Is the impact of Corporate Social Responsibility
on Corporate Financial Performance
heterogeneous across industries?
Jasmira WiersmaS3177300 M.Sc. Finance
Supervisor Prof. Dr. C.L.M. Hermes
University of Groningen- Faculty of Economics and Business June 2017
Abstract
Using a dataset of companies in 9 different industries this study examines whether the impact of Corporate Social Performance (CSP) on Corporate Financial Performance (CFP) differs across industries. This study analyzes the impact of CSP on CFP across industries by estimating panel fixed effects regressions with interaction terms between CSP and industry dummies. Firstly, the results indicate that CSP does not impact CFP when one does not account for industry effects. Secondly, the findings suggest that the sign and magnitude of the impact of CSP on CFP is different in each industry. However, the mixed results make it difficult to observe a clear pattern, no strong evidence is found regarding in which specific type of industry the impact of CSP on CFP is stronger/weaker or negative/positive/no relationship.
1. Introduction
Corporate Social Responsibility (CSR) is a widely-discussed topic nowadays. Companies are increasingly accused of being the main cause of economic, environmental and social problems (Kramer, 2011). Consequently, the pressure on firms to engage in CSR has increased (Lin, Yang, & Liou, 2009). The increase in CSR engagement indicates an evolvement of CSR into a mainstream business practice: firms are not only expected to serve their customers and
thereby generating profit, they must also act in a socially responsible way. Companies became even more interested in CSR activities when it was argued that there is a positive link between CSR and Corporate Financial Performance (CFP). Conversely, the existing literature shows ambiguous results on the effect of CSR on financial performance, without agreement about its nature or even its very existence (Margolis, Elfenbein, & Walsh, 2009; Orlitzky, Schmidt, & Rynes, 2003).
A common control variable used in these studies is industry and is introduced as a dummy in the model. It is often used as a control variable for CFP and does not control for industry drivers of CSR (Margolis et al., 2009). Hoepner et al. (2010) and Daszynska-Zygadlo, Slonski, & Zawadzki (2016) argue that there is no consensus on the relationship of CSR on CFP because previous studies have failed to account for these industry effects on CSR. Literature shows that CSR activities tend to differ across industries and different CSR activities also impact CFP differently. According to Daszynska-Zygadlo et al. (2016), companies in different industries benefit or experience losses from CSR activities to a different extent. They argue that some industries might be more sensitive to CSR activities than others (e.g. industries whose activities are more harmful to the environment are more sensitive to environmental CSR activities) and some industries might be immune to CSR activities. For example, CSR activities (e.g. reduction of emissions or waste) that might be highly important for stakeholders in the manufacturing sector are not as important for stakeholders in the services sector (focuses more on CSR
activities related to employees and consumers). Shell in the oil industry might focus on reducing the possibilities of an oil spill while ING bank may focus on CSR policies regarding better
the CSR impact on CFP across industries, prior research conducted using multiple industry datasets should have to be interpreted with caution. Furthermore, evidence would imply that future research should also account for the differing impact of CSR on CFP.
The aim of this study is to investigate if the impact of CSR performance on CFP is heterogeneous across industries. Therefore, the research question that this study tries to answer is as follows:
Does the impact of CSP on CFP differ across industries?
This is done by using data of 817 firms in 9 different industries during the years 2005-2015. Standard Industrial Code (SIC) obtained through Orbis is used to assign the companies to their corresponding industry. A unique dataset consisting of only publicly listed Western European countries is used. CSR is measured using ESG data of these firms provided by the ASSET4 database by Thomson Reuters. This study employs three different measures of CFP. Namely, Return on Assets (ROA), Return on Equity (ROE) and Market-to-book value (MBTV). The first two are accounting-based measures while the latter is a market-based measure. Moreover, firm size and risk are controlled for. Fixed effects panel regressions with interaction terms between industry and CSR are estimated to answer the research question. In order to test the robustness of the results, two additional tests are conducted.
Even though a vast number of researchers have called for research in this field, up to date there has been only a few studies that have investigated if the impact of CSR on CFP differs per industry. This study is unique from the prior studies because of the methodology it
employs. This study uses interaction effects between CSR and industry dummies whereas prior studies investigate the impact of CSR on CFP by estimating a regression in each industry
based) and long-term financial performance (market-based) across industries. An explanation for this is provided in the latter chapters.
This study makes the following contributions to existing literature. First, this research contributes to the existing literature by further clarifying the link between CSR and financial performance. Because of its wide industry coverage, the longitudinal nature of the sample and the different proxies for financial performance this research provides a picture of the effect of CSR on financial performance. It contributes to the lack of research on how industry effects might play a role on the CSR-CFP relationship. Second, it offers insight on how different CSR activities in different industries might impact a company’s CFP. This has important implications for policy makers. Third, this study argues that the impact of CSR on CFP is still undetermined because previous studies have failed to account for industry effects. The results of this study offer insight into the CSR-CFP relationship when one does not take industry effects into account and when industry effects are accounted for.
The results of this study indicate that CSR does not impact CFP in general when one does not account for industry effects. The results also show that when different measures of CFP are used, very mixed results are obtained making it difficult to find patterns in the result. Consequently, theories regarding why one might expect the impact of CSR on CFP in a certain industry type to be different from the other could not be supported. However, the results of panel regressions and robustness test show that the sign and magnitude of the impact of CSR on CFP differ per industry. Supporting the theory that the impact of CSP on CFP depends on the industry in which a company operates.
The remainder of this paper is organized as follows: Chapter 2 provides a review of existing literature on CSR/ CFP and the CSR- CFP relation. Chapter 3 will discuss the
2. Literature Review
2.1 The concept of Corporate Social Responsibility and why firms engage in it
CSR is a widely-discussed topic in businesses and research nowadays. The rapid evolution of CSR can be attributed to company’s stakeholders becoming more aware of the impact that companies can have on the environment. The concept of CSR has existed for some decades already and there is a substantial amount of research on the definition of CSR.
Nonetheless, there is still no agreed upon definition of this concept. Many concepts have been used over the years to describe CSR such as business ethics, corporate citizenship, sustainability and stakeholder theory (Carrol & Shabana, 2010). Dahlsrud (2008) analyzed 37 definitions of CSR and identified through content analysis five dimensions of CSR that were common across all definitions. These were the stakeholder, the societal, the economic, the voluntariness and the environmental dimension. Dahlsrud (2008) concluded that the lack of an accepted
definition of CSR is due to the fact that most of the definitions are congruent. Dam, Koetter and Scholtens (2009) state that ‘’there are many definitions of CSR, but for most scholars CSR occurs when firms engage in activities that appear to advance a social, environmental or ethical agenda beyond that which is required by law (Siegel and Vitaliano, 2007; Lyon and Maxwell, 2008; Heal, 2008).’’ In the study of Dahlsrud (2008) the definition of CSR by The Commissions of the European Communities was the most used definition of CSR in the academic literature. They define CSR as “Corporate social responsibility is a concept whereby companies integrate social and environmental concerns in their business operations and in their interaction with their stakeholder on a voluntarily basis.” (Commission of the European Communities, 2001 p. 6). This appeared to be most widely accepted definition of CSR and is used in this study as the
definition of CSR.
