The influence of corporate social responsibility on
corporate financial performance
Name: Demi Jannink Student number: 11302348
Program: Economics and Business Track: Economics and Finance Supervisor: Patrick Stastra Credit: 12 points
Date 30-06-2020
Statement of Originality
This document is written by Demi Jannink who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it.
The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Abstract
This research aims to examine what the influence of corporate social responsibility is on the corporate financial performance of firms in the energy and pharmaceuticals, biotechnology & life science industry. Panel data is obtained from the databases MSCI ESG and COMPUSTAT and a large sample for the years 2010-2016 is used. Control variables for the effect of size, risk, R&D and industries are added to the model. After performing four regressions, we find ambiguous results. Therefore, this research cannot conclude that corporate social responsibility has a positive effect on corporate financial performance and that there is a difference in effect in different industries.
Table of contents
Abstract ... 3
Table of contents ... 4
1. Introduction ... 5
2. Literature review ... 7
2.1 Corporate Social Responsibility: ... 7
2.2 Measuring CSR ... 7
2.3 Measuring Corporate Financial Performance: ... 8
2.4 The relationship between CSR and CFP: ... 8
2.4.1 Positive relationship ... 8 2.4.2 Negative relationship ... 9 2.4.3 No significant relationship... 9 2.5 Hypotheses... 10 3. Methodology ... 11 3.1 Dependent variable ... 11 3.2 Independent variable ... 11
3.2.1 Industry dummy and interactive dummy variable ... 11
3.3 Control variables ... 12
3.3.1 Size ... 12
3.3.2 Risk ... 12
3.3.3 R&D... 12
3.4 Model ... 13
3.5 Sample and data collection ... 13
4. Results ... 16 4.1 Correlation ... 16 4.2 VIF... 16 4.3 Regression results ... 17 5. Conclusion ... 19 5.1 Limitations... 19 References ... 21 Appendix ... 24
1. Introduction
For a couple of decades now, sustainable development is becoming increasingly important. Occurrences as holes in the ozone layer, global warming, but also the loss of biodiversity, ensure there is a rising belief that the current way of life is no longer tenable. As the above mentioned is only expected to become a bigger problem in the near future, many corporates are willing to do their part (Ambec & Lanoie, 2008). It can no longer stay unnoticed that corporate responsibility is becoming increasingly important.
Looking at the energy industry specifically, Pätäri, Arminen, Tuppura and Jantunen (2014) state that it is no longer an option to only focus on deriving profit. Firms in the energy industry should be implementing multiple aspects of corporate social responsibility (CSR) in their strategy and at the same time being profitable. They also mention the difficulty of these conditions.
In recent decades, many studies have been conducted into the relationship between CSR and corporate financial performance (CFP), which will be described in the literature review. According to Dimson & Karakas (2015) the outcome of these studies has been ambiguous and inconsistent. There is also little literature that focuses on the difference between industries that have a high CSR score and industries that have a low CSR score.This score measures to what extent companies have integrated CSR in their business operations. In this research it is measured by adding up the total measured strengths in certain dimensions of CSR and subtracting the concerns. This is a commonly used measurement of CSR (e.g. Chan, Chou & Lo, 2017)
This research extends the existing research about the relation between CSR and CFP by focussing on two specific industries. The choice for the industries is based on the fact energy is one of the most polluting industries. The pharmaceuticals, biotechnology and life science industry is not as polluting as the energy industry, which will cause a difference in the expectation of the CSR scores. It could be that being responsible already means that money does not have to be spent on CSR and ensures a higher CFP. Another reasoning could be that polluting industries will compensate for this by engaging more in other CSR dimensions. Therefore the following research question will be asked: “What is the influence of corporate
social responsibility on corporate financial performance in the energy compared to the pharmaceuticals, biotechnology and life science industry?” In this research, CFP will be
measured by accounting-based measurements, which will be explained more in detail in the literature review.
In this research 4 regressions were performed. First, two regressions are done to test whether CSR affects CFP and a large sample of all available industries is taken. Second, a subset of the complete dataset is taken with only the energy and pharmaceuticals, biotechnology & life science industry, in order to answer the second hypothesis. Results for both hypotheses gave an insignificant outcome.
This thesis is divided in 5 sections. After this section, a description of the existing literature concerning the relation between CSR and CFP is given, after which the hypotheses will be formed. In the third section the methodology will be presented where the method used in this research will be explained. Variables are defined and the information about the data and sample collection will be provided. Section 4 shows the results. A correlation table is provided, and the regression results are discussed. The last section will give the conclusion of this research.
2. Literature review
The following section describes the existing literature in the field of CSR and CSP. First the concepts CSR and CSP will be defined. Then the different ways of measuring CSR and CSP are described. Finally, an overview of existing literature that researches the relation between CSR and CSP is given.
