Socially responsible investment and Financial Performance
A study to indicate the causal relationship between SRI and the financial performance.
Bachelor Thesis
Student name: Femke Ebbelaar Student number: 11015489 Date of submission: 13 July 2018
Bachelor: Economics & Business – Finance and Economics Track Supervisor: dhr. P.M. (Pascal) Golec
Second reader:
2018
08
Fall
Abstract
In this study, the causal relation between socially responsible investment and firm’s return on assets is tested with the use of statistical techniques. Consistent with prior research, there exists a correlation between the firm’s return on assets and socially responsible investment. The effect is slightly negative. The causality between social responsible investment and financial performance is tested upon interaction effects. This study is looking at data form U.S. firms from 1991 till 2013, and shows a significant positive moderator effect of leverage and the ratio of research and development expense to sales. Time trend has no influence on correlation between ethical investing and financial profitability.
Statement of Originality
This document is written by Student Femke Ebbelaar who declares to take full responsibility for the contents of this document.
Table of Contents
ABSTRACT ... 2
INTRODUCTION ... 4
1.1 PROBLEM STATEMENT ... 6
1.2 RELEVANCE ... 7
2.1 SOCIAL RESPONSIBLE INVESTING ... 8
2.2 CORRELATION SRI AND FIRM PERFORMANCE ... 8
2.3 PREVIOUS RESEARCH ON THE VARIABLES ... 10
2.3.1 Age ... 12
2.3.2 Research & Development ... 12
2.3.3 Size ... 13
2.3.4 Leverage ... 13
2.3.5 Time trend ... 14
2.3 CONCEPTUAL MODEL ... 15 3 METHODOLOGY ... 16 3.1 DATA SOURCES ... 16
3.2 REGRESSIONS: DATA ANALYSIS ... 16
4 RESULTS ... 18
4.1 DESCRIPTIVE STATISTICS ... 18
4.1.1 SRI & Firm performance ... 19
4.1.2 Research & Development ... 19
4.1.3 Leverage ... 20
4.1.4 Time trend ... 20
5 DISCUSSION ... 21 5.1 SUMMARY OF RESULTS ... 21 5.2 LIMITATIONS ... 22 5.2.1 Multicollinearity ... 22
5.2.2 Endogeneity ... 23
5.2.3 Misinterpretation ... 24
5.3 FURTHER RESEARCH ... 24 CONCLUSIONS ... 26 REFERENCES ... 28 APPENDIX ... 30 Regression 1: ... 30
Regression 2: ... 30
Regression 3: ... 31
Table 2: ... 31
Table 3: ... 31
Regression 4: ... 32
Introduction
Ethical investment was the terminology used by investors before the broader umbrella term social responsible investment was introduced. Ethical investment found its roots in churches, where there was a wider concern about alcohol, tobacco and gambling. During the apartheid era, ethical practices began to include human rights. Ethical investment is the practice of selecting portfolios, mainly consisting of firm’s shares, based upon ethical and social concerns (Sparkes and Cowton, 2004). Nowadays, social responsible investment is part of government-controlled funds, mutual funds, exchange-traded funds, community investment, and shareholder advocacy. Mutual funds and ETFs are weighted upon environmental, social and corporate governance (ESG) criteria. In recent years, firms who engage in socially responsible investments are becoming increasingly successful in the adaption to deliver financial benefits and in the implementation of environmental, social, and governance criteria (Dimson et al., 2015). Social investors try to influence SRI by using several strategies to maximize financial returns and social good. These strategies include divestment, shareholder activism, negative screening, positive investing and community investment (Sparkes and Cowton, 2004).
Shareholder advocacy is investors’ role to provide a positive influence for social resolutions (Schueth, 2003). Shareholder activism is using shareholders’ influence to have a positive effect upon firm’s behavior. Institutional investors and firm’s executives engage in discussions about corporate social responsibility issues. Investors’ engagement is executed over a long period of time to find a low-profile solution for social responsible issues.
Shareholders have the rights to attend corporations’ meetings to discuss firm’s activities. The goal of shareholder engagement is to improve immoral activities of a firm. During those discussions, shareholders can express their financial, social and ethical concerns. Shareholder activism can effectively change a firm’s behavior taking into account those concerns (Sparkes and Cowton, 2004). For example, shareholder activism can lead to an improvement of the wellbeing of stakeholders. The financial benefits may increase as a consequence of better circumstances for employees, because they are willing to work harder (Schueth, 2003). Firms can also use an increase of financial returns to improve the safety and wellbeing of employees (Nelling & Webb, 2009).
With divestment, institutional investors select ethical funds upon the absence of immoral actions and in alliance with the wellbeing of shareholders. Morals are bases upon personal decisions of wrong and right. Ethics are practical conducts of shared morals in companies and communities. This absence strategy provides a threshold of moral actions. Moral purity is not achievable, because ethical funds within a portfolio are weighted upon a percentage level of moral responsibility. In this way, some immoral activities are still accepted. Large companies are capable of doing more immoral actions compared to smaller ones focusing on this percentage rate of moral responsibility. Investors can overstate their moral actions, and ignore other social practices. Divestment is the oldest and most dominant strategy for ethical funds (Sparkes and Cowton, 2004). More active ethical practices, such as donations, employment of ethical minorities and investing in disadvantaged communities, will provide a stronger social image of companies. Community investment is a part of social responsible investment. However, measurement to assess positive impacts within
communities is not yet available (Sparkes and Cowton, 2004). Sparkes & Cowton (2004) states consequently that, “there is relatively little agreement on what such positive issues should be”.
Another active approach is positive investing. Companies try to gain a green image by having a positive social impact. Investors can engage in activities whereby they choose companies that take into account social justice and sustainability. Selection makes it possible to invest in green image companies without sacrificing portfolio diversification or long-term performance. Especially in the environmental area, companies would gain customers’ interest from individuals that prioritize sustainability. In this case social responsibility and financial benefits are met. However, the environmental technology sector that provides dual benefits is small (Sparkes and Cowton, 2004).
