theory. Accounting, Auditing &
Accountability Journal, 15(3), 344-371
Orlitzky, M., Schmidt, F.L., & Rynes, S.L., (2003). Corporate social and financial performance: a meta-analysis. Organization studies, 24(3), 403-441.
30
Renneboog, L., Ter Horst, J., & Zhang, C. (2008). Socially responsible investments: Institutional aspects, performance, and investor behavior. Journal of Banking & Finance, 32(9), 1723-
1742
Richard, P. J., Devinney, T. M., Yip, G. S., & Johnson, G. (2009). Measuring organizational performance: Towards methodological best practice. Journal of management
Roberts, R. W. (1992). Determinants of corporate social responsibility disclosure: An application of stakeholder theory. Accounting, Organizations and Society, 17(6), 595-612
Ruf, B. M., Muralidhar, K., Brown, R. M., Janney, J. J., & Paul, K. (2001). An empirical investigation of the relationship between change in corporate social performance and financial performance: A stakeholder theory perspective. Journal of Business
ethics, 32(2), 143-156
Schreck, P. (2011). Reviewing the business case for corporate social responsibility: New evidence and analysis. Journal of Business Ethics, 103(2), 167-188
Soana, M. G. (2011). The relationship between corporate social performance and corporate financial performance in the banking sector. Journal of Business Ethics, 104(1), 133-148 Suchman, M. C. (1995). Managing legitimacy: Strategic and institutional approaches. Academy
of management review, 20(3), 571-610
Surroca, J., Tribó, J. A., & Waddock, S. (2010). Corporate responsibility and financial performance: The role of intangible resources. Strategic Management Journal, 31(5),
463-490
Tsoutsoura, M. (2004). Corporate social responsibility and financial performance. center for
responsible business
Waddock, S. A., & Graves, S. B. (1997). The corporate social performance-financial performance link. Strategic management journal, 18(4), 303-319
Wood, D.J. and Jones, R.A. (1995), Stakeholder mismatching: a theoretical problem in empirical research on corporate social performance, International Journal of Organizational
31
7 Appendix
Appendix 1: All recent studies found testing the CSR-CFP relationship
Author and year Sample Method Results
Carpentier, C., & Suret, J. M. (2013)
170 events of major industrial disasters
Event study Non-significant
Fisher-Vanden, K., & Thorburn, K. S. (2011) Voluntarily corporate environmental initiatives
Stock returns Negative
Garcia-Castro, R., Ariño, M. A., & Canela, M. A. (2010) 658 firms from 1991-2005 Accounted for endogeneity Non-significant
Jo, H., & Harjoto, M. A. (2011) 7750 no CSR firms and 5639 CSR firms Tobin’s Q, accounted for endogeneity Positive
Jiao, Y. (2010) 4027 observations for 822 firms Tobin’s Q Positive Margolis, J.D., & Elfenbein, H.A. (2007) 167 studies from 1972-2007 Meta-analysis Positive
Nollet, J., Filis, G., & Mitrokostas, E. (2016).
S&P500 firms from 2007-2011 Accounting and market measures U-shaped Oberndorfer, U., Schmidt, P., Wagner, M., & Ziegler, A., (2013)
51 events between 1999 and 2002
Event study Negative
Renneboog, L., Ter Horst, J., & Zhang, C. (2008) 16 studies from 1992 until 2006 Overview SRI performance Non-significant Ruf, B. M., Muralidhar, K., 496 firms from 1992 until 1995
32 Brown, R. M., Janney,
J. J., & Paul, K. (2001)
Schreck, P. (2011) 300 firms in 2006 Accounted for endogeneity,
accounting and market measures
Non-significant
Surroca, J., Tribó, J. A., & Waddock, S. (2010) 599 companies from 2002 until 2004 Use intangible resources Non-significant Tsoutsoura, M. (2004) 1996-2000, S&P500 firms
Accounting measures Positive
Waddock, S. A., & Graves, S. B. (1997)
469 companies for the year 1990
33
Appendix 2: overview of the CSR dimensions used by the MSCI database and the sub-scores per category
Category Strengths Concerns
Corporate Governance Limited compensation, ownership, reporting quality, political accountability, public policy, corruption & political instability, financial system instability, other strengths
High compensation, ownership, accounting, reporting quality, political accountability, public policy, governance structures, controversial investments, business ethics, other concerns Community Charitable giving, innovative
giving, support for housing, support for education, non-US charitable giving, volunteer programs, community engagement, other strengths
Investment controversies, community impact, tax disputes, other concerns
Diversity CEO, promotion, board of
directors- gender, work-life benefits, women and minority contracting, employment of the disabled, gay and lesbian policies, employment of underrepresented groups, other strengths
Workforce diversity, non-representation, board of directors- gender, board of directors- minorities, other concerns
Employee relations Union relations, no-layoff policy, cash profit sharing, employee involvement, retirement benefits, employee health & safety, supply chain labor standards, compensation
Union relations, employee health & safety, workforce reductions, retirement benefits, supply chain, child labor, labor-management relations
34 & benefits, employee relations, professional development, human capital management, controversial sourcing, other strengths Environment Environmental opportunities,
waste management, packaging materials & waste, climate change, property, plant, equipment, environmental management systems, water stress, biodiversity & land use, raw material sourcing, other strengths
Hazardous waste, regulatory compliance, ozone depleting chemicals, toxic spills & releases, agriculture chemicals, climate change, impact of products & services, biodiversity & land use, operational waste, supply chain management, water management, other concerns Human Rights Indigenous peoples relations,
labor rights, human rights policies & initiatives
Support for controversial regimes, labor rights,
indigenous peoples relations, operations in Sudan, freedom of expression & censorship, human rights violations, other concerns
Products Quality, R&D & innovation, social opportunities, access to finance, other strengths
Product quality & safety, marketing & advertising, anticompetitive practices, customer relations, other concerns
35 Appendix 3: summary of the industry SIC codes
SIC code Industry sector Percentage of sample
01-09 Agriculture, forestry, fishing 0.36
10-14 Mining 5.24
15-17 Construction 1.57
20-39 Manufacturing 43.85
40-49 Transportation & public utilities 6.75
50-51 Wholesale Trade 2.98
52-59 Retail trade 8.02
60-67 Finance, insurance, real estate 13.91
70-89 Services 16.99