2.1 Corporate Social Responsibility:
The definition of Corporate Social Responsibility (CSR) has evolved over time and varies across industries. Daszynska-Zygadlo, Slonski and Zawadski (2016) state that agreeing on a general definition that is precise at the same time is very difficult. Some researchers even claim that there is not a definition of CSR (Jackson and Hawker, 2001). According to Dahlsrud (2006) that is not the case, instead the problem is caused by the existence of too many definitions. Definitions that are often biased and not supported by empirical evidence. His research concluded that even though there are many different definitions, they consistently refer to the following five dimensions: the stakeholder dimension, the social dimension, the economic dimension, the voluntariness dimension, and the environmental dimension. CSR is also specified as corporate social performance (CSP) in multiple studies (Daszynska-Zygadlo et al., 2016; Waddock & Graves, 1997; Tsoutsoura, 2004)
According to Berman, Wicks, Kotha and Jones (1999), the idea behind CSR is that corporations have moral responsibilities towards society that go beyond the goal of making profits for their owners and shareholders. The most frequently used definition of CSR is the definition given by the European Commission (2008) which states: “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 in this thesis.
2.2 Measuring CSR
Galant & Cadez (2008) give two reasons for the complication of measuring CSR. First, there is no consistently used definition. Secondly, the concept consists of several dimensions that are relatively heterogeneous. As a result, CSR is measured by using many different approaches in previous studies.
Some researchers tried to bring those different elements together into one specific measurement (Mahoney and Roberts, 2007). Tsoutsoura (2004) mentions the use of subjective
indicators, such as surveys. He also states that the measurements of these indicators are significantly unclear.
According to Marsat and Williams (2011) most academic studies focus on the three pillars: Environmental, Social and Corporate Government (ESG). Environmental is the most considered link and is about companies trying to do business without affecting nature. Social refers to the intern and extern relationships of the company. Governance mainly concentrates on the treatment of shareholders and agency problems.
Database MSCI, formerly KLD, measures CSR by the use of 7 dimensions which are: community relations, customers, diversity, employee relations, environment, human rights and investors. The MSCI database is commonly used in previous studies (Schreck, 2011; Jo & Na (2012) and will be used in this thesis.
2.3 Measuring Corporate Financial Performance:
Measuring corporate financial performance (CFP) is, unlike measuring CSR less uncomplicated. According to Margolis, Elfenbein & Walsh (2007) CFP can be measured in two broad categories: accounting-based measures of financial returns (e.g., Return on Assets, Return on Equity) and market-based measures of financial value (e.g., stock returns, market/book value ratio). They also state that CSR is better in predicting accounting-based measures than market-based measures. Galant & Cadez (2017) state that accounting-market-based measures are reasonably comparable and that they are better in measuring unsystematic observations of CSR compared to market-based measures. Besides, accounting-based measures are easily accessible and related to economic returns (Richard, Devinney, Yip & Johnson, 2009). Therefore, CFP will be measured on the basis of ROA and ROE in this thesis.
2.4 The relationship between CSR and CFP:
Most studies show a positive relation (e.g. Waddock & Graves, 1997; Tsoutsoura, 2004; Margolis et al., 2007). A negative relation is rarely shown, but also found in some studies (e.g. Vance, 1975). In addition, no (clear) relationship is also a common outcome (e.g. Brine, Brown & Hackett, 2007; Schreck, 2011; Daszynska-Zygadlo et al., 2016).
2.4.1 Positive relationship
Waddock and Graves (1997) concluded that there is a positive relation between CSR and financial performance. Their conducted research did not only research whether CSR affects
CFP, but also the other way around. Additionally, their results showed that there is indeed a positive relation when CSR is both the dependent and the independent variable. They argue that the outcome makes sense, since financially healthy companies are more capable of making investments in CSR activities, which have more long-term impact, in comparison to companies that are in financial trouble. Tsoutsoura (2004) conducted a similar research in which CSR is positively linked to financial performance, while in this research CSR is also both the dependent variable and the independent variable. Also, Margolis et al. (2007) also found that there is a strong link when the effect of financial performance on CSR was tested instead of the other way around.
Malik (2005) states that both the theoretical and the empirical findings imply that CSR significantly increases company value by multiple outcomes. The possibility that social goals may align with corporate goals where CSR is used as a strategic tool to maximize value, is also mentioned in their research. This would imply that companies that have a higher CSR score, have more chances of performing better on a corporate level.
2.4.2 Negative relationship
Vance’s (1975) dated research showed a negative relation. This however does not give an adequate view of the current situation. Nollet, Filis and Mitrokostas (2016) use ESG ratings as a measurement for CSR in their research. They do find a negative effect of the lagged ESG coefficient on CFP, but it is not significant. Also, in the study of Mittal, Sinha and Singh (2008) negative relations between CSR and business performance are found. They state that although their results cannot conclusively prove the negative relation, they are however strong evidence against the claim that CSR positively influences financial performance of companies.
The few studies that imply a negative relationship between CSR and CFP argue that socially responsible companies are drawing resources and management effort away from the main areas of their business, which eventually will result in lower profit. From that point of view, managers cannot make both social and competitive improvements (Klassen and Whybark, 1999).
2.4.3 No significant relationship
Schreck (2012) did not find causality in the CSR-CFP relation. According to him there are two possible explanations. The first would be that social and environmental product features are related to extensive expenditure. This does not go along with compensating returns. Secondly,
managers may use excessive social expenditure to compensate for their low financial performance. Another remarkable result is that although it was not as significant as they hoped for, high-quality CSR reports showed a stronger positive influence than other cases.