Subsequently, avoidance approaches lead to a decrease in diversification by only searching for the best moral attributes to add to the investment portfolio. This lowers opportunities for risk sharing. Lower risk-adjusted returns do not provide further grow opportunities. In theory, SRI is expected to lead to lower financial returns. Evidence provides an ambivalent view on the link between social responsible investment and financial returns. Pension funds, as part of government-controlled funds, approach financial cost by offsetting risk to return in a portfolio. Ethical shares are selected upon similar economic characteristics as immoral shares. This enables investors to provide portfolios without certain immoral market sections of the stock market. However, those portfolios still have similar financial performance as portfolios that didn’t take into account ethical shares. In the U.S., the Domini Social Index uses this avoidance approach to minimize financial risks and costs. The
risk-optimization approach is associated with screening, whereby investments are selected upon SRI criteria and financial returns (Sparkes and Cowton, 2004). The practice of investors to search for profitable companies engaged in the improvement of society is defined as screening (Schueth, 2003).
A central debate from the academic financial community still rages to what extent SRI affects firm performance. The SRI market is growing rapidly. In the United States one-ninth of the total assets under management are held in SRI. The firms that are incorporating social responsibility rose with 33% since 2014 (Riedl & Smeets, 2017). According to the US SIF 2014 rapport, the SRI market contains 6.6 trillion USD in assets under management (AUM) (Henke, 2016).
1.1 Problem statement
Social responsible investing deviate the financial market regarding investments on high scores on return, or high scores on ESG criteria (Riedl & Smeets, 2017). Fundamental evidence significantly shows that socially responsible investment and performance are related. Empiric research shows that the results about the effect are ambivalent (Aktas, 2011). On one side of this debate are those who argue that high investment in socially responsible funds affects the firm’s performance positively (Kempf & Osthoff, 2007). Those on the other side, however, suggest that investing in socially responsible funds reduces the financial performance of the firm (Fisher-Vanden & Thorburn, 2011).
The standpoint of shareholders on SRI and their effect on firm performance is changing throughout the years, and the negative point of view is weakening. The perspective of a more positive effect has come into play. Our quantitative analysis adds time-trend to the effect of SRI on firm’s performance (Dimson, 2015). The effect is changing due to external time circumstances, e.g. a financial crisis. Nevertheless, time trend is added because literature reviews suggest that the popularity of SRI has shown a rapid increase. Divestment leads to a decrease of diversifying portfolios, because certain sectors are excluded from the portfolios (Sparkes & Cowton, 2004). Diversification can also increase by adding social responsible mutual funds to corporations’ portfolios. Prior research shows that diversifying the firm’s portfolio with stocks that have a high score on socially responsible ratings lead to abnormal high returns of up to 8.7% per year (Kempf &Osthoff, 2007). Less risk-taking goes
accompanied by a decrease in firm’s performance. During crisis, the firms are exposed to more risks. The risky social responsible investments will have higher payoffs in times of crises, because those investors are prepared to take on more risk (Henke, 2016). Despite the growing data found on the positive effect between SRI, and firm performance, data
limitations about the relation are still open for further research (Dimson, 2015).
1.2 Relevance
This study contributes to the literature by adding the effect of moderators to the link between firm performance and responsible investing. Hereby, moderators include leverage, R&D expense, and time-trend. Other firm characteristics control the effect of social responsible investment on the firm’s performance. The control variables consist of independent variables that have a proportional link with the firm’s performance. Whereby this research takes into account firm’s age, capital structure, sales, and research & development expenses into account. This paper examines the effect on the U.S. public financial sector during the period 1991 until 2013. This paper provides a new insight in the examination of the link between firm performance and social responsible investment.
The research question:
What is the effect of socially responsible investing (SRI), with the six dimensions: environment, product, governance, employee relations, community, and diversity, in the financial sector in U.S. markets on the firm’s financial performance: return on assets?
To provide the answer, sub questions are added to the research question about the effect of social responsible investment on the firm’s performance:
1 What the effect of the firm’s research and development expense on the relation
between socially responsible investments and the return on assets?
2 What is the effect of the firm’s leverage ratio on the relation between socially
responsible investments and the return on assets?
3 What is the effect of the time-trend on the relation between socially responsible
investments and the return on assets?
The paper is structured as follows. First, a literature review about previous research that examined the link between SRI and firm performance is provided. Ambivalent
approaches to the link are taken into account. Certain important variables are outlined. Second, methodology of this research is discussed. Third, results of the OLS regressions are outlined. Fourth, limitations and implications of our research to the correlation between SRI and firm performance is explored. Some ideas for future research are mentioned. Fifth, the conclusion about this research is provided.
2. Literature review
In this chapter, relevant information is provided to answer the main and sub questions. Hypotheses are based upon historical literature, and are given chronologically with explanation added.
2.1 Social responsible investing
Social responsible investment (SRI) has its roots in the 1960s and has matured throughout the years. Firms are required to comply with social and environmental criteria before investors include them in the portfolio. Social responsible investment is used interchangeably with the definitions green investment, social investment, ethical investment, and values-based investment. For most, SRI is about improving the quality of life while still gaining financial benefits. Nowadays, the social firms listed in the portfolios provide more information (Schueth, 2003). Firms have an incentive to be socially responsible due to improvement of their brand image and reputation improvements. Consumers drive SRI-growth, and
consultants see the potential. The increase in a firm’s reputation leads to an increase in sales, which ultimately leads to an improvement in profits (Adam & Shavit, 2008). The majority of SR investors are women, and their attendance to the workforce is growing in the last decade. Another important reason for the growth in ethical investment is the growing evidence of a positive link between social responsible investment and firm’s performance provided by academic studies and real-world cases (Schueth, 2003).
2.2 Correlation SRI and firm performance
What is the effect of socially responsible investing (SRI), with the six dimensions: environment, product, governance, employee relations, community, and diversity, in the financial sector in U.S. markets on the firm’s financial performance: return on assets?
SRI can be calculated as a vector of strengths minus concerns based on six dimensions: community, corporate governance, diversity, environment, employee relations and product. Those scores give social investors criteria with which they can commit to environmental, social and corporate governance (ESG) (Galema, et al., 2008). According to one viewpoint of Friedman (1970), a negative relationship exists between corporate social responsibility and financial performance. Friedman argues that a positive effect on firm performance is accompanied by the engagement in activities that increase profits and wealth for owners.