Engaging in CSR is voluntary and is a cost to the companies that engage in it. Nevertheless, there are several reasons why firms are motivated to engage in costly and voluntary CSR activities (Sprinkle & Maines, 2010; Sharma, Sharma, & Devi, 2009; Pedersen, 2009). First, companies may have altruistic intentions. In other words, they believe that
(NGOs). This is called “windowdressing”, CSR engagement is then viewed as being a
requirement to avoid negative publicity or other actions from such stakeholders. Third, Sprinkle and Maines argue that engaging in employee-centered CSR may help firms recruit, motivate and retain employees. According to Bauwman and Skitka (2012), “CSR Policies and procedures that show concern for employees and promise a good working environment prompt attributions of corporate morality, support general fairness judgments, and foster trust in the company, which in turn should increase organizational attractiveness and commitment and decrease counterproductive workplace behavior.” Fourth, customers may prefer to buy products or services from companies that actively engage in CSR. Therefore, companies can charge price premiums or increase its market share. Fifth, focusing on environmental concerns (for example, using less packaging material) can lead to reductions in production costs. Furthermore, CSR engagement can help companies alleviate risks (for example, litigation risks). Reduced risks often go hand in hand with reduced costs. Finally, firms with better CSR performance (higher CSR scores) may face lower capital constraints. Cheng et al. (2012) propose two possible explanation for this. First, superior CSR performance is associated with superior stakeholder engagement which in turn leads to lower agency costs. Second, they argue that better CSR performance reduces asymmetric information between investors and the company due to higher levels of transparency. They argue that firms with high CSR scores tend to have higher levels of CSR disclosure. Consequently, these firms are more transparent and accountable. Sprinkle and Maines (2010) propose another argument for the above, stating that investors prefer to invest in projects that are aligned with their moral aims.
2.2 Previous research on the CSP- CFP relationship
Corporate Social Performance is defined here as a measure of the outcomes of CSR and will be discussed in detail later on. There have been many studies on the relationship between CSP and CFP. However, the results from these studies are mixed. The main focus of most
(1997), van Beurden & Gössling (2008), Martinez-Ferrero & Valeriano (2015). According to a meta-analysis conducted by Margolis et al. (2009) on several papers examing the CSP and CFP relationship only 2% of these supported a negative relationship. Some papers that have found a negative relationship between CSP and CFP are McGuire et al. (1988), Brammer, Brooks and Pavelin (2006) and Tsoutsoura (2004). All of these paper have a similar explanation for the negative relationship between CSP and CFP. They argue that when companies engage in CSR activities they face a competitive disadvantage because by engaging in CSR activities these firms incur costs that directly impact their profitability. On the other hand, 58% concluded that there is no significant relationship. Some papers that have concluded that there exists no relationship between CSP and CFP are Aupperle, Carroll and Hatfield (1985), Brine, Brown and Hackett (2007) and Aras, Aybars and Kutlu (2010).
2.3 The impact of CSR on CFP across industries
Academics argue that the effect of CSP on CFP can be influenced by the industry the company operates i.e. the impact of CSP on CFP differs per industry. Some arguments for this are presented below.
Hoepner et al. (2010) argue that the impact of CSP on CFP might differ per industry because industries differ in their proximity to end consumers. Baron et al. (2009), Lev, Petrovits, and Radhakrishan (2009) and Hoepner et al. (2010) find that the impact of CSP on CFP is higher in industries that serve end consumers. They argue that this is because end consumers tend to show more social concerns in their consumption. Baron et al (2009) found that the impact of CSP on CFP is positive in industries that serve end consumers and negative in industries that serve businesses. Lev, Petrovits, and Radhakrishan (2009) found a positive relationship between charitable donations and sales growth in companies that serve end consumers. They argue that an increase in the firm’s charitable donations is more easily observed in consumer-focused industries. This, in turn, leads to a better reputation/image and more demand by the
consumers. Based on the above, we can expect the impact of CSP on CFP to be more prominent in industries that are closer to operate closer to their end consumers.
society or the natural environment. Similarly, Chand (2006) and Baird et al. (2012) argue that firms in different industries deal with distinct social and environmental concerns. According to Padgett and Galan (2010), Peters and Romi (2014), Derwall et al. (2005) and Semenova and Hassel (2008), companies that operate in environmentally problematic industries (e.g. manufacturing) tend to face greater political and social pressures by their stakeholders to be environmentally responsible. This is because the consequences of the company’s activities impact the whole society. Environmentally problematic industries are defined as industries whose activities are more harmful to the environment. Companies in these industries tend to have higher levels of waste, CO2 emissions, and pollution. For example, companies in the agriculture industry often produce hazardous waste such as nitrogen, pathogens and more. These are detrimental to plant, aquatic, animal, and human life. Because of their potential harm to the society, these companies will tend to invest more in CSP practices, in particular, the environmental dimension to satisfy their stakeholders. However, academics argue that the CSR efforts of companies in such industries will have no impact on the CFP of these companies (Hoepner et al., 2010; Flammer, 2015). Flammer (2015) argue that in industries with lower institutional norms of CSR (“dirty”/ environmentally problematic industries) stakeholders are indifferent towards the CSR efforts of these companies. This is because these stakeholders view these efforts as mandatory and will not attach any additional value to these activities.
An example for the above can be seen in the oil industry, a company such as Shell may encounter higher risks because of their potential environmental damage. Shell may then invest more in the environmental dimension of CSR in order to reduce these risks such as to reduce waste, emissions and the possibility of oil spills. These efforts can be seen as being a necessity and produce no additional benefits for Shell. Contrastingly, when a company in the services industry like ING bank chooses to be more environmentally friendly by reducing paper waste and choosing to send letters via emails instead of via post these efforts will not go unnoticed by their stakeholders. Konar and Cohen (2001) find support for the above stating that financial markets react differently to industries that are environmentally clean and polluting. A real life example of the above can be seen after the British Petroleum’s (BP) oil spill in the Gulf of Mexico in 2010. Before this event, BP’s stock price was 59.50 US dollars and three months after the oil spill BP’s stock was trading at 28.90 US dollars per share. After this oil spill, BP increased its CSR activities but its financial performance remained more or less the same and their CSR activities had no impact on their CFP (Flammer, 2015). This example shows that when CSR environmental practices are considered a necessity stakeholders will be indifferent to these CSR efforts. Thus, based on the above one can expect CSP to have no influence on CFP in
environmental problematic industries such as manufacturing, mining, transportation and agriculture/ forestry.