Daszynska-Zygadlo et al. (2016) also researched the link between CSR and financial performance. They divided their sample into ten industry subsamples, in order to confirm heterogeneity across industries. They analyzed the impact of each category of ESG on the financial performance of the industries. The category of social performance was found to be relatively the least important. Most of their results are related to the environmental performance category, which implies that this category affects financial performance most. In eight sectors the hypothesis is rejected.
2.5 Hypotheses
In this section the hypotheses will be formed on the basis of the literature review. First, a short summary of the literature will be given.
As discussed earlier, previous studies have shown ambiguous outcomes. Where the majority of the results show a positive relation between CSR and financial performance (e.g. Waddock & Graves, 1997; Malik, 2005), a negative relationship is also found by some (e.g. Vance, 1975). No relation is also a common result (Schreck, 2011; Daszynska-Zygadlo et al., 2016). To conclude, previous research does not give conclusive evidence on a relationship between CSR and CFP (Galant & Cadez, 2008).
Based on above theory and keeping in mind that the majority of the studies show a positive relationship, we can form the following hypotheses:
H1: The CSR score does have a positive effect on financial performance.
Daszunska-Zygadlo et al. (2016) could not find a clear relationship. They found the CSR dimension “social” to have the least impact on companies’ value measures, on the other hand their research showed that the impact of CSR over different industries appeared to be inconsistent. As stated in the introduction, the energy industry is one of the most polluting ones. Expected is that polluting industries will try to compensate for this by engaging more in other CSR aspects. The following hypotheses will be tested:
H2: The CSR score does have more effect on financial performance in the energy industry compared to the pharmaceuticals, biotechnology and life science industry.
3. Methodology
3.1 Dependent variable
The dependent variable of the model is CFP. In the literature review it is concluded that accounting-based measures are better in measuring CSR and easily accessible. Therefore, Return on Equity (ROE) and Return on Assets (ROA) are used as a proxy for CFP. The calculation of these measures is as follows:
𝑅𝑂𝐸 = 𝑁𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑆ℎ𝑎𝑟𝑒ℎ𝑜𝑙𝑑𝑒𝑟′𝑠 𝑒𝑞𝑢𝑖𝑡𝑦
𝑅𝑂𝐴 = 𝑁𝑒𝑡 𝑖𝑛𝑐𝑜𝑚𝑒 𝑇𝑜𝑡𝑎𝑙 𝐴𝑠𝑠𝑒𝑡𝑠
3.2 Independent variable
Corporate social responsibility will be the independent variable in the model. As stated in the literature review, CSR will be measured in seven dimensions using the MSCI ESG database Waddock and Graves (1997) state that the MSCI database is the best measure of CSR compared to other measures because of four reasons. First, all companies in the S&P 500 are rated, so they have a wide range of companies. Second, they are rated on multiple CSR dimensions. Third, an independent group of researchers apply the same set of criteria to related companies. The last reason is that the scoring is being done consistently, with data gathered from internal and external sources to the company.
3.2.1 Industry dummy and interactive dummy variable
In model 3 and 4 we also measure the effect of the industry specific CSR. For this, a dummy is generated for the energy industry. The energy dummy variable will take the value 1 if the firm is in the energy industry and 0 otherwise. Since the dataset in model 3 and 4 only consists of firms from the energy industry and the pharmaceuticals, biotechnology & life science industry, pharmaceuticals, biotechnology & life science firms will take the value 0. Also, the interactive dummy variable is generated by multiplying the energy dummy variable times the CSR score. This means that if a firm is in the energy industry, the interactive variable takes the value of the CSR score. Pharmaceuticals, biotechnology & life science firms will take the value 0.
3.3 Control variables
Besides the dependent and independent variables, control variables are included in the model. They are important as they may influence the relationship between CSR and CFP. It serves to minimize omitted-variables bias. In multiple studies the following control variables are commonly used: RISK, SIZE, INDUSTRY and R&D (McWilliams & Siegel, 2000; Andersen & DeJoy, 2011)
3.3.1 Size
Size is proven to have a significant effect on financial performance when looked at the relationship between CSR and CFP. An argument given by Waddock and Graves (1997) is that bigger companies have more pressure of acting socially responsible, because they draw more attention from external constituents. This is also expected by Servaes & Tamayo (2013).
Another reason for this significant effect could be the fact that larger companies have more resources available for social investments (Wu, 2006)
The most common ways to measure size are total assets, total sales and total number of employees (e.g. Waddock and Graves, 1997) In line with Mahoney & Roberts (2007) size will be measured by the natural logarithm of the total assets.
3.3.2 Risk
Margolis et al. (2007) state that it seems more likely that low-risk companies that are stable are more likely to take part in CSR. They have a more steady and certain financial cash flows, which is why they face less risk concerning the future CSR opportunities (Orlitzky & Benjamin, 2001).
A regularly used proxy for risk is the total debt to total assets ratio (e.g. Andersen & Dejoy, 2011), which will also be used as a measurement for risk in this thesis.