Other alternative uses of the allocation of resources will not maximize firm performance. This negative correlation was examined in prior research papers by the effect of SRI on excess stock returns, stock price changes, or changes in earnings-per- shares (Kang et al., 2010). Empirical work does not suggest a trade-off between SRI and expected returns, nor does it suggest a price mechanism of socially responsible stocks by capital markets. Theoretical studies focus on differences in stock prices due to excess demand for socially responsible stocks. The excess demand will lead to overpricing stocks. In a neo-classical equilibrium model, we expect that overpriced SRI stocks will have lower returns and lower book-to-market ratios. The excess demand activates bigger risk sharing opportunities for people who invest socially responsibly. According to the risk-return tradeoff, the decrease of risks and uncertainty will lead to lower financial returns (Galema et al., 2008). Less social responsible companies are expected to generate more risk, because investors may expect a decrease in the value of companies that have less stable relationships with their community. For example, the government may pursuit lawsuits against a social irresponsible firm. This may threaten the company’s existence. Consequently, firms that are highly socially responsible have low levels of financial risk and may be less affected by external circumstances, such as government activities. With lower levels of risk, lower levels of financial returns are expected due to risk-return trade-off (Cornell & Shapiro, 1987).
Another group of investors argue that socially responsible investment has no specific effect on financial performance (Kang et al., 2010). According to Nelling and Webb (2009), the firm’s performance is not affected by corporate social responsible activities. The
feedback-effect of return on assets on SRI is also examined. Their research findings suggest that SRI is not significantly influenced by financial performance. Better financial
performance was expected to lead to more SRI, because of the possibility to spend returns on social responsible activities. With Tobit models, an insignificant feedback loop was found between return on assets and the sections community, diversity, and environment. However, there was a significant effect of performance upon employee relations. Firms can use their financial gains to benefit the employee safety, which leads to the causal relationship between the two. Conclusively, there was only evidence to have a causality running between the firm’s stock performance and the employee relations. Moreover, Granger causality models also do not result in an improvement of the lagged return on assets and stock performance due to an increase in the social responsible investments. If SRI provides firm benefits, they do not manifest in financial benefits (Nelling & Webb, 2009).
Nevertheless, various literature has shown that SRI can lead to higher value for stakeholders and ultimately to an increase in financial profits. Another approach of Friedman
(1970) is based on stakeholders’ theory, a theory that focuses on the creation of firm value for all stakeholders. Investors argue that SRI has a positive effect on financial performance. Firms take multiple stakeholders into account, including customers, employees, suppliers and investors. Hereby, social responsible activities can improve firm value by the improvement of firm reputation. When all stakeholders are satisfied, the firm will ultimately gain financial returns, e.g. employees work more effective; customers are willing to buy more goods and services and sales are improved; future governmental restrictions are avoided and suppliers deliver their products on time. There is a statistically significant, positive relation between financial performance and social responsibility (Kang et al., 2010). The improvement of agency problems as a consequence of strong shareholders rights lead to better financial performance. According to shareholder activism, shareholders can positively influence corporations’ behavior by discussing ethical and financial issues. Better social practices, in the case of large companies, lead to an upward movement of long-run stock performance (Nelling & Webb, 2009). Kempf and Osthoff (2007) found an increase of 8.7% in yearly returns when using screening to select their portfolio. SRI is expected to lead to higher financial returns, because satisfaction of stakeholders will generate firm value (Kang et al., 2010).
Hypothesis 1: Social responsible investment has a slightly positive effect on the financial performance of U.S. public firms.
2.3 Previous research on the variables
Many academic papers that study the link of SRI and financial performance are based on a comparison between ethical mutual funds and conventional funds performance. In those papers, the Fama-French time-series regressions or Fama-Macbeth cross-sectional regressions are used (Bauer, et al., 2005). Fama-Macbeth regressions underestimate standard errors when the dependent variable is correlated across time (Galema, et al., 2008). Performance ratios, such as book-to-market ratios, are extremely correlated in cross-section and across time (Galema, et al., 2008). The use of pooled linear regression models with robust standard errors is more suited to study the link between SRI and firm performance (Kang et al., 2010). Standard errors are clustered by firm and by time (Galema, et al., 2008). Ordinary least-squares (OLS) regressions are common in the search for initial evidence between the causal relationship between SRI and financial performance. Panel data is used to study observations of many firms throughout time periods. Many observations about each firm are examined to study the correlation between SRI and firm performance within each firm. Endogeneity is deleted from the sample with the use of fixed effects and controls. Omitted variable bias, simultaneity and measurement error are causes for endogeneity. There may exist a direct
relationship between social responsible investment and firm performance. Simultaneity bias is excluded by focusing independently on each effect per firm. Correlation between SRI and firm performance depends on the usage of standard OLS regressions or time series fixed effects. In the second case, the effect is more relatable and prior literature shows that this effect is much weaker (Nelling & Webb, 2009).
Measurements for firm performance can be based on accounting tools or stock-market returns. Both measurements focus on different aspects of firm performance (Nelling & Webb, 2009). Accounting based measurements focus on historical returns. Stock-market-based returns are adjusted for expected future earnings due to investors’ evaluations of firm performance. However, investors may have difficulties to consider all constituencies when evaluating future firm value. Accounting-based returns are operating income growth, return on assets, sales growth, assets growth, and total assets (Cornell & Shapiro, 1987). Research papers show that financial performance can be measured on a short-term as well as a long-term basis. The short-long-term measurement uses accounting-based tools, such as return on assets (ROA) or return on equity (ROE), to predict the profitability. On the long-term basis, firm value is measured by stock-market-based returns, such as Tobin’s Q and the price-earnings ratio (Kang et al., 2010).