Contradictorily, Marsat and Williams (2011), Slonski et al. (2014) and Daszynska-Zygadlo et al. (2016) find that CSP negatively impacted CFP in environmentally problematic industries. Derwall et al. (2005) and Semenova and Hassel (2008) provided an explanation for these results stating that the positive effects of CSR are difficult to achieve in environmentally problematic industries due to the higher cost of environmental performance. In polluting industries the costs of new technologies, clean processes and reforming costs to reduce emissions are high. For example, a company in the manufacturing industry might need to change its entire
On the social side, however, academics find that CSR efforts positively impacted CFP in socially problematic industries and these positive effects were more pronounced in these industries (Herremanns et al., 1993; Lee and Faff, 2009). Companies in retail or wholesale industries tend to face greater social issues regarding their employees. These industries are often scrutinized because they have bad working conditions for their employees. It is expected that when these companies improve on their social dimension this will have a stronger impact on their CFP. Herremans et al. (1993) find that when companies that operate in a socially problematic industry adopt a CSR agenda they are able to reduce their business risks
significantly. These business risks might include lawsuits, strikes, reputation and brand erosions and boycotts. All of these risks can influence a firm’s profitability because a reduction in risk often goes hand in hand with a reduction in costs. For example, companies in the textile industry often face scandals about child labor and sweatshops in less developed countries. When these companies incorporate CSR programs in their business strategies they are able to reduce risks, improve the company’s reputation, attract more investors and perform financially better (Palmer, 2012). A real-life example for the above is Nike’s sweatshop-scandal of 1990’s. In 1996, Nike was accused of providing bad working conditions and violating human rights by using sweatshops to reduce operating costs. When this happened Nike’s stock prices and sales fell dramatically. Nike developed a CSR program which incorporated opinions of shareholders and labor issues. After this Nike was able to improve its reputation, reduce its business risk and better its financial performance (Palmer, 2012). In the beginning of the 2000’s Nike was able to increase its stock price and doubled its profit. Vogel (2007) argues that Nike’s improved
financial performance stems from its CSR efforts. Based on the discussion above, we can expect CSR to positively impact CFP in socially problematic industries and the impact to be higher as compared to industries that face less social concerns.
likely to be a differentiating factor for more consumers since price differences among
comparable products are likely to be low, and since products themselves are likely to be similar. By investing in CSR activities firms in these industries are able to gain a competitive advantage, increase sales and attract more investors. Fisman et al., (2005) state that consumers in highly competitive markets often prefer to buy products from companies that have an active CSR agenda.
2.4 Hypotheses
Based on the discussion in section 2.2 it can be concluded that some studies find a positive, negative or no relationship. However, most of these use industry as a control variable related to CFP. As discussed above industry also influences a company’s CSR activities. Because of this, the impact of CSP on CFP might differ across industries. As stated previously, Daszynska-Zygadlo et al. (2016) argue that each industry experience benefits or losses from CSR actions to a different extent. The discussion above suggests that when analyzing the CSP-CFP relationship one must also account for industry effects and using control variables related to CFP alone is not sufficient. It can be argued that when one does not take these effects into account there will be no relationship between CSP and CFP. Therefore, it is concluded that the relationship between CSP and CFP is still undetermined because previous studies did not account for these industry effects. The following hypothesis is developed:
H1: The impact of CSP on CFP is undetermined
The existing literature suggests that the impact of CSP on CFP differs for companies in different industries. Previous studies on CSP- CFP relationship have mainly focused on one industry type and few studies explore the moderating effects of a specific property of industries. There is still a lack of research on the difference of the CSR impact on CFP across industries (Daszynska-Zygadlo et al., 2016). Therefore, the aim of this study is to investigate if the impact of CSP on CFP differs across industries.
larger in firms that operate closer to their end consumers. Industries that operate close to their end consumers are the retail, services, construction and public administration industry.
Furthermore, it was discussed that CSP may not impact or negatively impact the CFP of companies in environmentally problematic industries. This is because in these industries CSP is considered a necessity or the cost of CSR activities are too high. In this study, the CSP impact on CFP is evaluated in four environmentally problematic industries. These are the manufacturing (contributes to air and water pollution), mining (contaminates soil and groundwater,
contributes to the loss of biodiversity), agriculture (produce hazardous waste that is detrimental) and the transportation (contributes to air pollution) industry.
Contrarily, as discussed in section 2.3, CSP positively impacts CFP in socially problematic industries. Two socially problematic industries are considered in this study; the wholesale and retail industries. Companies in this industry have to avoid labor issues, poor working conditions and the violation of human rights. Therefore, as discussed to reduce the business risks of scandals these companies invest in CSP. The reduction of these risks directly impact the
company’s financial performance by reducing costs associated with lawsuits, brand erosion etc. Finally, we can expect the impact of CSP on CFP to be higher in competitive industries. This is because in these industries CSR is often used as a differentiation strategy and managers more often engage in profit-maximizing CSR activities. In this study, the construction and retail industry are considered to be highly competitive industries. The barriers to entry in these industries are considered to be low and many firms are already operating in these industries.
Based on the discussion in section 2.3 and the above the following hypothesis can be developed:
3. Data & Methodology
3.1 Data sourcesThe Standard Industrial Code (SIC) is used to distinguish the industries. SIC is a system used to classify industries by four digits. The SIC codes first start by dividing the companies into broad industries (two digits), then into sub-industries (three digits) and finally into specific specialization (four digits). In this study, only the first two digits of the SIC codes are used. The SIC defines 10 different sectors: Agriculture, Forestry, Fishing (SIC Sector Code: 01-09), Mining (10-14), Constructions (15-17), Manufacturing (20-39), Transportation and Public Utilities (40-49), Wholesale Trade (50-51), Retail Trade (52-59), Finance, Insurance and Real Estate (60-67), Services (70-89) and Public Administration (91-99). This data is obtained through Orbis. Only companies publicly listed in Western Europe are used for this study. First, most of the previous research has focused on finding the CSP-CFP link in U.S. and UK Firms (Preston & O' Bannon, 1997; Scholtens, 2008). To the best of my knowledge, few studies have focused on studying this relationship in Western European countries. Hoepner et al. (2010) studied the heterogeneity of the CSP-CFP relationship across industries in 19 countries and Daszynska-Zygadlo et al. (2016) used data from North America, Europe, and Asia. Since there is a lack of studies concentrating on the Western European region this study chose to focus solely on this region. This study investigates the impact of CSP on CFP by using aggregate Asset4 ESG data.
cleaning up the data, a total of 817 companies is left. The companies are grouped in industries and the number of observations per industry can be seen below. The countries to which these companies belonged to and the number of observations per country can also be seen in Table 1. The other countries were eliminated from the sample due to not having any observations or having less than 30 observations. Thus, after the clean up the listed firms that were left were from one of these seven countries. The finance, insurance and real estate sector (SIC code: 60-67) is excluded in the analysis since the companies in these industries are highly regulated (for example they have certain capital requirements) in comparison to firms in other industries and this might influence these firm’s CFP. Because of these regulations, the CSP-CFP relationship in these industries may need to be interpreted differently which is beyond the scope of this study. Finally, the final dataset consists of 8987 observations on 817 firms for the period of 2005 to 2015.