3.3.3 R&D
Van Beurden and Gössling (2008) conclude by reviewing existing studies, that R&D significantly affects financial performance. McWilliams & Siegel (2000) also mention the fact that R&D is an important determinant of firm performance. They state that their model was incorrectly specified since the control variable R&D was omitted from the model. They also conclude that this misspecification must have led to an upward bias in results on the link
between financial performance and CSR. In this thesis R&D will be included in the model, by dividing the total R&D expenses by total sales.
3.4 Model
This research uses regression analysis to test whether CSR affects the financial performance of a company. STATA16 is used to perform an ordinary least squares (OLS) regression. In the model financial performance is presented by ROE and ROA, which are the dependent variables.
Model 1 and 2 are estimated in order to test the first hypothesis to the effect of CSR on CFP.
(1) 𝑅𝑂𝐸𝑖𝑗= 𝛽0+ 𝛽1∗ 𝐶𝑆𝑅 + 𝛽2∗ 𝑆𝐼𝑍𝐸 + 𝛽3∗ 𝑅𝐼𝑆𝐾 + 𝛽4∗ 𝑅&𝐷 + 𝜀
(2) 𝑅𝑂𝐴𝑖𝑗= 𝛽0+ 𝛽1∗ 𝐶𝑆𝑅 + 𝛽2∗ 𝑆𝐼𝑍𝐸 + 𝛽3∗ 𝑅𝐼𝑆𝐾 + 𝛽4∗ 𝑅&𝐷 + 𝜀
Model 3 and 4 are estimated in order to test the second hypothesis to test if having a high CSR score in the energy industry has more impact on CFP compared to the pharmaceuticals, biotechnology & life science industry.
(3) 𝑅𝑂𝐸𝑖𝑗= 𝛽0+ 𝛽1∗ 𝐶𝑆𝑅 + 𝛽2∗ 𝑆𝐼𝑍𝐸 + 𝛽3∗ 𝑅𝐼𝑆𝐾 + 𝛽4∗ 𝑅&𝐷 + 𝛽5∗ 𝐸𝑁𝐸𝑅𝐺𝑌 + 𝛽6∗
𝐸𝑁𝐸𝑅𝐺𝑌𝑥𝐶𝑆𝑅 + 𝜀
(4) 𝑅𝑂𝐴𝑖𝑗= 𝛽0+ 𝛽1∗ 𝐶𝑆𝑅 + 𝛽2∗ 𝑆𝐼𝑍𝐸 + 𝛽3∗ 𝑅𝐼𝑆𝐾 + 𝛽4∗ 𝑅&𝐷 + 𝛽5∗ 𝐸𝑁𝐸𝑅𝐺𝑌 + 𝛽6∗
𝐸𝑁𝐸𝑅𝐺𝑌𝑥𝐶𝑆𝑅 + 𝜀
Where i=company, t=time
ROE = Return on Equity ROA = Return on Assets
CSR = Corporate social responsibility score SIZE = Natural logarithm of total assets RISK = Total debt to asset ratio R&D = R&D expenses to total sales
ENERGY = Dummy variable which takes the value 1 if the company is in the energy industry and 0 otherwise ENERGYxCSR = Energy dummy * corporate social responsibility score
3.5 Sample and data collection
Data collected for this research is in the form of panel data. Panel data is commonly used in similar researches (e.g. Marsat & Williams, 2011). As mentioned earlier CSR data is collected from the MSCI ESG. An adjusted CSR score is calculated based on the seven dimensions by
taking the sum of the strengths divided by the total amount of the measured strengths and subtracting the sum of the concerns divided by the total amount of the measured concerns for each dimension. As already mentioned in the introduction, this measurement is commonly used in other researches. Table 1 in the appendix gives an overview of these dimensions and the corresponding strengths and concerns within each dimension. By identifying firm’s tickers, financial data collected from COMPUSTAT is linked to the CSR data. With the variables found on COMPUSTAT, dependent variables and control variables are generated. Industries are categorized by applying global industry classification (GIC) codes, in this way energy and transport industry firms are identified. Table 2 in the appendix gives an overview of the distribution of industries per year and in total. The data is taken from 2011 until 2016, since this research is performed to show the most recent CSR-CFP relation. Also, North American firms are used in this sample. Other than that, companies are only selected based on their availability.
First, firms with missing ROE or ROA are excluded from the sample This resulted in a sample consisting of 12366 observations on firm’s CSR score and ROA. ROE counts 12365 observations. Second, industry dummies are generated by giving the dummies the value 1 if they equal the corresponding industry group and 0 otherwise. After that, the interactive dummy variable is generated, where we take the dummy value and multiply it with the corresponding CSR score. For the third and fourth regression a subset of the full dataset is taken. Only firms in the energy industry and pharmaceuticals, biotechnology and life science industry are observed, and all other firms are excluded from the dataset. From table 2 in the appendix it can be observed that these industries This is done in order to find the difference in effect of the industries. Last, all big outliers were excluded from the model, in order to prevent potential bias.