Moreover, SRI does not affect systematic risks because the market share of ethical funds is still small compared to other conventional funds. Cornell & Shapiro (1987) state that the impact of SR is minimal in systematically affecting all other corporations in the market. The Fama-French model is an extension of CAPM by adding size risk and value risk to market risk. In this way, the model leads to a generalization of CAPM to focus on systematic risk completely. However, no researches discuss big impacts upon systematic risk as a consequence of implementing SRI. The market capitalization of those firms is still too small to have an applicable impact. In CAPM idiosyncratic risk is diversified, but opponents of this model consider the impact of firm-specific risk and firm financial returns. For example, firms that highly invest in sustainability may lead to a green label that leads to excess demand of customers and higher sales. Higher sales will ultimately increase firm performance. The most appropriate measurement to focus upon the impact of increased sales is the operating ROA. Hereby, cost of goods sold is subtracted from the total number of revenues (EBITHA) divided by total assets. In general, accounting-based measurements tools, in particular ROA, proved more suitable measurements to study the effect between SRI and ROA. Accounting
measurements of return are more sensitive to variances in firm-specific performance. Firm performance is expected to be more influenced by SRI when taking into account past
reputation can be increased. Higher past returns and lower risk levels may make it affordable for companies to invest social responsibly (Cornell & Shapiro, 1987).
2.3.1 Age
Firm characteristics control for the effect of social responsible investment on financial performance. Control variables have a proportional effect upon financial returns. Also, the model is effective in predicting the dependent variable. Firms are mainly concerned about SRI-based engagements if they are large, mature, and showing poor performance (Dimson et al., 2015). Older firms can generate higher returns based on their expertise in the market. However, many younger startups show a higher increase in firm’s performance. This can be a result of startups’ flexibility (Dimson et al., 2015). Logarithm of age is an important variable to influence the profitability of a company and is added as control variable (Galema et al., 2008).
2.3.2 Research & Development
A less common control variable for return on assets is research & development expenses (Galema et al., 2008). R&D intensity is a strategic variable that highly correlates with corporate social responsibility. Firms that are using a differentiation strategy want to develop unique products and services that will attract customers. SRI can attract customers by using unique products and services that are socially responsible and sustainable. R&D expenses are highly correlated with social responsible expenses. Most firms that incorporate SR are also actively investing in R&D. Models that claim to explain financial performance correlated with SRI should include important strategic variables, such as R&D expenses. McWilliams and Siegel (2000) claim that R&D has a proportional impact on profitability, and excluding important predictors leads to misspecification of the model. The results are therefore not reliable and may lead to biased approximations (McWilliams & Siegel, 2000).
What the effect of the firm’s research and development expense on the relation between socially responsible investments and the return on assets?
Firm’s characteristics are not only considered as control variables, but also as interactions with SRI and their effect upon financial performance. These variables are called moderators, or in financial terms interaction variables (Nelling & Webb, 2009). The interacting variable, R&D expense, is an enlargement to prior research. R&D expenses may be necessary for ESG improvements (Dimson et al., 2015). Most firms that invest highly in research and
a brand label of social responsibility, they attract more customers and generate higher revenues. Social enterprises, such as Tony’s Chocolonely, actively innovate. SRI is highly correlated with research and development (McWilliams & Siegel, 2000). Interacting variables R&D and SRI have a positive, significant effect on firm performance (Hull & Rothenberg, 2008).
Hypothesis 2: High R&D expense will have a positive effect on the relation between SRI & performance.
2.3.3 Size
In their models, Kang, Lee and Huh (2010) included additional size, and leverage as control variables that also correlate with the financial performance measurements. The different sizes of firms indicate that small and large firms differ in financial performance. Smaller firms may be riskier due to information asymmetry. Lack of information provided to investors results in higher risks. Firm’s debt level increases to cover higher risks by holding on to more money to be prepared for implicit claims. This results in cost of capital increases in smaller firms, and a delay before firm value is provided. Large firms may provide more information to investors, a decrease in cost of capital and ultimately an increase in firm value. Furthermore, larger firms may be more profitable because of economies of scale. The size can be measured as log of sales, or with log of assets. Those two control variables are correlated with financial returns. However, total assets are used as denominator in ROA, so when log of assets is used endogeneity should be excluded from the model. This can be provided by regressions with fixed effects (Kang et al., 2010).
2.3.4 Leverage
Another control variable, leverage, identifies the stimulus of a firm to make use of its capital structure. Operating leverage measured by the debt to assets ratio, is an accounting-based measurement of risk. Firms that actively invest in a social responsible manner generate lower levels of risk, because of better relationships with stakeholders. In this manner, those firms have low level of debt to assets (Cornell & Shapiro, 1987). Furthermore, the market defines firms with excessive debt as too risky. One positive aspect of debt is that firms can subtract tax of interest expenses while dividends are not tax-deductible. A firm may take on more debt to make use of the tax-deductibility (Kang et al., 2010). Mostly, leverage is assumed to have a negative correlation with future growth. Firms with high leverage ratios suggest bad
investment opportunities, and a negative correlation with firm growth. This negative relationship is consistent with the agency costs of debt. Shareholders may have different
interests containing debt and equity (Lang et al., 1996).
What is the effect of the firm’s leverage ratio on the relation between socially responsible investments and the return on assets?
Operating leverage is added as interaction variable to cover the complication to invest highly in SRI when there are high debt levels. With the increase of firm’s leverage, the investors have less cash available, and this results in the deviation from value maximization and a decrease of the effect of SRI on firm’s performance (Barnea, 2010). Firms that are less financially constrained hold larger cash holdings and lower leverage ratios (Dimson et al., 2015). High SRI firms generate lower levels of risk, and lower leverage ratios (Cornell & Shapiro, 1987). Leverage makes a negative impact upon the correlation between social investing and firm performance. The interacting variable of SRI and leverage is negative (Kang et al., 2010). Dimson, Karakas, and Li (2015) observe that leverage has a negative effect in obtaining financial success.
Hypothesis 3: High leverage ratios will have a negative effect on the relation between SRI & performance.
2.3.5 Time trend
What is the effect of the time-trend on the relation between socially responsible investments and the return on assets?
The risk-mitigation view of SRI consists of the idea that investors that take on bonds with high social ratings are exposed to less risks, because they examine other firm risks as well. While they screen to identify SRI risk, they also pay attention to other risks. During a crisis, investors are exposed to more risks. With screening for risky investments, investors prepare themselves for the exposure of risks in other situations. In this way, portfolios with socially responsible bonds outperform other portfolios. SRI results in a downside risk-reduction effect and an outperformance of 0.65%-0.70% than other investments during bear markets or economic recessions. The correlation between SRI and financial performance is influenced by time, and this is considered as an interaction between time trend and SRI (Henke, 2016).