Table 1. Number of observations per industry
Industry Spain (ES) United Kingdom (GB) France (FR) Germany (DE) Sweden (SE) Turkey (TR) Italy (IT) N
Agriculture, Forestry, Fishing (AGR) 15 12 11 15 7 2 10 77
Mining (MIN) 20 14 11 7 5 4 2 63
Construction (CON) 7 5 7 5 7 3 5 39
Manufacturing (MANU) 64 53 37 23 18 5 7 207
Transportation and Public Utilities (TRANS)
27 20 25 18 8 3 7 108
Wholesale Trade (WHOT) 24 30 32 27 2 6 5 121
Retail Trade (RET) 12 11 11 10 6 3 1 54
Services (SER) 34 23 24 10 5 7 12 115
Public Administration (PAD) 7 8 9 8 1 0 0 33
3.2 CSP measurement (independent variable)
CSR as a term for an entire field of research and practice is too complex to be narrowed down to one definition and a measurable concept (Gössling, 2011). Academics argue that the reason why there is still no consensus on the relationship between CSR and CFP is that there is still no agreed upon measurement of CSR. A reason might be that “CSR is a difficult concept to measure because it is not a variable and therefore cannot directly be measured” (van Beurden & Gössling , 2008). CSP, however, can be seen as a way of making CSR applicable and putting it into a measurable variable (van Beurden & Gössling , 2008). According to van Beurden and Gössling (2008), the definition of CSP according to Wood (1991) makes CSP suitable for measurement of a company’s CSR. Wood (1991) defines CSP as “a business organization’s configuration of principles of social responsibility, processes of social responsiveness, and policies, programs, and observable outcomes as they relate to the firm’s societal relationships”. According to Tuppura et al. (2016), CSP is a measure of a company’s engagement in CSR
activities. Therefore, in this study, CSP is used as a measure of a company’s engagement in CSR.
measure a firm’s CSP in this study. Thomson Reuters defines the social pillar as “the social pillar measures a company's capacity to generate trust and loyalty with its workforce, customers, and society, through its use of best management practices. It is a reflection of the company's
reputation and the health of its license to operate, which are key factors in determining its ability to generate long-term shareholder value.” The environmental pillar is defined as “a company's impact on living and non-living natural systems, including the air, land, and water, as well as complete ecosystems. It reflects how well a company uses best management practices to avoid environmental risks and capitalize on environmental opportunities in order to generate long-term shareholder value.” The way the firm performs on these two indicators is combined into one measure of CSR performance: CSP. Asset4 assigns a z-score to each pillar that can be used to benchmark a company’s performance against others. This z-score is a relative measure; it reflects a company’s CSP relative to the average CSP of all other companies that are rated by ASSET4. The z-scores are normalized. This normalization entails that ASSET4 scales the z-scores in order to make them fit into the range of zero to hundred. The higher the Z-score the better the company’s performance. A detailed explanation of the methodology used by ASSET4 to calculate its pillar scores is given in Appendix A.
3.3 CFP measurement (dependent variable)
Many different CFP measurements have been used in previous studies to test the relationship between CSP and CFP. According to Cochran and Wood (1984), there is no
based measures are forward-looking and they reflect the long-term financial performance of the company. These measures reflect the idea that the primary stakeholders of a firm are the shareholders. They are also less susceptible to accounting regulations and managerial
manipulation. Market-based measures previously used are stock market returns, the market-to- book Ratio (Tobin’s Q) and price per share.
Although researchers agree that the measurement of CFP is more straightforward than the measurement of CSP, there are still disagreements on which of the CFP two measures is better for predicting the CSP- CFP relationship. This study will use both accounting- and market- based CFP measures. As for the first measure, ROA and ROE are chosen as proxies for CFP. For the second measure Market- to - book value (MTB) / Tobin’s Q is used. The necessary data are obtained through DataStream.
3.3.1 Return on Assets (DataStream code: WC08326)
Return on assets (ROA) is a profitability measure. It has been frequently used in many previous types of research and is generally accepted a measure of a firm’s financial
performance. Therefore, it is used as an accounting- based measure in this study. ROA is calculated as follow:
𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐴𝑠𝑠𝑒𝑡𝑠 = 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
3.3.2 Return on Equity (DataStream code: DWRE)
Another profitability measured is the Return on Equity (ROE). ROE is calculated as follows:
𝑅𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝐸𝑞𝑢𝑖𝑡𝑦 = 𝑁𝑒𝑡 𝐼𝑛𝑐𝑜𝑚𝑒
𝑠ℎ𝑎𝑟𝑒ℎ𝑜𝑙𝑑𝑒𝑟𝑠′𝑒𝑞𝑢𝑖𝑡𝑦
3.3.3 MTBV / Tobin’s Q (DataStream code: MTBV)
𝑇𝑜𝑏𝑖𝑛′𝑠 𝑄 =𝐸𝑞𝑢𝑖𝑡𝑦 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 + 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 𝑀𝑎𝑟𝑘𝑒𝑡 𝑉𝑎𝑙𝑢𝑒 𝐸𝑞𝑢𝑖𝑡𝑦 𝐵𝑜𝑜𝑘 𝑉𝑎𝑙𝑢𝑒 + 𝐿𝑖𝑎𝑏𝑖𝑙𝑖𝑡𝑖𝑒𝑠 𝐵𝑜𝑜𝑘 𝑣𝑎𝑙𝑢𝑒
Regularly, it is assumed that the market and book value of liabilities are the same. Therefore, the formula for Tobin’s Q is the equity market value divided equity book value equal to the MTB ratio.
3.4 Control Variables
Previous studies suggest that firm size and risk affect a company’s financial performance (Ullmann, 1985; McWilliams et al., 2006; Orlitzky et al., 2003). These are also the two most used control variable when investigating the CSP-CFP relationship. Therefore, these variables are controlled for and the necessary data is obtained through DataStream.
3.4.1 Firm size
According to Margolis et al. (2009), larger firms have more resources than smaller firms. Gooding and Wagner, (1985) and Orlitzky, (2003) argue that CFP is positively related to firms’ size because larger firms have better access to resources, can profit from economies of scale and scope and have greater market power. Because of these larger firms usually generate relatively stronger competitive capabilities than smaller firms (Waddock & Graves 1997). On the other hand, a study by Konar and Cohen (2001) found that shareholder value decreased by about 20% when the firm size increased. They explained this by arguing that larger firms have higher transaction costs which lead to a lower CFP. Thus, previous studies suggest that the effect of firm size on CFP can be positive or negative. Therefore, in this study, no expectations are made regarding the sign of the effect of firm size on CFP. There are several measurements of firm’s size: number of employees, sales, and total assets (Orlitzky et al., 2003). However, the study of Orlitzky et al. (2003) determined that the natural log of total assets is one of the most commonly used measures of firm size. Hence, firms’ size is controlled for and measured using the natural log of total assets (Makni et al., 2009; Madorran & Garcia, 2016).
3.4.2 Risk/ leverage
debt ratio, negatively affects CFP. This is because firms with higher debt ratios have higher debt levels and this may lead to financial distress risk (Waddock & Graves 1997). These firms may face more restrict borrowing terms and higher interest rates, making borrowing costlier. This can, in turn, lead to difficulties raising funds for investments.
3.5 Methodology 3.5.1 Models
This study makes use of panel data analysis. Therefore, OLS fixed effects regressions are used to test if the impact of CSP aggregate on CFP varies across industries. In order support the decision of using fixed effects a Hausman test is conducted. The results show that the fixed effects should be used. The results of the test can be obtained in Appendix C. Both cross-sectional/ firm- fixed and period fixed effects are used.
CSP will be the independent variable and CFP will be the dependent variable. The control variables mentioned earlier are also included in the model as independent variables. The following regressions will be estimated to test hypothesis 2.