Table 1:
Descriptive Statistics
Variable Obs Mean Std.Dev. Min Max
CSR 12366 .152 1.167 -1.333 4.5 ROE 12365 .076 .348 -1.938 1.444 ROA 12366 .031 .109 -.567 .281 SIZE 12366 7.763 1.741 4.033 12.376 RISK 10970 .226 .211 0 .903 RD 6600 .196 .792 0 6.703 Energy 12366 .051 .22 0 1 Energy*CSR 12366 .019 .36 -2.917 6.477
Full sample, used in regression 1 and 2
In table 1, the descriptive statistics of the sample are shown, which is meant to increase the understanding of the data used in this research. These observations are already excluded big outliers. Including R&D in the model would decrease the sample size, however excluding R&D from the sample set could lead to omitted variable bias and wrong conclusions could be drawn. Since the sample would still include 6600 observations, which is still sufficient for this research, R&D is included in the model.
The table shows that the average Return on Equity is 7,6% and the average Return on Equity 3,1%, with a standard deviation of respectively 0.348 and 0.109. The average CSR score is 0.152 with a relatively high standard deviation of 1.167, while the max score is 4.5. Looking at the industry specific CSR score, we see that the energy industry scores on average 0.019 and the pharmaceuticals, biotechnology and life science industry even lower with an average of 0.001. This contradicts expectations at the beginning of this research. Both standard deviations are relatively small compared to the standard deviation of the average CSR of all industries.
Hypotheses are stated in the literature review. In order to test these, the betas of CSR and the interactive dummy variables are measured and tested if these significantly differ from zero. This can be written as:
𝐻10: 𝛽𝐶𝑆𝑅 = 0 𝐻11: 𝛽𝐶𝑆𝑅 > 0
𝐻20: 𝛽𝐶𝑆𝑅,𝐸𝑁𝐸𝑅𝐺𝑌 = 0 𝐻21: 𝛽𝐶𝑆𝑅,𝐸𝑁𝐸𝑅𝐺𝑌 > 0
4. Results 4.1 Correlation
Correlation between the variables is presented in table 2 below. Farrar and Glauber (1967) state that correlations should not exceed a value higher than 0.8, otherwise multicollinearity could exist. With multicollinearity, the linear relation between two or more variables is meant (Alin, 2010). No correlation values exceeding 0.8 are found in table 2.
From the table we observe that both ROE and ROA are positively correlated with CSR with a correlation of 0.137 and 0.152, respectively. This is in line with most of the previous researches discussed in the literature review. The interactive dummy variables both have a positive correlation with ROE and ROA. Also, the natural logarithm of SIZE also positively correlates with CSR, which is in line with previous mentioned findings, that state that bigger companies have more pressure to act responsible and at the same time have more resources available for social investments. Remarkable is the positive correlation between CSR and RISK. Which suggests that firms that have a higher leverage ratio score higher on CSR.
Table 2:
Correlation of the variables
Variables CSR ROE ROA SIZE RISK R&D Energy Energy
*CSR CSR 1.000 ROE 0.137 1.000 ROA 0.152 0.560 1.000 SIZE 0.571 0.218 0.302 1.000 RISK 0.067 -0.044 -0.080 0.331 1.000 RD -0.102 -0.301 -0.578 -0.276 -0.086 1.000 Energy 0.085 0.006 0.015 0.098 -0.028 -0.026 1.000 Energy*CSR 0.211 0.021 0.028 0.133 -0.015 -0.013 0.398 1.000 4.2 VIF
In section 4.1 we found a high correlation between SIZE and CSR. Although we stated that this should not be a problem for multicollinearity since it does not exceed the value 0.8, a VIF test is done to secure this. The VIF test calculates variance inflation factors (VIFs) for the independent and control variables that are included in a linear regression model. Although there are no clear rules about what value of VIF should lead to concerns, Field (2014) states that a value of 10 should cause concerns about possible multicollinearity. From the results, represented in table 3, we can conclude that there is no strong evidence for multicollinearity.
Table 3:
Variance inflation factor
VIF 1/VIF SIZE 1.810 0.553 CSR 1.530 0.652 RISK 1.150 0.871 R&D 1.080 0.923 Mean VIF 1.390 4.3 Regression results
Results of the OLS regression on financial performance are presented in table 4 below. Regression 1 and 2 test model 1 and 2. Table 3 in the appendix contains the full regression output, including the dummy variables for all industries. Besides the effect of CSR, they measure the effect of SIZE, RISK, R&D and the industries. In regression 3 and 4 the corresponding models are tested, where we only observe the energy industry and the pharmaceuticals, biotechnology & life science industry. Both the energy dummy and the interaction dummy variable are included in these models. These dummies are added in order to test for the difference in effect between the energy industry and the pharmaceuticals, biotechnology and life science industry.
First, we look at the R-squared. This variable shows the percentage of the dependent variable that is explained by the independent variables. Adding an extra variable that improves the model leads to an increase of R2. Looking at the R2 values of all four models, we can
conclude that looking at a specific measurement of CFP the third and fourth model are better at predicting CFP, therefore we can conclude that adding the interactive variable improves the model. Although, the model is better in predicting CFP by measuring the ROA instead of the ROE. It can be observed from the table that the third model explains for 1,5% and the fourth model for 30,4% of the variation of CFP.