The effect of SR investment on financial performance is influenced over time. When social responsibility was first introduced, this period was considered as more risky than the years that follow. The popularity of SRI has shown a rapid increase, which evolved in less risk-taking for investors. Screening made it possible for investors to focus on riskiness. Social responsible investments were able to outperform financially in times of high risks (Henke,
2016). Social responsible funds are selected upon financial returns, and include those funds that pay higher financial returns in more risky periods, such as time of introduction and crises (Kempf &Osthoff, 2007). The interaction between time trend and SRI on return on assets is changing negatively due to external time circumstances, except during risky periods. In times of crises, they are outperforming financially and there a positive correlation exists (Henke, 2016).
Hypothesis 4: Time trend will have a negative effect on the relation between SRI & performance.
2.3 Conceptual model Figure 1: Conceptual model.
3 Methodology 3.1 Data sources
Data is obtained from the U.S. market, because this market has a high level of transparency and data availability (Henke, 2016). The data used, is found on the Wharton WRDS database. This data platform has access to different data sections, whereby CRSP, Compustat – Capital IQ, and MSCI (formerly KLD and GMI) are very commonly used. The latter provides us information about social ratings. The ratings are based upon strengths and weaknesses in the sections community, corporate governance, diversity, employee relations, environment, human rights, product, alcohol, firearms, gambling, military, nuclear power and tobacco. Strengths and weaknesses are measured from zero till two, and are obtained in the period 1991 until 2013. However, human rights have a timeline till 2000, and are excluded from the regression. Furthermore, screens that focus consistently on concerns are left outside analyses. Those sections include alcohol, firearms, gambling, military, nuclear power and tobacco. Strengths and weaknesses that will generate a firm’s social ratings are the six dimensions: community, corporate governance, diversity, employee relations, environment, and product. These sections stand for social environment, diversification on the workforce, relationships of corporations with their employees, frameworks of firm governance, management of
environmental issues, and quality of products. Screens are measured at the end of the year (Galema et al., 2008). By downloading the entire database of records of these sections from January 1991 till December 2013, ticker and CUSIP codes are downloaded as well. The accounting data is found on the Compustat database beneath the section North America Daily - Fundamentals Annual. Accounting data consists of ipodate, total assets, total long-term debt, net sales, net income, research and development expense, ticker symbols, and CUSIP codes.
3.2 Regressions: data analysis
𝑅𝑂𝐴! = 𝛽!+ 𝛽!𝑆𝑅𝐼!+ 𝛽!𝑋!+ 𝜀!
𝑅𝑂𝐴! = 𝛽!+ 𝛽!𝑆𝑅𝐼!+ 𝛽!𝑆𝐼𝑍𝐸!+ 𝛽!𝐿𝐸𝑉!+ 𝛽!𝐴𝐺𝐸!+ 𝛽!𝑅𝐷!+ !
Vector 𝑋! includes several control variables known to correlate with return on assets. Control variables consist of leverage, age, sales, assets, and R&D ratio. R&D ratio is R&D expense divided by total assets. Leverage ratio is provided by debt-to-assets ratio. The dependent variable, firm’s financial performance, is return on assets, which is the firm’s operating income divided by total assets. SRI may lead to increased sales, equivalent to higher EBITDA. Increased debt correlates with firm’s performance, where too much debt may be
dangerous for a firm. However, tax deductibility of debt may be profitable. Different sizes of firms are correlated with financial performance, and consist of the control factor log of sales. Smaller firms have on average higher expected returns than larger firms, and they take on more risk. Larger firms have the advantage of economies of scale, and less information asymmetry. Furthermore, firm performance is influenced by firm age. Whereby, control variable log of age is used to express that older firms have gained experience, but young start-ups are flexible. The control variable R&D-to-assets ratio is used to indicate that firm’s performance is related to the amount firms spend on innovation to search for new ideas, and technologies. After downloading the entire database of the accounting data, the two STATA-files are merged. Afterwards, the newly formed dataset contains variables that have no impact and missing values can be left out of STATA-file. The remaining merged data is left to form needed ratios. Next, ratios are winsorized by 1% to limit the influence of outliers. Another method to minimize data from an influence of outliers is provided by robust regressions.
The next step is fitting linear fixed- and random-effects in STATA by using panel variable. The model will be tested with panel data. First of all, xtset and xtreg are used to provide regressions with one level of fixed effects. This makes it difficult to cluster per firm and per year. Firms-level clustering starts with changing ticket codes to new ascending id numbers. Clustering per firm and per year is achieved in xtreg analyses by putting the two effects together in one variable. Regdfe is a generalization of xtreg, whereby regressions can be made with two or more fixed effects. Step two is making use of reghdfe fixed effects panel data. Two-way clustering is based upon firm id and year. Reghdfe drops singleton groups, which have effects on the variance–covariance matrix (VCE). Keeping those observations leads to an overestimation of significance levels of coefficients within the model and an underestimation of the standard errors.
Ultimately, interacting variables are also tested upon statistical significance by fixed effect regressions with a new regression analysis. The new regression analysis changes prior controls R&D ratio and leverage to interacting variables with SRI. This regression includes time trend as interaction with SRI, and will be tested with regdfe. Controls in the model are log of age and log of sales.
𝑅𝑂𝐴! = 𝛼!+ 𝛼!𝑆𝑅𝐼!+ 𝛼!𝑆𝑅𝐼!∙ 𝑅𝐷!+ 𝛼!𝑆𝑅𝐼!∙ 𝐿𝐸𝑉!+ 𝛼!𝑆𝑅𝐼!∙ 𝑇𝑅𝐸𝑁𝐷!+ 𝛼!𝑆𝐼𝑍𝐸! + 𝛼!𝐴𝐺𝐸!+ 𝜀!
4 Results
In this chapter, results between SRI and ROA are described. Moreover, interacting variables are discussed step by step. STATA-files of regressions are reported in the appendix. Based upon STATA-findings, results can be compared to the four hypotheses.