𝑅𝑂𝐴𝑖,𝑡 = 𝑎𝑖 + 𝜇𝐶𝑆𝑃𝑖,𝑡−1+ 𝜒𝐶𝑆𝑃𝑖,𝑡−1∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖+ 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (1) 𝑅𝑂𝐸𝑖,𝑡 = 𝑎𝑖+ 𝜇𝐶𝑆𝑃𝑖,𝑡−1+ 𝜒𝐶𝑆𝑃𝑖,𝑡−1∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (2) 𝑀𝑇𝐵𝑉𝑖,𝑡 = 𝑎𝑖 + 𝜇𝐶𝑆𝑃𝑖,𝑡−1+ 𝜒𝐶𝑆𝑃𝑖,𝑡−1∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖+ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (3)
year time lag is taken to reflect that changes in CSP does not immediately affect CFP. Marom (2006), argue that stakeholder’s reaction to CSR activities unfold over the long run. Risk and size are also lagged by one year.
This analysis is further extended by disaggregating the scores into the two pillars. These models can be seen below.
𝐶𝐹𝑃𝑖,𝑡 = 𝑎𝑖+ 𝜇𝐸𝑁𝑉𝑠𝑐𝑜𝑟𝑒𝑖,𝑡−1+ 𝜒𝐸𝑁𝑉𝑖,𝑡−1∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖,𝑡+ 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (4)
𝐶𝐹𝑃𝑖,𝑡 = 𝑎𝑖 + 𝜇𝑆𝑂𝐶𝑠𝑐𝑜𝑟𝑒𝑖,𝑡−1+ 𝜒𝑆𝑂𝐶𝑖,𝑡−1∗ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖+ 𝐼𝑁𝐷𝑈𝑆𝑇𝑅𝑌𝑖 + 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (5)
4. Results and Discussion
4.1 WinsorizingBefore analyzing the descriptive statistics, the normality of the distribution of ROA, ROE, MTBV, Size, and Risk are checked. The CSP, ENV and SOC variables are not controlled for since by construction these variables are forced to be in a range from 0 to 100. The analysis shows that there are outliers in the CFP and control variables. These outliers can have an effect on the model and the values of the coefficients estimated by the regression equations. A common technique to minimize the effect of outliers in research is called winsorizing. Winsorizing is a process that involves bringing down outliers to a specified value. Extreme values are replaced with a certain percentile from each end. The variables above are winsorized at 2.5% to 97.5% level. This is a common level chosen in research. A 1% to 99% winsorization was not used because the variables still had major outliers after winsorization at this level. A higher level of winsorization (5% to 95% and above) was not chosen because this would result in a significant transformation of valuable data. However, after winsorization, some outliers were still
observed in the ROA and ROE variables. Therefore, a 5% to 95% winsorization level was chosen for these two variables. After the winsorization these variables appear to be almost normally distributed. The winsorized variables are further used in the study.
4.2 Descriptive Statistics
The table below shows the descriptive statistics of all the variables used in this study, including the total number of observations, mean, standard deviation, minimum value, and maximum value. The tables contain winsorized data for the ROA, ROE, MTBV variables.
Table 2. Descriptive Statistics whole data set
From the table, it becomes evident that this is an unbalanced panel. This is because based on the number of firms (817) and a time period of 11 years the number of observations should be 8987 and all of the variables have fewer observations than this. The mean of the measures of CFP all exhibit positive values. The average ROA, ROE, and MTBV for the firms in the dataset are 7.144%, 15.45% and 2.856 respectively. However, looking at their minimums ROA and ROE exhibit negative values. A negative ROA indicates that the utilization of capital could be more effectively managed. A company that has reported a negative income might also report a negative ROA or ROE. All of the CSP ratings are below 100 as expected. This is because by construction these variables are forced to be in a range from 0 to 100. The average CSP rating was 58.88. The highest CSP score is achieved by Repsol YPF in the mining industry and the lowest by BB Biotech in the services industry. Size, as measured by total assets in US$, shows large values. Therefore, as already mentioned, the logarithm of total assets is used. By construction, the minimum value of the level of risk is zero. The level of risk cannot take a negative value because it is calculated by dividing total debt by total assets, which both cannot take a value below zero. When the firm holds no debt, the level of risk is equal to zero.
Table 3. Averages of variables in each industry
From the table above, it can be concluded that the services industry has the highest ROA and MTBV. The retail industry has the highest ROE. Firms in the manufacturing industry have the highest debt ratio. This suggests that that firms in this industry have high levels of debt and might be riskier than firms in other industries. Finally, looking at the average size,
companies in the transportation industry are the largest compared to firms in the other industries.
The construction industry has the highest CSP score, an explanation for this as discussed previously is that this is a highly competitive industry. As discussed, managers from companies in this industry often invest in CSP in order to differentiate themselves from other companies. Therefore, one might expect firms in this industries to have higher CSP ratings.
As discussed in section 2.3 environmentally problematic industries are defined as industries whose activities are more harmful to the environment. These are the mining,
manufacturing, transportation and agriculture, fishing & forestry industries. The activities of the firms in this industry tend to have a higher impact on the environment. For example, firms in the transportation industry tend to have higher levels of CO2 emissions compared to firms in the services industry. As discussed in the literature review these firms are expected to have higher ratings on the environmental dimension of CSP as well because of the pressures they face by their stakeholders to be environmentally responsible. The findings do show that the transportation industry an environmentally problematic industry has the second highest CSP score and the highest environmental score.
The table shows that the retail industry has the highest score on the social dimension. Firms in the retail industry are more competitive and are more socially problematic. As discussed in the literature review, firms in this industry might invest more in the social dimension of CSP as a differentiation strategy and to reduce business risks associated with being socially irresponsible. The services industry has the lowest score on CSP and both of the dimensions. Companies in these industries have less social and environmental concerns to deal with. Because of this, they might be less involved in CSP.
the literature the impact of CSP on CFP might differ in different industries. In order to test if the CSP, environmental and social scores statistically differ across industries an Analysis of Variance (ANOVA) test is conducted for each of these variables. ANOVA is used to test if the group means for the different industries statistically differ from each other. The null hypothesis states that the industry means on the 3 variables do no statistically differ. If this hypothesis is rejected this is an indication that CSR activities do vary across industries. The results (F-stat: 40.44, p-value: 0.00) show that the CSP means of the various industries do statistically differ.
Furthermore, the results indicate the means on the environmental (F-stat: 43.09, p-value: 0.00) and social dimension (F-stat: 26.76, p-value: 0.00) do differ across industries. Therefore, it can be concluded that industries differ when it comes to their CSP and performance on two dimensions of CSP.
4.3 Correlation Analysis CSP, CFP, and control variables
Table 4. Correlation Matrix
Notes: *P < 0.01, **P < 0.05, ***P < 0.10. For size the natural log is taken and CSP, RISK and Size is lagged by one year.
Table 4 presents the correlation coefficients between the CFP measures, CSP measures, the CSP dimensions and the control variables. The null hypothesis here states that the variables do not correlate. The correlation coefficients presented above are based on Pearson correlation test. The Pearson correlation test is the most appropriate test for correlation between
parametric variables. A Pearson correlation coefficient of -1 indicates a perfect negative correlation, +1 indicates a perfect positive correlation and 0 indicates no relationship.
Variables ROA ROE MTBV CSP ENV SOC SIZE RISK
Most of the coefficient exhibit a weak to moderate correlation and are statistically significant at a 1%- level based on a two-tailed test. As mentioned previously, CSP is lagged by one year. The correlation coefficients between CSP and the CFP variables suggest that higher levels of CSP are associated with lower levels of ROA and MTBV and higher levels of ROE in the next year. These coefficients might indicate that CSP negatively affects ROA and MTBV but positively affects ROE. However, correlation does not indicate causation. Therefore, panel regressions are run and discussed in section 4.5 to determine the impact of CSP on the different CFP measures.
4.4 Correlation Analysis CSP, CFP and control variables per industry
Correlations matrixes between the variables in the industry subsamples are presented in Appendix C. The tables indicate that CSP is negatively related to ROE and MTBV in the
transportation industry an environmentally problematic industry. However, CSP is positively correlated to ROA and ROE in the manufacturing industry which is also an environmentally problematic industry. One would expect the first finding based on the literature review however the latter finding contradicts the theory. Based on the literature, one would expect CSP to have no correlation to CFP or higher levels of CSP to lead to lower CFP. This is because CSR is considered a necessity thus stakeholders attach no value to these CSR activities or that the costs of being environmentally responsible outweighs the benefits in industries with higher environmental impact (Flammer, 2015; Derwall et al., 2005; Semenova and Hassel, 2008).
The environmental score is statistically and negatively correlated to MTBV in 3 environmentally problematic industries (agriculture, mining, and transportation). Based on what was discussed above this is what one would expect based on the literature review. Thus, one would expect higher levels of CSP to lead to lower/ worse CFP in environmentally
problematic industries.
social score in negatively correlated to all the CFP measures in the wholesale industry a socially problematic industry.
The correlation matrixes reveal that the correlation between CSP and CFP depends on the CFP measure that is being analyzed. For example, we see that for the agriculture and the transportation industry CSP is sometimes positively and other times negatively correlated to CFP depending on the CFP measure.
All in all, the results are very mixed. Sometimes CSP is positively related to CFP, sometimes negatively related and other times there is no correlation depending on the industry. This indicates that CSP correlates differently to CFP in all the industries. Correlation, however, does not address issues of causality. Therefore, panel regressions are estimated in order to test the hypotheses that the CSP impact on CFP differs across industries. The preliminary results, however, indicated that these regressions might produce no or contradictory results.
4.5 Panel data regression hypothesis 1
To test the first hypothesis the impact of CSP on CFP must be analyzed. This is done by estimating a regression with only the CSP measure and control variables.
Table 5 Cross-sectional and time fixed effects model without industry dummy variables
Notes: Standard errors are reported in parentheses. Reported F-statistic corresponds to a Wald test of the hypothesis that all coefficients excluding the constant are equal to zero*P < 0.01, **P < 0.05, ***P < 0.10. For size the natural log is taken and CSP is lagged by one year.
Hypothesis 1 states, as discussed in the literature review, the impact of CSP on CFP is undetermined because none of the previous studies accounted for industry effects on CSP. It
Independent Variable ROA ROE MTBV
can be argued that when one does not take these effects into account there will be no relationship between CSP and CFP. The model without the inclusion of industry interactions shows that the CSP of the previous year has a nonsignificant relationship with the next year’s CFP. The p-values of all these variables were larger than 0.10, suggesting that this relationship is not statistically significant. The ROE and MTBV models suggest that none of the variables play a significant impact on a firm’s CFP. However, in the ROA model size and risk are negative and significant suggesting that firm size and risk negatively affect CFP. Based on the results above we find support for the first hypothesis. Thus, when one does not account for industry effects CSP does not impact CFP.
4.6 Panel data regressions hypothesis 2
In order to test hypotheses 2 that the impact of CSP on CFP varies across industries, interactions between industry dummies and CSP are included in the model. The first model contains the interaction between CSP and the agriculture plus the dummy variables of the industries. The second model contains the interaction between CSP and the mining plus the dummy variables of the industries and so forth. If the coefficient of the interaction term is significant, it implies that the effect of CSR on CFP is sensitive to industry. In other words, the effect of CSR on financial performance is different in industries. When the interaction term is positive the impact of CSP on CFP is enhanced by the value of the interaction term as compared to the reference category which is the public administration industry. Thus, for example, if the impact of CSP on CFP is 0.5 and the interaction term between CSP and agriculture is 0.2. This would suggest that the impact of CSP on CFP in the agriculture industry is (0.5+0.2= 0.7) higher as compared to companies in the public administration industry. A total of 8 models are estimated and the results of these models can be obtained in Appendix D. As mentioned in section 3.5 the decision to use fixed effect was supported by the Hausman test.
Before discussing the results, it is important to note that when using interaction effects problems might arise if the correlation between the CSP, industry dummy variables and the interaction variables is high (Aiken & West, 1991). If the correlations are high a data
correlate. The correlation matrix between CSP, industry dummies and CSP times the industry dummies show that all the variables are correlated. However, the variables are weakly correlated. The highest correlation being between CSP and the interaction between CSP and the transportation industry dummy. Therefore, it is concluded that the correlation between the 3 variables are low and we do not need to transform these data.
The models in Appendix D show that the sign and value impact of the control variables on CFP vary. This is also the case for the impact of CSP on CFP. The results depend on the measure of CFP being used. When ROA is used a significant CSP and CSP interaction coefficient is observed only in the manufacturing industry. However, when ROE is used as a proxy for CFP a positive significant impact of CSP on CFP is observed in all of the industries except the
agriculture industries. Only the interaction between CSP and services dummy was statistically significant and negative when ROE is used. However, when MTBV many of the industries that previously showed a positive CSP impact on ROE now show a negative impact (agriculture, construction, transportation and manufacturing industries). Almost none of the interactions between CSP and the industry dummies were significant. The results are very mixed because they vary over industries as well as over the CFP measure that is being used. Therefore, it is very difficult to find support for the theories that were discussed in the literature review and find a clear picture. For example, based on the theory on would expect the impact of CSP on CFP to be either nonexistent or negative in an environmentally problematic industry. When one looks at the CSP coefficient in the manufacturing industry it is sometimes negative and
sometimes positive. On the other hand, in the agriculture industry which has a high
significant, their signs and magnitude vary. This indicates that the impact of CSP on CFP does differ across industries.
To further support the above conclusion that CSP impacts CFP differently across
Table 6. Cross-sectional and time fixed effects model all interaction effects
Notes: ROA, ROE, and MTBV are proxies for the dependent variable CFP. Standard errors are reported in parentheses. Reported F-statistic corresponds to a Wald test of the hypothesis that all coefficients excluding the constant are equal to zero*P < 0.01, **P < 0.05, ***P < 0.10. For the size, the natural log is taken and CSP, Risk, and size are lagged by one year.
4.7 Robustness test
In order to test the robustness of the results, two different analysis are conducted. First, in this study interaction terms between CSP and industry dummies are used to investigate the impact of CSP on CFP differs across industries. Hoepner et al. (2010) and Daszynska-Zygadlo et al. (2016) both study the heterogeneity of CSP’s impact on CFP across industries by estimating an OLS regression in each industry subsample. Therefore, to test if the results will still hold, the
whole sample is divided into subsamples based on industry and the following regression is estimated per industry:
𝐶𝐹𝑃𝑖,𝑡 = 𝑎𝑖+ 𝜇𝐶𝑆𝑃𝑖,𝑡−1+ 𝜏𝑅𝐼𝑆𝐾𝑖,𝑡−1+ 𝜂𝑆𝐼𝑍𝐸𝑖,𝑡−1+ 𝜀 𝑖,𝑡 (6) where 𝐶𝐹𝑃𝑖,𝑡 is either ROA, ROE or MTBV. The i and t are the firm and year index. 𝐼𝑛𝑑𝑢𝑠𝑡𝑟𝑦𝑖,𝑡 are dummy variables for industry. 𝑅𝐼𝑆𝐾𝑖,𝑡 is the proxy for firm risk as measured by the debt ratio. 𝑆𝐼𝑍𝐸𝑖,𝑡 is the control variable for firm size as measured by the natural log of total assets. The εi,t is the error term for firm i at time t.
The results of these regressions can be seen in Tables 7 and 8. The results show that the impact of CSP on CFP differs when different measures of CFP is used. For example, in the agriculture industry when ROA and ROE are used as a measure of CFP, the impact of CSP is positive. However, when MTBV is used the impact is negative. The impact of CSP on CFP has proven to be insignificant in all the models. Thus, here again, the models produce very mixed results. However, looking at the sign and the value of the coefficients it can be concluded that the impact of CSP on CFP is different in each industry. Thus, the robustness test also supports the second hypothesis that the impact of CSP on CFP varies across industries.
The second robustness test uses disaggregated CSP data to test if the impact of different CSP dimensions on CFP differs across industries. One would expect different CSP dimensions to impact CFP differently depending on the industry in which a firm operates. Also, if the impact of the dimensions on CFP differs across industries this will further support hypothesis 2. As
discussed previously, Flammer (2015) and Semenova and Hassel (2008) argue that the impact of the environmental dimension of CSP on CFP might be stronger in clean “industries” than in “dirty” industries. This is because in “dirty” industries CSP efforts in the environmental
A correlation test is conducted to see if the ENV, industry dummy variables and the interaction variables are high. The correlation matrix between ENV, industry dummies and ENV*industry show that all the variables are correlated. However, the highest correlation (0.3249) is between ENV and the interaction between ENV and manufacturing dummy variable which is a weak correlation. The correlation matrix between SOC, industry dummies and SOC*industry show that all the variables are correlated. However, the correlations are pretty low, with the highest correlation (0.2397) being between SOC and the interaction between SOC and retail dummy variable.
Table 9 presents the results of the 2 estimated models. For the models using environmental score (models 4 to 6) as the dependent variable, very few coefficients are significant. This indicates that the environmental score does not impact CFP and that the industry to which a firm belongs to does not matter. This is not in line with the results of Daszynska-Zygadlo et al. (2016) who find a significant negative effect in 8 of the 10 industries. The estimated coefficients although not significant show that the impact of the ENV score in each industry is different. However, no clear pattern can be established here.
The models using social score (models 7-9) again produce very few significant
coefficients. In model 8, the interaction between the social score and the transportation and manufacturing industry is positive and significant. Both of these industries are environmentally problematic and their actions are more harmful to the society. Based on theory one would not expect the social dimension to impact the CFP of companies in these industries. An explanation for this is that these industries also have to deal a large number of employees and consumers, they will face greater pressures from these stakeholders to be socially responsible. Companies in the manufacturing industry also tend to have bad working conditions. Therefore, when these companies make improvements on these areas their stakeholders might attach value to these improvements and they may reduce business risks resulting in a higher CFP.
Table 7. Cross-sectional and time fixed effects models per industry: ROA and ROE
Dependent Variable: ROA Dependent Variable: ROE
Sector N 𝒂 CSP RISK SIZE R2 F-
Table 8. Cross-sectional and time fixed effects models per industry: MTBV
Dependent Variable: MTBV
Sector N 𝒂 CSP RISK SIZE R2 F- stat.
Table 9. Cross-sectional and time fixed effects model CSP dimensions
Notes: ROA, ROE, and MTBV are proxies for the dependent variable CFP. Standard errors are reported in parentheses. Reported F-statistic corresponds to a Wald test of the hypothesis that all coefficients excluding the constant are equal to zero*P < 0.01, **P < 0.05, ***P < 0.10. For the size the natural log is taken ENV and SOC is lagged by one year.
ROA ROE MTBV ROA ROE MTBV
5. Conclusion
The objective of this study was to examine whether the impact of Corporate Social
Responsibility performance as measured by its Corporate Social Performance on its Corporate Financial Performance differs across industries. The first hypothesis states that the impact of CSP on CFP is still undetermined. This is based on the fact that previous research on this relationship has failed to account for the possibility of a differing impact of CSP on CFP across industries. The second hypothesis states that the impact of CSP on CFP differs across industries. This is supported by the theory that because of industry characteristics such as being in a competitive, consumer-oriented or socially or environmentally problematic industry shape the impact of CSP on CFP differs across industries. According to theory, firms in different industries react differently to CSR activities. These CSR activities all impact CFP differently. Therefore, the impact of CSP on CFP might differ across industries. CSP and CFP data over a period of 11 years (2005-2015) of 817 firms operating in 9 different industries is used to test these hypotheses. The first hypothesis is tested by estimating panel data regressions without any industry
dummies. The second hypothesis is tested by estimating panel data regressions with interaction terms between industry dummies and CSP.
The first conclusion of this study based on the first hypothesis is that CSP does not impact CFP when one does not account for industry effects. This is in line with the findings of Aupperle, Carroll, and Hatfield (1985), Brine, Brown, and Hackett (2007) and Aras, Aybars and Kutlu (2010) who find no relationship between CSP and CFP. In these studies, the industry was used as a control variable for CFP and they failed to account for the possibility of a differing effect of CSP on CFP across industries.
The second conclusion of this research is that the impact of CSP on CFP does differ per industry. However, the models produced very mixed results making it difficult to draw clear conclusions about in which industry the impact of CSP on CFP is higher or lower. Thus, it is challenging to find support for the theories that argue that the impact of CSP on CFP might differ because the company operates in either a consumer-focused, competitive,
results. Despite this based on the sign and magnitude of the CSP and interaction coefficients, it can be concluded that the impact of CSP on CFP differs per industry. This is in line with the findings of Hoepner et al. (2010) and Daszynska-Zygadlo et al. (2016) who find that the impact of CSP on CFP was different in all their sectors. These findings are further supported by two robustness tests. In the first robustness test, the sample is split into subsamples based on industry. Then a regression measuring the impact of CSP on CFP is estimated for each
subsample. The results again are very mixed and prove to depend on the CFP measure being used. Nevertheless, the results confirm the previous findings that the impact of CSP on CFP differs per industry. The second robustness test uses disaggregated CSP to test whether the impact of the CSP dimensions on CFP differs per industry. The results again indicate that the impact does indeed differ. This is in line with the findings of Daszynska-Zygadlo et al. (2016). They find that the impact of three different CSP dimensions (social, environmental and corporate governance) on CFP differs per industry. They argue that this is because some
industries are immune to CSR actions while sectors that have a high impact on the environment are sensitive to environmental actions only.
5.1 Limitations of this research
5.2 Recommendations for future research
References
Aupperle, K. E., Carroll, A. B., & Hatfield, J. D. (1985). An Empirical Examination of the Relationship between Corporate Social Responsibility and Profitability. The Academy of Management Journal,
28(2), 446-463.
Aiken, & West. (1991). Multiple Regression: Testing and interpreting interactions. CA, Newbury Park: SAGE.
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.
Baird, P., Geylani, P., & Roberts, J. (2012). Corporate Social and Financial Performance Re-Examined: Industry Effects in a Linear Mixed Model Analysis. Journal of Business Ethics, 367-388. Barnea, A., & Rubin, A. (2010). Corporate Social Responsibility as a Conflict Between Shareholders.
Journal of Business Ethics, 71-86.
Baron, D., Harjoto, M., & Jo, H. (2009). The Economics and Politics of Corporate Social Performance.
Stanford University Graduate School of Business Research Paper No. 1993; Rock Center for Corporate Governance at Stanford University working paper No. 45. Retrieved from
https://ssrn.com/abstract=1202390
Bauman, C. W., & Skitka, L. J. (2012). Corporate social responsibility as a source of employee satisfaction.
Research in Organizational Behavior, 32, 63-86.
Brammer, S., Brooks, C., & Pavelin, S. (2006). Corporate social performance and stock returns: UK evidence from disaggregate measures. Financial Management, 35(3), 97-116.
Brine, M., Brown, R., & Hackett, G. (2007, May 31). Corporate social responsibility and financial. Retrieved from Australian Government: The Treasury performance in the Australian context.: http://www.treasury.gov.au/~/media/Treasury/Publications%20and%20Media/Publications/20 07/Economic%20Roundup%20Autumn%202007/Downloads/PDF/04_CSR.ashx
Campbell, J. L. (2007). Why would corporations behave in socially responsible ways? An. Academy of
Management Review, 32, 946-967.
Carrol, A., & Shabana, K. (2010). The business case for corporate social responsibility: A review of concepts, research and practice. International Journal of Managemnet Reviews, 12, 85-105. Carroll, A. B. (1979). A three-dimensional conceptual model of corporate social performanc. Academy of
Management Review, 497-505.
Carroll, A. B. (1991). The Pyramid of Corporate Social Responsiblity: Toward the Management of Organizational Stakeholders. Business Horizons, 39-48.
Carroll, A. B. (1999). Corporate social responsibility: Evolution of a definitional construct. . Business and
Carroll, A. B. (2008). A History of Corporate Social Responsibility: Concepts and Practices. In A. Crane, A. McWilliams, D. Matten, J. Moon, & D. Siegel, The Oxford Handbook of Corporate Social
Responsibility. (pp. 19-46). Oxford University Pres.
Cheng, B., Ioannou, I., & Serafeim, G. (2013). Corporate social responsibility and access to finance.
Strategic Management Journal.
Chih, H.-L., Chih, H.-H., & Chen, T.-Y. (2010). How Well Do Social Ratings Actually Measure Corporate Social Responsibility. Journal of Economics and Management Strategy, 125-169.
Clarkson, M. B. (1995). A stakeholder framework for analyzing and evaluating corporate social performance. Academy of Management Review,, 20, 92-117.
Cochran, P. L., & Wood, R. A. (1984). Corporate Social Responsibility and Financial Performance. The
Academy of Management Journal, 27, 42-56.
Cooper, S. (2004). Corporate Social Performance: A Stakeholder Approach. Burlington, VT: Ashgate. Dam, L., Koetter, M., & Scholtens, B. (2009). Why Do Firms Do Good? Evidence From Managerial
Efficiency. SSRN working paper.
Daszynska-Zygadlo, K., Slonski, T., & Zawadzki, B. (2016). The Market Value of CSR Performance Across Sectors. Inzinerine Ekonomika-Engineering Economics, 230-238.
Endrikat, J., Guenter, E., & Hoppe, H. (2014). Making sense of conflicting empirical findings: A meta-analytic review of the relationship between corporate environmental and financial performance.
European Management Journal, 735-751.
European Commission. (2001). EUR-Lex. Retrieved November 18, 2015, from EUR-Lex: http://eur-lex.europa.eu/legal-content/EN/TXT/?uri=URISERV:n26039
Flammer, C. (2015). Does Corporate Social Responsibility Lead to Superior Financial Performance? A
Regression Discontinuity Approach. Retrieved May 20, 2017, from Management Science:
https://doi.org/10.1287/mnsc.2014.2038
Freeman, E. R. (1984). Strategic management: A stakeholder approach. Advances in strategic
management 1.1, 31-60.
Friedman, M. (1970, September 13). The social responsibility of business is to increase its profits. New
York Times Magazine, pp. 32-33.
Gössling, T. (2011). Corporate Social Responsibility and Business Performance: Theories and Evidence
about Organizational Responsibility. Edward Elgar Publishing.
Graves, S., & Waddock, S. (1994). Institutional owners and corporate social performance. Academy of
Management Journal, 1034-1046.
Hahn, R., & Kühnen, M. (2013). Determinants of sustainability reporting: a review of results, trends, theory, and opportunities in an expanding field of research. Journal of Cleaner Production, 5-21. Herremans, I. M., Akathaporn , P., & McInnes, M. (1993). An Investigation of corporate social
responsibility reputation and economic performance. Accounting, Organizations and Society, 587-604.
Hoepner, A., Yu, P.-S., & Ferguson, J. (2010). Corporate Social Responsibility across Industries: When can who do well by doing good? SSRN Electronic Journal. Retrieved from
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=1284703
Hull, C., & Rothenberg, S. (2008). Firm Performance: the interactions of corporate social performance with innovation and industry differentiation. Strategic Management Journal, 781-789. Kanji, G., & Chopra, P. (2010). Corporate social responsibility in a global economy. Total Quality
Management, 21(2), 119-143.
Kramer, M. R. (2011). Creating shared value. . Harvard business review, 62-77.
Lee, M. (2008). A review of the theories of corporate social responsibility: Its evolutionary path and the road ahead. International Journal of Management Reviews, 10, 53-73.
Lev, B., Petrovits, C., & Radhakrishan, S. (2009). "Is doing good good for you? How corporate charitable contributions enhance revenue growth.". Strategic Management Journal, 31(2), 182-200. Lin, C., Yang, H., & Liou, D. (2009). The impact of corporate social responsibility on financial
performance: Evidence from business in Taiwan. Technology in Society, 56.63.
Madorran, C., & Garcia, T. (2016). Corporate Social Responsibility and financial performance: The spanish case. Revista de Administração de Empresas, 20-28.
Makni, R., Francoeur, C., & Bellavance, F. (2009). Causality Between Corporate Social Performance and Financial Performance: Evidence from Canadian Firms. Journal of Business Ethics, 89, 409–422. Margolis, J. D., Elfenbein, H., & Walsh, J. (2009). Does it Pay to be Good? A Meta-analysis and
Redirection of Research on the Relationship between Corporate Social and Financial Performance. Working paper Harvard University.
Marsat, S., & Williams, B. (2011). CSR and Market Valuation: International Evidence. International
Conference of the French Finance Association (AFFI). SSRN. Retrieved from
http://ssrn.com/abstract=1833581
Martinez-Ferrero, J., & Valeriano, F. A. (2015). Relationship Between Sustainable Development and Financial Performance: International Empirical Research’. Business Strategy and the
Environment, 20-39.
McGuire, J. B., Sundgren, A., & Schneeweis, T. (1988). Corporate social responsibility and firm financial performance. Academy of Management Journal, 31(4), 854-872.