From the table we observe inconsistent effects of CSR on corporate performance. In model 1 and 3, CSR shows a positive effect on ROE, but it is only significant in the third model (t=0.76, p=0.447; t=1.67, p=0.096). Regression results show a significant negative effect of CSR on ROA in both model 2 and 4 (t=-3.34, p<001; t=-3.15, p=0.002). Besides the fact that we only found three of the four variables to be significant, these effects found are also contradictory. The coefficient of the interactive dummy variable is found to be negative in both model 3 and 4, but the coefficients are not significant (t=-1.51, p=0.132; t=-1.53, p=0.126).
The effect of the control variables is almost consistent over the regressions. We find a positive effect of SIZE in model 1, 2 and 4 and a negative effect only in model 3. The coefficients of RISK give a negative effect, also in model 1, 2 and 4 and a positive effect only
in model 3. The effect of R&D is almost equal to 0. For most control variables, we only find significant coefficients in the second and fourth model, from which it can be observed that these variables significantly impact return on assets. An exception is the significant effect of SIZE in model 3. The effect of SIZE on corporate performance is in line with earlier mentioned previous research, a clear effect of R&D was not given in earlier mentioned research, but coefficients of RISK contradicts previous outcomes.
Altogether we cannot conclude from the regression results that a higher CSR results in a higher CFP. Although coefficients suggest a negative effect of CSR on ROA, outcomes of the CSR coefficient in the model with ROE as a measure for CFP suggest a positive effect. Besides, not all results found are significant. Also, we do see a consistent negative effect of the interactive variable but fail to find statistical significance. From this we can only observe that there seems to be a trend of having a negative impact of CSR in the energy industry on CFP, but we cannot assume this.
Table 4:
Regression output model 1, 2, 3 and 4
(1) (2) (3) (4)
ROE ROA ROE ROA
CSR 0.0287 -0.00572**** 0.373* -0.033*** (0.76) (-3.34) (1.67) (-3.15) Energy*CSR -0.175 -0.014 (-1.51) (1.53) Energy 0.411* -0.007 (1.75) (-0.29) SIZE 0.0169 0.0282**** -0.265 0.090*** (0.34) (12.32) (-1.31) (11.54) RISK -0.190 -0.120**** 2.182* -0.218*** (-0.51) (-8.71) (1.89) (-5.02) RD 0.000146 -0.000288*** -0.001 0.000** (0.36) (-3.27) (-1.25) (-2.42) _cons -0.0209 -0.156**** 1.410 -0.663*** (-0.07) (-8.84) (1.09) (-11.92) N R2 6562 0.0033 6563 0.224 747 0.015 747 0.304 t statistics in parentheses * p <0.1, ** p < 0.05, *** p < 0.01, **** p < 0.001,
5. Conclusion
The aim of this research is to examine what the influence of corporate social responsibility on the corporate financial performance of firms in the energy and pharmaceuticals, biotechnology & life sciences industry is. A large sample is examined for the first two regressions. The third and fourth regression contained a smaller sample in which only two industries are observed. From looking at theR2 we concluded that using ROA as a measurement of CSR improves the
model compared to the model in which ROE measured CSR.
From the regression table we observed that many control variables are significant, which confirms their contribution to the CSR-CFP relation. The results of the main independent variables are inconsistent, we found both negative and negative effects of CSR on CFP. Only some are found significant. Because of these ambiguous results, we fail to reject the first null hypothesis and thus we cannot assume that CSR has a positive effect on corporate financial performance. This finding is in line with some of the existing literature about the relationship between CSR and CFP.
Looking at the interactive dummy variable, we found negative effects on both ROA and ROE but failed to find significant effects. Despite these consistent results, we also fail to reject the second null hypothesis and therefore cannot assume that a higher CSR score in the energy industry has more effect on corporate financial performance.
These above-mentioned outcomes suggest that there is an extra effect of CSR on CFP in the energy industry, but a negative effect instead of positive. Also, there seems to be a difference in effect between the CFP measurements. But none of these outcomes are significant enough to draw conclusions.
Reasons for these non-significant and ambiguous outcomes could be that having a high CSR score was quite an achievement once. At this moment, being socially responsible is more of a requirement and not very exceptional anymore. Expectations are different nowadays; people already expect companies to be socially involved. This thought is comparable to Herzberg's motivation-hygiene theory.
5.1 Limitations
From the descriptive statistics table we observed that the energy industry has a higher average CSR score compared to the pharmaceuticals, biotechnology and life science industry. This is different from what was predicted in the introduction. The energy industry was taken as a highly polluting industry and the pharmaceuticals and life science industry as a less polluting industry. The reason for this unexpected result might be that firms in the energy industry try to improve
their CSR scores more at other dimensions. Also, CSR is measured by taking the sum of the measured strengths and subtracting the sum of the measured concerns in this research. CSR can be and is already measured by adding a score between 0 and 100 for each firm. This would result in a more consistent measure of CSR.
Future research can be done comparing a high CSR scoring industry with a low CSR scoring industry. Although this was the intention, we found the average CSR scores of the main industries in this research to be almost equal.