4.1 Descriptive statistics Variables (1) OLS model (2) Panel data, 1-level fixed effects (xtreg) (3) Panel data, 2-levels fixed effects (reghdfe) Variables (4) Panel data with interaction variables, 2-levels fixed effects (regdfe) ROA ROA SRI -0.0027701*** (0.000) -0.0023633** (0.002) -0.0003262 (0.474) SRI -0.0106152*** (0.000) 𝐿𝐸𝑉 - 0.11781*** (0.000) -0.121219*** (0.000) -0.1107797*** (0.000) SRI ∗ LEV 0.0001286*** (0.000) 𝑅𝐷 -0.784738*** (0.000) -0.7962927*** (0.000) -0.7603809*** (0.000) SRI ∗ RD 0.0616474*** (0.000) LOGSALES 0.0430996*** (0.000) 0.0446171*** (0.000) 0.0753293*** (0.000) LOG- SALES 0.0836008*** (0.000) LOGAGE 0.0052091* (0.020) 0.0027029 (0.335) -0.053239*** (0.000) LOGAGE -0.0678323*** (0.000) SRI ∗ TREND 0.0001286 (0.206) Observations 9,321 9,321 9,276 9,276 Adj. R-squared 0.6051 - 0.8900 0.8571 Figure 2: Results
4.1.1 SRI & Firm performance
It is interesting to see that correlation between SRI and ROA remains negative in all our regressions. The results show, overall, that the coefficient of SRI is small. In OLS-regression, there exists a highly significant coefficient of -0.0028. The robust statistical OLS-regression shows an R-squared level of 0.6051. All control variables are significant. Fixed-effect models are expected to improve measurability of the correlation between SRI and ROA. Panel models based on 1-level (xtreg) and 2-levels fixed effects (reghdfe) are used. The latter is expected to lead to more reliable measurements with interpreting cross-sectional and time-series effects separately. Also, the drop of singleton groups makes two-way clustering more generalized. Model 2 with 1-level fixed effects shows a significant coefficient of -0.0024, which is close to the effect found in OLS-regression. Only the control variable log of age is insignificant within the model. As expected model 3 improves R-squared compared to an OLS-regression. The R-squared level becomes 0.8900. All control variables are significant in this model. However, the effect of SRI on ROA is insignificant. This may indicate that there is no effect of SRI upon ROA.
Interaction variables of SRI with leverage, R&D ratio and time trend are added in a 2-level fixed effects model. Leverage and R&D ratio are now expected to be interacting
variables with SRI instead of control variables. This model shows a high significant
correlation of -0.011 between SRI and ROA. The control variables log of age and log of sales are significant. R-squared level of this model is 0.8571. That may indicate that the model has a high level of fit. Further details about interacting variables will be given in the next section.
4.1.2 Research & Development
Having a closer look, interaction effects of R&D with SRI can be added to show, with significance, an effect of 0.062 upon firm performance ratio. Firm’s return on assets is positively influenced by SR and R&D investments at the same time. SRI and innovation are correlated in firms across time, and can improve financial performance.
4.1.3 Leverage
Interacting effects between leverage and SRI are significantly correlated with ROA. With an effect of 0.0001286, leverage and SRI do no have an important effect together upon return on assets. Leverage ratios and SRI are correlated with each other. Increasing SRI is correlated with firm’s leverage ratio and will result in an upward movement of ROA. When more investments are made, debt may increase and as an effect profitability of investments may pay off in higher return on assets. Even though, this effect makes such a small impact that it may be negligible.
4.1.4 Time trend
Interactions between SRI and time trend have an insignificant effect upon ROA. The effect is positive, but this coefficient is not reliable. It may indicate that firms have no time-bound impact on financial returns. In a specific year, firms may be checked upon SRI, and they would generate comparable ROA as in other years from 1991 till 2013. Time trend interacting with SRI will not lead to changes in financial returns. Conclusively, SRI has not a changing effect on ROA throughout the years.
5 Discussion
5.1 Summary of results
As expected results show an improvement of R-squared levels between fixed-effects and OLS-regressions. This may indicate that cross-sectional and time-series dimensions improve the reliability of the model. In other papers, results of the correlation between SRI and ROA differentiate in OLS regressions and fixed effects models. Prior literature shows that in the latter models, the correlation is much weaker between the two (Nelling & Webb, 2009). The ability of social signaling to search for profitable and social responsibility at the same time, followed up with the conclusion of a slightly positive correlation. Social signaling and safe relationships within the community, were leading to high returns of SRI. The results contradict this view, and show a negative effect of SRI on ROA, or even an insignificant correlation. The latter correlation may indicate that SRI may not lead to financial benefits. SRI was expected to result in more sales, because SRI would improve firm reputation. This may lead to more customers that are interested in social responsible products, and services. However, environmental technology sector that provides dual benefits may be a niche market, and may not lead to huge impacts to sales after SRI involvement.
Misinterpretation about the riskiness of SRI can indicate why a positive effect was expected between SRI and ROA. SRI was expected to be less risky after social screening. However, investors may still associate SRI with financial costs. Especially, pro-active SRI, such as community building, can be considered more as costs than benefits. Also, another SRI approach, called disinvesting, lowers the opportunities of diversification and risk-sharing. Disinvestment disregards certain market sectors. Some SRI strategies may indicate high risks. However, screening was expected to lead to lower levels of risks. If this was the case, results found in OLS and 1-level fixed effects regressions would indicate a risk-return payoff. However, it can be argued that SRI has not a significant effect on ROA. Fixed effect model 3, which may be the most reliable model, shows no causal relation of SRI on ROA. SRI may not result in financial benefits.
Time trend has not a specific effect on SRI that may lead to an improvement of ROA. R&D expense is, as expected, associated with more SRI. The correlation will lead to positive improvements of return on assets. Interaction between leverage and SRI has a small positive impact on ROA. When SRI may be considered as financial cost, the linkage between leverage and SRI can be misinterpreted. In theory, interaction of leverage and SRI was expected to
have negative effect on ROA. Firms with high leverage ratios suggest bad investment opportunities, and a negative correlation with firm growth.
5.2 Limitations
Furthermore, this research is still incomplete in certain areas. Multicollinearity, endogeneity, and misinterpretations will be discussed here. Based on those limitations, other research ideas are outlined to give insights for future research.
5.2.1 Multicollinearity
Multicollinearity exists when independent variables are correlated. Normally, in linear regressions the coefficient is the change in dependent variable as an effect of 1-unit change in independent variable, while other independent variables are held constant. However, the coefficient is unequal to this change when multicollinearity exists within the model. Multicollinearity may result in opposite effects of an independent variable on dependent variable. Logical thinking is not in line with what the model shows. No conclusions can be made based on coefficients in multicollinearity models. Collinearity can be tested with VIF of all independent variables. Each VIF-number should be beneath the rule of tumb VIF<5. If no collinearity between independent variables exists within the model, VIF should be equal to one.
In correlation table (table 3), log of assets and log of sales are highly correlated. They are both measurements of firm size, as discussed in literature review. Moreover, only one control variable of firm size is added in the model to exclude the possibility of
multicollinearity. Log of sales is this control variable. However, R&D ratio and leverage ratio consist of total assets as numerators. Total assets may be correlated with log of sales. This can indicate the high correlation of -0.5526 between R&D ratio and log of sales (table 3). The correlation between log of sales and leverage is not an indication for multicollinearity with a level of 0.2231 (table 3). Moreover, multicollinearity is most likely present in a model when F-test values are high and p-values are high as well. In our model, p-values of leverage, R&D ratio, and log of sales are highly significance (p-value<1%). It can be expected that there exists no multicollinearity in our model. When multicollinearity is present interpretations of coefficients within the model are not reliable. When VIF-tests are made, correlated
independent variables with high VIF-scores that threaten the existence of the model can be found. One solution of multicollinearity may be to eliminate independent variables with high VIF scores in regressions. Another solution may be to neglect making conclusions based on coefficients in a multicollinearity model.
5.2.2 Endogeneity
Endogeneity is an umbrella term for biases that indicate correlation between variables in the model and error term. It is a common problem in regressions. Two important biases are omitted variables, and measurement error. Those are outlined in the following two sections. Panel data minimizes the impact of omitted variables, and measurement errors. By making use of clustering, panel data models assume that ROA for the same firm is correlated over time, but it is independent across firms. Cross-sectional and time-series dimensions control impacts of omitted variables by fixed effects (Hsiao, 2007).
5.2.2.1 Omitted variables
Higher past returns may make it more accessible for firms to spend money on SRI. This indicates correlation between dependent and independent variable. Omitted variable bias indicates that an important variable is not included into the model. The omitted variable is correlated with an independent variable and dependent variable. Conclusions can be interpreted wrongly from models that eliminate important variables. Investors may be expected to be poor Bayesian practitioners, and do not choose wisely. Successfulness in the past, may lead to investors overreacting to current events. In their choices, they may deviate from real underlying values. An increase in return on assets, may lead to an overreaction of investors to those investments. Bondt & Thaler (1987) show overreaction of investors may lead to an outperformance of unsuccessful portfolios compared to successful portfolios by 31.9% (Bondt & Thaler, 1987). Investors’ perception of SRI may be extremely important. Intangibles, such as culture, human capital, reputation, and innovation, may have indirect effects between SRI and firm performance. It may not be measurable, but could lead indirect effects of ROA on SRI (Surroca et al., 2010).
5.2.2.2 Measurement error
The vector of socially responsible investment depends on many factors and is measurable in many ways. The qualitative aspect to screen for social conscious firms is considered more complicated to measure than the quantitative analysis of profitability. The screening decision of investors is based upon this double bottom line analysis. The research provided to screen companies is ambivalent, because the investors’ social criteria and returns must be met (Schueth, 2003). The six dimensions may be an incomplete measure. Return on assets is considered to measure firm’s performance, but the dependent variable can also take into account stock-performance measurements, such as Tobin’s Q or market-to-book ratio.
5.2.3 Misinterpretation
Prior academic papers show evidence of negative, neutral and positive correlations between SRI and ROA. It may be difficult to form hypotheses based upon biased papers.
Another error in our approach to measure the link between SRI and ROA depends on interaction effects. Our theoretical research showed that leverage, R&D and time trend has positive as well as negative effects upon the relationship. Positive subjects may outweigh negative characteristics, or the other way around. We expected that the effect of SRI upon performance would increase during a crisis. Theoretical statements could be made that the effect may increase throughout the years, due to effectiveness of implementing SRI. To assess whether moderator effects are leading to a decrease or increase in our sample of U.S. firms, may be inconsistent.
5.3 Further research
Future research may eliminate endogeneity with two-least squares regressions. In this case, there is no assumption of one-way effects of SRI upon ROA, but feedback effects are tested as well. This means that an increase of firm performance may lead to an increase of SRI. In this study, endogeneity was made as small as possible with the use of fixed panel data. Still, instruments (Z-values) that have a correlation with SRI and no effect on ROA, can be included in the model. Before starting 2SLS-regressions, it is important to test the model on endogeneity with STATA-command “estat endog”. When endogenous covariates are found, instrumental-variables regressions can be made to test upon feedback effects. Previous research papers found that SRI can improve intangibles, such as culture, innovation, human capital, and reputation. These intangibles have an indirect effect of SRI on firm performance. However, results also show an opposite effect. These findings may indicate that improved intangibles result in a virtuous circle between SRI and ROA (Surroca et al., 2010).
A difficulty of intangible resources may be their measurability. Culture, human capital, innovation and reputation are hard to measure. R&D expense is assumed to be a measurement of innovation, but can still be considered incomplete. In this study, innovation is included as interaction variable with SRI and their effect on ROA. However, finding an appropriate instrument for a two-least squares regression that is only correlated with the independent variable SRI may be difficult, and is still open for further research. It is important to find instruments that are uncorrelated with intangibles (Surroca et al., 2010).
It may be interesting to study other considerations of investors to hold back in green investments next to firm performance. Riedl & Smeets (2017) found that investors hold back in SRI mutual funds, because of social preferences. Socioeconomic intangibles are open for future research. The measurement of SRI is also topic for further discussion. KLD database provides environmental, social and governance screens up to 2013, which could be improved to more recent data availability. Moreover, SRI is a broad term that can be measured in numerous qualitative ways.