References
Alin, A. (2010) Multicollinearity. WIREs Computational Statistics, 2(3), 370- 374
Ambec, S. & Lanoie, P. (2008). DOES IT PAY TO BE GREEN? A SYSTEMATIC OVERVIEW. Academy of Management Perspectives, 22(4), 45-62
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
Berman, S. L., Wicks, A. C., Kotha, S., & Jones T. M. (1999). Does Stakeholder
Orientation Matter? The Relationship between Stakeholder Management Models and Firm Financial Performance, Academy of Management Journal, 42, 488–506
Chan, C.Y., Chou, D.W. & Lo, H.C. (2017). Do financial constraints matter when firms engage in CSR? The North American Journal of Economics and Finance, 39, 241-259 Daszynska-Zygadlo, K., Slonski, T., & Zawadzki, B., (2016). The Market Value of CSR
Performance Across Sectors. Inzinerine Ekonomika-Engineering Economics, 27(2), 230–238
Dimson, E., Karakas, O., & Li, X. (2015). Active Ownership, The Review of Financial
Studies, 28(12), 3225–3268
EU. (2008). European Competitiveness report. Retrieved from: http://aei.pitt.edu/45440/1/Competitiveness_2008.pdf
Farrar, D. E., & Glauber, R. R. (1967). Multicollinearity in Regression Analysis: The Problem Revisited. The Review of Economics and Statistics, 49(1), 92-107.
Field, A. (2014). Discovering Statistics Using IBM SPSS Statistics. London, United Kingdom: Sage Publications.
Galant, A. & Cadez, S. (2008). Corporate social responsibility and financial performance relationship: a review of measurement approaches, Economic Research-Ekonomska
Istraživanja, 30(1), 676-693
Jo, H. & Na, H. (2012) Does CSR Reduce Firm Risk? Evidence from Controversial Industry Sectors. J Bus Ethics, 110, 441–456
Klassen R.D. & Whybark D.C. (1999). The impact of environmental technologies on manufacturing performance. Academy of Management Journal, 42(6), 599–615 Margolis, J. D., Elfenbein, H. A., & Walsh, J. P. (2007). Does it pay to be good? A
meta-, analysis and redirection of research on the relationship between corporate social and financial performance. Harvard University, University of California – Berkeley,
Marsat, S. & Williams, B. (2011). CSR and Market Valuation: International evidence
International Conference of the French Finance Association (AFFI), 123, 29-42
McWilliams, A., & Siegel, D. (2000). Corporate social responsibility and financial
performance: Correlation or misspecification? Strategic Management Journal, 21(5), 603-609
Mittal, R.K., Sinha, N. and Singh, A. (2008), An analysis of linkage between economic value added and corporate social responsibility, Management Decision, 46(9), 1437-1443 Nollet, J., Filis, G., & Mitrokostas, E. (2016). Corporate social responsibility and financial
performance: A non-linear and disaggregated approach. Economic Modelling 52, 400-407
Orlitzky, M., Schmidt, F.L., & Rynes, S.L. (2003). Corporate social and financial performance: a meta-analysis. Organization studies, 24(3), 403-441.
Pätäri, S., Arminen, H., Tuppura, A., Jantunen, A. (2014), Competitive and responsible? The relationship between corporate social and financial performance in the energy sector,
Renewable and Sustainable Energy Reviews, 73, 142-154
Richard, P. J., Devinney, T. M., Yip, G. S., & Johnson, G. (2009). Measuring organizational performance: Towards methodological best practice. Journal of management, 35(3), 718-804
Sassen, R. & Hinze, A.K. (2016). "Impact of ESG factors on firm risk in Europe," Journal of
Business Economics, Springer, 86(8), 867-904
Schreck, P. (2011). Reviewing the business case for corporate social responsibility: New evidence and analysis. Journal of Business Ethics, 103(2), 167-188
Servaes, H. & Tamayo, A. (2013) The Impact of Corporate Social Responsibility on Firm Value: The Role of Customer Awareness. Management Science, 59(5), 1045– 1061
Tsoutsoura, M. (2004). Corporate social responsibility and financial performance. Center for
Responsible Business.
Van Beurden, P., & Gössling, T. (2008). The worth of values–a literature review on the relation between corporate social and financial performance. Journal of Business
Ethics, 82(2), 407- 424
Vance, S. C. (1975). Are socially responsible corporation’s good investment risks?
Management review, 64(8), 19-24
Waddock, S. A., and Samuel B. Graves (1997). “The corporate social performance financial performance link.” Strategic Management Journal, 18(4), 303-319
Wu, M. L. 2006. Corporate social performance, corporate financial performance, and firm size: A meta-analysis. Journal of American Academy of Business, 8(1), 163–171
Appendix Table 1
MSCI ESG performance indicators
Strengths Concerns
Environmental Environmental Opportunities; Pollution & waste; Climate Change; Environmental Management Systems; Natural Capital
Toxic Emmissions and Waste; Energy & Climate Change; Biodiversity & Land Use; Operational Waste; Supply Chain Management; Water Stress; Environment
Community Community Engagement;
Innovative Giving; Community Impact
Impact on Community
Human Rights Indigenous Peoples Relations Strenght; Human Rights Policies & Initiatives
Civil Liberties; Human Rights Concerns; Human Rights – Other Concerns
Employee Relations Union Relations; Cash Profit Sharing; Employee Involvement; Employee Health & Safety; Supply Chain Labor Standards; Human Capital Development; Labor Management; Controversial Sourcing; Human Capital – Other Strengths
Collective Bargaining & Unions; Health & Safety; Supply Chain Labor Standards; Child Labor; Labor Management Relations; Labor Rights & Supply Chain – Other Concerns
Product Product Safety and Quality; Social Opportunities; Product Safety
Product Quality & Safety; Marketing & Advertising; Anticompetitive Practices; Customer Relations; Privacy & Data Security; Other concerns
Governance Corruption & Political Instability; Financial System Instability
Governance Structures; Controversial Investments; Bribery & Fraud; Governance – Other Concerns
Diversity Representation; Board Diversity – Gender
Discrimination & Workforce Diversity; Board Diversity - Gender
If a company meets the assessment criteria established for an indicator, then this is signified with a “1”.
If a company does NOT meet the assessment criteria established for an indicator, then this is signified with a “0”.
Table 2
Distribution of industries per year
GIC Groups year
2011 2012 2013 2014 2015 2016 Total 1010 124 117 117 116 84 73 631 1510 123 120 130 126 98 94 691 2010 219 210 205 204 176 171 1185 2020 79 81 77 79 66 62 444 2030 47 44 46 47 41 38 263 2510 27 26 28 29 24 26 160 2520 76 73 64 62 55 54 384 2530 86 78 68 66 58 56 412 2540 16 14 8 13 8 4 63 2550 103 100 100 97 89 86 575 3010 20 21 19 19 16 16 111 3020 58 58 55 56 49 49 325 3030 19 19 17 18 16 15 104 3510 166 150 132 131 115 117 811 3520 163 138 108 118 104 91 722 4010 196 200 150 149 138 147 980 4020 113 109 100 90 77 79 568 4030 92 89 86 86 71 70 494 4040 15 16 15 14 6 0 66 4510 148 140 138 127 105 94 752 4520 135 125 110 104 91 78 643 4530 92 87 70 68 61 56 434 5010 31 28 24 22 23 23 151 5020 44 47 48 51 42 40 272 5510 74 73 75 76 65 62 425 6010 108 114 122 126 115 115 700 Total 2374 2277 2112 2094 1793 1716 12366 Where:
1010 = Energy 3510 = Health Care Equipment & Services
1510 = Materials 3520 = Pharmaceuticals, Biotechnology & Life Science
2010 = Capital Goods 4010 = Banks
2020 = Commercial & Professional Services 4020 = Diversified Financials
2030 = Transportation 4030 = Insurances
2510 = Automobiles & Components 4510 = Software & Services
2520 = Consumer Durables & Apparel 4520 = Technology Hardware
2530 = Consumer Services 4530 = Semiconductors & Semiconductor Equipment
2550 = Retailing 5010 = Communication Services
3010 = Food & Staples Retailing 5020 = Media & Entertainment
3020 = Food, Beverage & Tabacco 5510 = Utilities
Table 3
Full regression output of regression 1 and 2
(1) (2) ROE ROA CSR 0.0287 -0.00572*** (0.76) (-3.34) SIZE 0.0169 0.0282*** (0.34) (12.32) RISK -0.190 -0.120*** (-0.51) (-8.71) RD 0.000146 -0.000288** (0.36) (-3.27) IND1_dummy -0.0831 -0.0338* (-0.57) (-2.33) IND2_dummy 0.350 0.0155 (1.55) (1.89) IND3_dummy 0.0231 0.0164* (0.14) (1.97) IND4_dummy -0.0567 0.0344*** (-0.32) (3.45) IND5_dummy 0 0 (.) (.) IND6_dummy -0.0522 0.00774 (-0.34) (0.68) IND7_dummy 0.188 0.0577*** (1.73) (5.79) IND8_dummy -0.188 0.0594*** (-0.56) (6.13) IND9_dummy 0.0734 0.0454*** (0.70) (4.99) IND10_dummy -0.534 0.0114 (-0.90) (1.11) IND11_dummy 0.455 0.0423*** (1.41) (4.27)
IND12_dummy 0.131 0.0728*** (0.65) (5.63) IND13_dummy -0.179 -0.0134 (-1.08) (-1.34) IND14_dummy 0.0882 -0.130*** (0.39) (-9.62) IND15_dummy 0 0 (.) (.) IND16_dummy -0.0439 -0.00863 (-0.34) (-0.76) IND17_dummy 0.0692 0.0969*** (0.61) (10.64) IND18_dummy -0.0871 0.0205* (-0.67) (2.23) IND19_dummy -0.0332 0.0121 (-0.29) (1.34) IND20_dummy 0.00304 0.0202* (0.03) (2.03) IND21_dummy -0.127 -0.0128 (-0.76) (-0.82) IND22_dummy -0.0849 -0.0269 (-0.84) (-1.89) IND23_dummy 0 0 (.) (.) IND24_dummy 0.0462* 0.0152*
(2.02) (2.03) _cons -0.0209 -0.156*** (-0.07) (-8.84) N R2 6562 0.0033 6563 0.224