Conclusions
In conclusion, multiple regressions with 9,321 observations of firms in the U.S. market reveal that social responsible investment has negative, or even no effects on financial performance. The hypothesis that SRI will, overall, have a slightly positive effect on ROA is rejected. Results show a slightly negative effect that is even insignificant in the model with the highest level of fit. Control variables show high significance within the models. SRI consists of six dimensions, which are based upon qualitative research. Strategies that implement SRI are divestment, shareholder activism, negative screening, positive investing, and community investment. This broad umbrella term can be measured in multiple ways, and our research is based on the implementation of environmental, social and corporate governance (ESG). This approach is in line with screening, whereby investors search for economic returns and high scores on ESG criteria. This decreases the riskiness of social responsible portfolios.
Misinterpretation can be caused by the many ways SRI and ROA can be measured. Firm performance can also be measured based on stock market or accounting-based returns. It was expected that SRI began to outperform other investments throughout the years due to their successful implementation. The timeline 1991 until 2013 may not be relevant enough to show this shift. It may indicate a more negative correlation between SRI and ROA than nowadays would have been the case. It was highly risky to first invest social responsible, but during times of crisis SRI may have outperformed other investments. Contradictive aspects of correlation between SRI and ROA in the time period 1991 until 2013 may suggest that the effect is not proportionally changing throughout the years, and time trend is insignificant. The hypothesis that time trend will have a negative interaction with SRI on ROA is rejected with insignificance.
When SRI would lead to negative returns, the risk of SRI may increase. This goes accompanied with high levels of debt. High debt levels are negatively correlated with firm performance. Moreover, the hypothesis that leverage will have a negative effect with SRI on ROA is rejected. Instead, high debt levels and high SRI levels are followed with positive effects on firm performance. In theory, this correlation would have been different. The results show that firms with high leverage ratios and high SRI have negative correlation with firm performance.
This research contributes to the literature by including fixed effects models next to common OLS, Fama-French and Mac-Beth regressions. In this study, the ambivalence view between SRI and firm performance is confirmed. Multiple historical researches show neutral, negative, and positive correlations. Theory and practice may not always be in line. The link between SRI and firm performance is still open for future research. It may be important to take into account the effect of firm performance on SRI as well. Instruments can be used for further research. Another reason of improvement compared to other literature, is taking into account innovation. R&D ratio is in all our models highly significant. When R&D was used as an interacting variable with SRI, it showed a high significant positive effect. The
hypothesis that high R&D expense will lead to positive effects on correlation between SRI and ROA is accepted.
Table 1: Results OLS regressions hypotheses
Hypotheses Results
H!: Social responsible investment has a slightly positive effect on the financial performance of U.S. public firms.
Rejected
H!: High R&D expense will have a positive effect on the relation between SRI
& performance.
Accepted
H!: High leverage ratios will have a negative effect on the relation between
SRI & performance.
Rejected
H!: Time trend will have a negative effect on the relation between SRI &
performance.
References
Adam, A. M., & Shavit, T. (2008). How can a ratings-based method for assessing corporate social responsibility (CSR) provide an incentive to firms excluded from socially responsible investment indices to invest in CSR?. Journal of Business Ethics, 82(4), 899-905.
Barnea, A., & Rubin, A. (2010). Corporate social responsibility as a conflict between shareholders. Journal of business ethics, 97(1), 71-86.
Bauer, R., Koedijk, K., & Otten, R. (2005). International evidence on ethical mutual fund performance and investment style. Journal of Banking & Finance, 29(7), 1751-1767.
Bondt, D., Werner, F. M., & Thaler, R. H. (1987). Further evidence on investor overreaction and stock market seasonality. The Journal of finance, 42(3), 557-581.
Dimson, E., Karakaş, O., & Li, X. (2015). Active ownership. The Review of Financial
Studies, 28(12), 3225-3268.
EUROSIF, 2014, European SRI study 2014, available at http://www.eurosif.org. Fisher-Vanden, K., & Thorburn, K. S. (2011). Voluntary corporate environmental initiatives and shareholder wealth. Journal of Environmental Economics and management, 62(3), 430-445.
Henke, H. M. (2016). The effect of social screening on bond mutual fund performance.
Journal of Banking & Finance, 67, 69-84.
Hull, C. E., & Rothenberg, S. (2008). Firm performance: The interactions of corporate social performance with innovation and industry differentiation. Strategic Management Journal,
29(7), 781-789.
Hsiao, C. (2007). Panel data analysis—advantages and challenges. Test, 16(1), 1-22.
Galema, R., Plantinga, A., & Scholtens, B. (2008). The stocks at stake: Return and risk in socially responsible investment. Journal of Banking & Finance, 32(12), 2646-2654.
Kang, K. H., Lee, S., & Huh, C. (2010). Impacts of positive and negative corporate social responsibility activities on company performance in the hospitality industry. International
journal of hospitality management, 29(1), 72-82.
performance. European Financial Management, 13(5), 908-922.
Lang, L., Ofek, E., & Stulz, R. (1996). Leverage, investment, and firm growth. Journal of
financial Economics, 40(1), 3-29.
McGuire, J. B., Sundgren, A., & Schneeweis, T. (1988). Corporate social responsibility and firm financial performance. Academy of management Journal, 31(4), 854-872.
McWilliams, A., & Siegel, D. (2000). Corporate social responsibility and financial
performance: correlation or misspecification?. Strategic management journal, 21(5), 603-609.
Nelling, E., & Webb, E. (2009). Corporate social responsibility and financial performance: the “virtuous circle” revisited. Review of Quantitative Finance and Accounting, 32(2), 197-209.
Riedl, A., & Smeets, P. (2017). Why do investors hold socially responsible mutual funds?.
The Journal of Finance.
Schueth, S. (2003). Socially responsible investing in the United States. Journal of business
ethics, 43(3), 189-194.
Social Investment Forum (SIF), 2014, Report on socially responsible investing trends in the United States, Available at http://www.socialinvest.org.
Sparkes, R., & Cowton, C. J. (2004). The maturing of socially responsible investment: A review of the developing link with corporate social responsibility. Journal of Business Ethics,
52(1), 45-57.
Surroca, J., Tribó, J. A., & Waddock, S. (2010). Corporate responsibility and financial
Appendix
Regression 1:
Regression 3:
Table 2:
