Sustainability and stock performance after the financial crisis

1

University of Amsterdam Faculty Economics and Business

BSc Economics & Business Specialization Economics and Finance

Sustainability and stock performance after the financial crisis

Author: D.G. Adel

Student number: 10354476

Thesis supervisor: Mr. M. Droës

2

Statement of Originality

This document is written by student Dior Adel who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of the completion of the work, not for the content.

3

Abstract

There have been several studies done to find out whether sociable responsible companies could outperform the market or not. The results of these studies have been very diverse. This study conducts the same research question. We examine whether social responsible companies have outperformed the market by constructing four portfolios of each having 15 companies and one other portfolio taken the other four portfolios together and act as a social responsible market proxy. The companies have been chosen using the program ASSET4, which allowed us to find the companies which were the most social responsible in their sector. With the use of the Carhart factor model we examined the research question. The constant of the Carhart four-factor model tells us whether the excess return can be called abnormal or not. Using the Carhart four-factor model we constructed a second model to test whether additional social responsible variables would have an influence on the excess return of the social responsible companies. The first model regresses the excess return on the variables: RM_RF, SMB, HML and MOM. The second model regresses the excess return on the previous variables, and also the following variables: GBD, HR and DIV. The results of these regressions show that the socially sustainable companies did not outperform the market proxy. In fact, they significantly scored lower than the proxy market.

4

Table of contents

ABSTRACT 3

TABLE OF CONTENTS 4

CHAPTER 1 Introduction 5

CHAPTER 2 Literature review 7

Chapter 3 Data and Methodology 10

3.1 Data 10

3.2 Methodology 18

Chapter 4 Empirical results 23

Chapter 5 Conclusion and discussion 29

References 31

5

Chapter 1

Over the last few years it has become more important in the eyes of consumers and investors how companies treat their employees and the environment. This can be shown by the fact that there is more money invested today in socially responsible companies than in the past. ‘Today, consumers have access to a wide selection of pro-social goods ranging from ‘‘ecologically produced’’ milk and ‘‘fairly traded’’ coffee to ‘‘green’’ electricity and ‘‘hybrid’’ cars’ (Nilsson, 2008). According to The Forum for Sustainable and Responsible Investment institutional investors in 2005 invested 1,490 billion dollars in environmental, social and governance (ESG), while in 2016 it was 4,725 billion dollars. This is a raise of more than 317%.

According to MSCI1 _{67 percent of millennials believe investments are a way to express social,}

political and environmental value. This is contrary to 36 percent of baby boomers.

Consequently, this means that there is a shift in how people invest their money. Today an upcoming movement is social responsible investing (SRI). Sometimes SRI is also called ethical investment, sustainable investment or socially conscious investing (Nilsson, 2008). Although SRI sounds like something of the last few years, it originated in Jewish, Christian and Islamic traditions (Renneboog, 2007). For example, Renneboog (2007) mentions that in 1139 the

Catholic church imposed a prohibition on usury, this prohibition was not relaxed until the 19th

century.

This all raises the question if social responsible investments are not scrutinized by the restraints it gives to the investing opportunities of individuals. Here for, what this paper examines is:

Do the stocks of the top of the class social responsible companies outperform the stocks of the non-social responsible countries in the US from 2008 to 2015.

According to Renneboog (2007) modern SRI is, in contrast to the ethical investing ways of religious companies, more based on the ethical and social convictions of individual investors. He also argues that the anti-war and anti-racist movements have made investors more aware of the social consequences of their investments. The definition of SRI according to Statman (2006), Schueth (2003) and Shank et al. (2005) is “Integrating personal values and societal concerns with investment decisions”.

Renneboog (2007) addresses that since the early 1990s social responsible companies experienced strong growth in the US, Europe and the rest of the world. According to Nilsson (2008) there are two reasons for that. The first reason is that in the general population there is a transition going on from being a ‘saver’ to becoming an ‘investor’. Regular people are moving their savings into mutual funds, because of this trend, the mutual fund industry is one of the

larger growth industries of the 20th _{century. At the end of 2004 there existed over 8000}

1 https://www.msci.com/documents/10199/1283513/MSCI_Cyclical_and_Defensive_Sectors_Indexes_Methodology_Jun14.pdf/f05126ab-65f6-4f39-b9d2-ad031858475a

6

different mutual funds, this in contrast to only approximately 70 in 1940 (ICI, 2005). Nearly half of all U.S. households owned mutual funds in 2004 compared to less than 6% in 1980 (ICI, 2005). The second reason according to Renneboog (2007) is that people became more keen on the fact to pay an extra fee for products that were consistent with their personal values. This was called ethical consumerism. Because of various corporate scandals, corporate governance and responsibility have become an additional focal point of social responsible investors. Hence, criteria like environmental protection, human rights policy and labor relations are nowadays common in SRI screenings.

To examine our research question the data used will be retrieved from ASSET4, datastream and the database of Kenneth French. Asset4 provides us with the information we need to know about companies that are the best in their sector with respect to social responsibility. With this information four portfolios of the four different sectors will be constructed and there will be examined whether they outperform the proxy market. The proxy market is designed by Fama and French. The values of the data retrieved from the database of Kenneth French are made with the use of constructing a proxy market for the US stock market. The four sectors as a whole act as a proxy market for social responsible companies. The sectors will be examined individually as well as taken as a whole. This is to investigate whether the social responsible proxy market can outperform the conventional proxy market and to examine how the different sectors perform with regards to the conventional proxy market. Another question that will be examined is whether it will improve the abnormal return of the social responsible companies if to the existing regression are added three extra variables that enhances the social responsibility of the companies.

The methodology that this paper uses will be a standard ‘event study’. With the help of the constant in the Carhart four-factor model it will be determined whether the sectors have outperformed the market or not. This will be done for the time period from 2008 to 2015. There are two main models in this paper. One is the standard Carhart four-factor model and the other one is the Carhart four-factor model expanded with three extra variables. These variables are added to determine whether making a company more social responsible makes a significant difference for the outcome of the abnormal returns for the company.

The results show that none of the sectors outperformed the proxy market, nor did the sectors taken as a whole. In fact, when there was found a significant result for the abnormal return of the sectors taken as a whole, but those abnormal returns would be negative. This goes for both the models used in this paper.

The remainder of this thesis is structured as follows: chapter 2 provides a brief theoretical summary of the past researches done on social responsible companies and the question if they could outperform the market. In chapter 3 the data sample is being explained and the methodology is defined. Chapter 4 shows the empirical results that has been achieved given the data and the methodology given in the methodology section. Chapter 5 discusses the obtained results and the conclusion. Furthermore, the limitations will be discussed and possible ideas for future researches.

7

Chapter 2

Social responsible investments have become of more significance the last approximately 25 years in the US. This is on one side because people went from ‘savers’ to ‘investors’ and on the other side, because for people ethical consumerism started to become more and more important. Due to the fact that SRI picked up, a lot more studies were set up to investigate whether participating in SRI also meant if there was an abnormal return to be made (Revelli and Viviani, 2014).

Studies analyzing SRI come to different findings regarding the question if SRI can improve the portfolio return of investors. SRI opponents like Markowitz (1952) and Le Maux & Le Saout (2004) find that because of the selection and exclusion of certain equities, SRI reduces the opportunities an investor has and therefor the ability to diversify the portfolio. Due to this, the portfolio manager faces a smaller universe of equities, which should therefor result in poorer performance by shifting down the efficient frontier. This explanation is also backed by Clow (1999), who argues that because of the choice constraint, a bias towards specific sectors could arise and therefor increase the risk of the investments.

Despite all of this Diltz (1995a,1995b) argues that because stock markets are so liquid, vast and efficient that the effect of being under-diversified will have almost no effect on the portfolio performance of the social responsible investor. According to Merton (1987) the opposite happens as opposed to the SRI opponents. This is because when there is incomplete information, a portfolio that is perfectly diversified is no longer efficient. This makes that the returns on assets which have concentrated information increases compared with the returns on assets that benefit from more information. In other words, if the information on social responsible equities is mostly only available for social responsible investors, who only hold social responsible equities, then those investors can earn higher returns. This is despite the fact that they have an under-diversified portfolio.

In the studies done today there is no real consensus on what the relationship is between SRI and the financial performance of social responsible companies. Some studies argue that there is a negative relationship, others argue that there is a positive relationship, while others believe there is a neutral relationship. This all despite the fact that many authors tried to simulate, or interpret the existing empirical results and try to define the best practice to do an empirical analyses (Kurtz 1997, 2005; Renneboog et al. 2008; Chegut et al. 2011).

Schleifer and Vishny (1997) envisioned that SRIs that are more concerned about governance achieve better performance than SRIs that are more concerned about other criteria. This is due to the fact that governance refers to ‘the ways in which suppliers of finance to corporations assure themselves of getting a return on their investment’ (Schleifer, 1997).

Another explanation for why some studies find that SRI outperform the market is that SRIs are still an emergent movement. This can lead to wrong rational expectations, this incomplete information in effect can cause stock price to be mispriced. Here for, the market of social responsible stocks and conventional stocks can be temporarily inefficient and therefor arbitrage can take place by social responsible stock specialists (Hudson, 2006; Bénabou and

8

Tirole, 2010). Another possibility regarding incomplete information is that when enough investors underestimate the probability of negative information on conventional companies, this will lead to higher than expected returns of social responsible stocks in contrast to the expected returns of conventional stocks (Hamilton et al., 1993).

In the US it is more common to screen on negative criteria, where as in Europe it is more common to screen on positive criteria. Due to the ‘negative’ approach in the US it can happen that entire business industries are excluded from the portfolio. This approach can, according to Markowitz (1952), lead to underperformance of the US portfolios, because of under diversification.

One other explanation of why the social responsible portfolios can also differ with regards to conventional portfolios is the fact that some investors focus on environmental, social and governance (ESG) criteria. These criteria will allow for a bias in the social responsible portfolios towards certain industries. In fact, various empirical studies find that SRI funds may be orientated to ESG criteria and hence, depending on their orientations, the performance of these funds may vary as empirically observed (Dimson et al., 2012).

Bénabou and Tirole (2010) argue that the difference in performance between social responsible stocks and conventional stocks can be explained by systematic risk exposure to known factors or by factors that are not included in often used models which measures stock performance. This explanation is based on Arbitrage Pricing Theory and is validated by empirical studies. Therefor it must be that social responsible stocks are less sensitive to these factors than conventional stocks (Revelli and Viviani, 2014). Fama and French (1993) and Carhart (1997) showed in their studies that sample construction methods can partially correct for the influence of many factors. However, their methods cannot easily correct factors that are valued by the market, but not included in the empirical studies. According to Grossman and Stiglitz (1980) the performance measured by factors that are not included in risk models could be explained by the compensation for research costs. The firm specific risk for social responsible companies is likely to be lower than for conventional companies (Aupperle et al., 1985; MacGuire et al., 1988). There is evidence that this risk is priced by the market (Malkiel and Xu, 1997; Goyal and Santa-Clara, 2003). This could be another explanation of the difference in returns between social responsible stocks and conventional stocks.

The way financial performance is measured is also important for the outcome of the measurement. Different methods have been conducted, ranging from simple evaluations based on raw returns, to single-factor models derived from the capital asset pricing model (CAPM), to complex multifactorial models (Sharpe, 1966; Jensen, 1968; Fama and French, 1993; Carhart, 1997). The more complex models can better explain the SRI effect, taking into account the potential disturbance caused by size, risk and growth potential (Revelli and Viviani, 2014).

Another explanation of why the performance of conventional stocks and social responsible stocks differs is that the time horizon influences the performance of the portfolios. A study with a long-term horizon is more likely to be obtain a more robust and reliable result. When the period of the study is short then there could be a certain result obtained, because of a specific event. This happened for example during the internet bubble. The prices of the social

9

responsible stock increased during this period, but this was not because they were social responsible (Revelli and Viviani, 2014).

Another fact that is of big importance for this paper is the question on how to choose the social responsible stocks. In a theoretical sense, there is no underlying financial framework to relate the marginal social responsibility of an investment to an investment’s performance. In other words, there is no theoretical model to determine how much social responsibility is appropriate, or to define the optimal trade-off between social responsibility and other investment criteria, primarily risk and return. Thus, SRI lies outside the common efficient markets framework used in finance theory to decide on the attractiveness of an investment (Berry and Junkus, 2012).

10

Chapter 3 Data and methodology

The data on Social responsible companies has been, as mentioned before, provided by the database of ASSET4, which in return has been provided by Thomson Reuters. ASSET4 gives the option of retrieving the companies that are the most social responsible per chosen sector and the program also ranks them by the size of their capitalization. This all gives us the name of the different companies used in this paper, these companies are listed in list 1 in the appendix.

This paper uses two models to investigate the research question. The determinants used in the first model are excess market return (Rm_Rf), Small Minus Big (SMB), High Minus Low (HML) and Momentum (MOM) these determinants are provided by the database of Kenneth French. The determinants used in the second model are the same ones used in the first model, but including the next determinants: Green Buildings(GBD), Board Diversity (DIV) and Human Rights Policy (HR) these determinants have been provided by ASSET4. Of these three determinants GBD and HR are dummy variables.

The data on the stock returns of the social responsible companies are provided by datastream. The data goes from 2008, the beginning of the crisis to 2015. This paper only analyses companies that are listed in the US. It does not matter if there are other companies that are listed somewhere else in the world, but also have activity in the US or if the US listed companies have activities anywhere else in the world. All the companies are US stock market listed companies.

All the results of the regressions have been tested for heteroskedasticity and all of them were found the be indeed heteroskedastic. For this reason, all the regressions have been tested again, but then robust. All of the results of the test if the results are heteroskedastic are listed in list 2 in the appendix.

The model used in this paper is the Carhart four-factor model. This model has four independent variables, namely; the excess market return (Rm_Rf), small minus big (SMB), high minus low (HML) and momentum (MOM). The dependent variable in this model is the excess return. For this paper the excess return is the excess return computed for every individual sector and the individual sectors taken as a whole. Next to the Carhart four-factor model there will be made use of a second model. This second model is an expansion of first model, the Carhart four-factor model, used with the extra variables: green buildings (GBD), human rights policy (HR) and board diversity (DIV).

The value for the factors Rm_Rf, SMB, HML and MOM are the same for every company. This is due to the fact that all the data concerning these factors are retrieved from the same source and are calculated the same for all the sectors. This means that the means, standard deviations, minimums and maximums for these variables are the same for every sector from 2008 to 2015.

11

The following variables are regressed in the analysis: Excess return

The excess (monthly stock) returns of the different companies stand for the (monthly stock) returns that the companies realized minus the risk-free rate. This number can be, as will be seen, highly volatile. Baldauf and Santoni (1991) argue that one of the explanations for this could be the rise of programmed trading, which allows an investor to make much faster trades and build in stop loss mechanisms.

A test has been performed to see whether the excess returns are normally distributed. This is not the case for any of the sectors, nor all the sectors taken together. This result can be seen in the appendix in list 3. Due to the non-normality of the variable not much can be said about the standard deviation.

Excess market return (Rm_Rf)

Rm-Rf stands for the market return minus the risk-free rate. The market return calculated by Fama and French is done on the basis of the value-weight return of all the firms

that are listed at The Center for Research in Security Prices2_.

For this paper, only the returns for American based companies are applicable, as the website of Kenneth French also provides the numbers of the factors for only American based companies. Only those numbers were used in this research. The firms listed in the database were all listed on the NYSE, AMEX or NASDAQ. The risk-free rates are all retrieved from Ibbotson Associates. Small minus big (SMB)

The SMB determinant with an average of 0.175 suggests that during the period of research the portfolios constructed to calculate SMB, the stock returns of small-cap firms have outperformed the stock returns of large-cap firms. This suggests that the portfolio Fama and French compiled to estimate these numbers were predominately made with small-cap stocks. According to the research of Fama and French (1997) the stock returns of small-cap firms have outperformed the stock returns of their large-cap competitors. So, the result for this time period is in accordance with the theory.

High minus low (HML)

The HML determinant has a negative mean of 0.144. This can be explained by the fact that the portfolio constructed by Fama and French consisted of low book-to-market (i.e. growth) stocks that outperformed high book-to-market (i.e. value) stocks. This in return suggests that the portfolios were mainly made up by growth stocks. Fama and French found in their research evidence for a value premium. This refers to the higher risk adjusted return of value stocks than for growth stocks. The result found here is therefore contrary to what Fama and French found. This could be due to the chosen period, as the period chosen to research this paper on is during a financial crisis, where the value premium was lower, or maybe even

12

existent. There are experts though, like J. Bogle (2001) who say that the value premium is non-existent and that Fama and French found evidence for it because of time dependency.

The shown descriptive in the tables below show normal results for the means. Except for the mean of HML. The mean of HML here is negative which could mean that the coefficient of HML most of the stocks in this portfolio act like growth stocks. To see if this is correct there has to be taken a deeper look into the data. This will be done by looking at the beta, coefficient, of the determinant to see if that is also negative and if so if it is statistically significant.

Momentum (MOM)

With a mean of 0.006 the variable MOM is close to zero. This suggests that this variable will not have that much influence on the regression. Whether this is the case can only be examined by the regression itself though.

The factor momentum is built into the model to describe the tendency for stock prices to continue rising, when they go up and vice versa for stock prices to continue declining when they go down. This phenomenon has been shown by Carhart (1997). To calculate MOM one needs to subtract the equal weighted average of the lowest performing firms from the equal weighted average of the highest performing firms, lagged one month (Carhart, 1997).

The determinant MOM is close to zero, which would mean that it would not have a big, if any, impact on the return of the stocks. This though should be tested with the Carhart four-factor model to see whether this is true.

Green buildings (GBD)

The dummy green buildings (GBD) tells us whether a firm owns green buildings or not. When a firm does own green buildings, the dummy will have the value one, if not it will have

the value zero. According to the US EPA3 _{the definition of green buildings is the next: “Green}

building is the practice of creating structures and using processes that are environmentally responsible and resource-efficient throughout a building’s life-cycle from siting to design, construction, operation, maintenance, renovation and deconstruction. This practice expands and complements the classical building design concerns of economy, utility, durability, and comfort. Green building is also known as a sustainable or high performance building.”

Human rights policy (HR)

As the variable GBD, also HR is a dummy variable. If a company is involved in human rights policy policies than the dummy will have a value of one. If a company is not involved in human rights policies, then the dummy will have a value of zero. According to the United

Nations Human rights policy4_{, human rights policy can be expressed as follows: ‘’ Human rights}

policy are rights inherent to all human beings, whatever our nationality, place of residence, sex, national or ethnic origin, color, religion, language, or any other status. We are all equally

3_{https://archive.epa.gov/greenbuilding/web/html/about.html#1} 4_{http://www.ohchr.org/EN/Issues/Pages/WhatareHumanRights.aspx}

13

entitled to our human rights policy without discrimination. These rights are all interrelated, interdependent and indivisible.’’

Board diversification (DIV)

The variable board diversity gives us information about the percentage of women that are on the board of a firm. This number can be between zero and one-hundred percent. Erhardt et al. (2003) found in their study between 1993 and 1998 that there is a positive link between having a diverse board of directors (in the case of this paper this means the diversity with regards to women) and the financial performance of the firms.

Table 1: summary data for the sector energy

When looking at table 1, for the sector energy, the excess return for the sector is approximately 0.094 percent which is much lower than the excess market return of 0.643

percent. This could be because the energy sector is, according to MSCI5_{, an anti-cyclical sector.}

Which means that the returns of the sector move up or down less volatile than that of the market.

The dummy variable green buildings has a mean of 0.6 which means that 60 percent of the companies in the energy sector own green buildings. Next to that approximately 53.3 percent of the companies are involved in some form of human rights policy. In this sector the lowest percentage of women in the board of directors can be found. This is the only sector where there is at least one company with zero women in the board. Also, with a maximum of 30 percent that is the lowest that can be found compared to the other sectors. Lemons and Parzinger (2001) explain in their paper that one of the causes for this low percentage could be the glass ceiling that most of the women face in this sector. The average amount of women that take a position in the board of directors is 17.94 percent.

5 https://www.msci.com/documents/10199/1283513/MSCI_Cyclical_and_Defensive_Sectors_Indexes_Methodology_Jun14.pdf/f05126ab-65f6-4f39-b9d2-ad031858475a DIV 1,440 17.94067 7.632316 0 30 HR 1,440 .5333333 .499061 0 1 GBD 1,440 .6 .4900681 0 1 MOM 1,440 .0064583 4.424197 -25.17 11.17 HML 1,440 -.1438542 2.78253 -11.25 7.85 SMB 1,440 .1753125 2.309791 -4.29 6.11 rm_rf 1,440 .6429167 4.829178 -17.23 11.35 excessreturn 1,440 .0939857 13.30714 -54.50289 56.10836 Variable Obs Mean Std. Dev. Min Max

14

Table 2: correlation between the explanatory variables for the sector energy

In table 2 it can be seen that there is a moderate correlation between DIV and GBD, -0.6247 and a moderate correlation between HML and MOM, -0.4521. Although these

correlations seem to be high, as stated in Keller, G.: Managerial Statistics, 9th _{edition, ISBN-13:}

978-1-111-53463-9 the correlation will be called highly when it is (-)0.7 or higher (lower). When this is the case there is so called multicollinearity. Multicollinearity causes the standard error of the coefficient to rise and in response to that it produces a low t-value which leads to a non-significant p-value of the coefficient (Graham, 2003).

Table 3: summary data for the sector technology

In table 3 it can be seen that in the technology sector there is an average excess return of 0.703 percent this is higher than the average excess stock return of the market. This could be due to the pro-cyclical behavior of technology stocks. The excess return does not have a much higher excess return than that of the market, suggesting that the beta, according to the CAPM model, should be close to zero.

In this sector 73.3 percent of the businesses own green buildings and 80 percent are involved in human rights policies. There is an average of 24.14 percent of women on the board of directors. DIV 0.0000 0.0000 -0.0000 -0.0000 -0.6247 0.2709 1.0000 HR 0.0000 0.0000 -0.0000 -0.0000 0.0546 1.0000 GBD -0.0000 0.0000 -0.0000 -0.0000 1.0000 MOM -0.2837 -0.0697 -0.4521 1.0000 HML 0.3605 0.1686 1.0000 SMB 0.3434 1.0000 rm_rf 1.0000 rm_rf SMB HML MOM GBD HR DIV DIV 1,440 24.13667 8.526872 11.11 41.67 HR 1,440 .8 .400139 0 1 GBD 1,440 .7333333 .4423703 0 1 MOM 1,440 .0064583 4.424197 -25.17 11.17 HML 1,440 -.1438542 2.78253 -11.25 7.85 SMB 1,440 .1753125 2.309791 -4.29 6.11 rm_rf 1,440 .6429167 4.829178 -17.23 11.35 excessreturn 1,440 .7031969 13.86007 -70.40686 139.6126 Variable Obs Mean Std. Dev. Min Max

15

Table 4: correlation between the explanatory variables for the sector technology

In this correlation table, it can be seen that there is a moderate negative correlation between HR and DIV, -0.5023, this could affect the t-score as there is a moderate correlation, but not a high correlation, that could influence the standard error of the coefficient. For the variables HML and MOM goes the same as they have a correlation of -0.4521.

Table 5: summary data for the sector healthcare

For the sector healthcare, it can be seen in table that there was an average excess return of 1.023 percent. This is much higher than that of the excess market return. According to MSCI the healthcare sector is an anti-cyclical sector. With respect to this the result can be called considerately strange. One reason for this could be the fact that during the research time there was a financial crisis and as healthcare is a anti-cyclical company this led to higher excess returns than that of the market.

66.67 percent of the firms owned green buildings and 46.67 percent of the firms were active in human right policies in one way or another. The boards were on average made up of 26.39 percent women. DIV -0.0000 -0.0000 -0.0000 0.0000 0.2267 -0.5023 1.0000 HR -0.0000 -0.0000 0.0000 0.0000 0.0754 1.0000 GBD 0.0000 -0.0000 -0.0000 -0.0000 1.0000 MOM -0.2837 -0.0697 -0.4521 1.0000 HML 0.3605 0.1686 1.0000 SMB 0.3434 1.0000 rm_rf 1.0000 rm_rf SMB HML MOM GBD HR DIV DIV 1,440 26.386 8.026797 11.11 37.5 HR 1,440 .4666667 .499061 0 1 GBD 1,440 .6666667 .4715683 0 1 MOM 1,440 .0064583 4.424197 -25.17 11.17 HML 1,440 -.1438542 2.78253 -11.25 7.85 SMB 1,440 .1753125 2.309791 -4.29 6.11 rm_rf 1,440 .6429167 4.829178 -17.23 11.35 excessreturn 1,440 1.023256 10.4476 -72.60282 99.99 Variable Obs Mean Std. Dev. Min Max

16

Table 6: correlation between the explanatory variables for the sector healthcare

Again, there is a moderate correlation between HML and MOM, -0.4521.The rest of the

correlation values correspond with those found earlier.

Table 7: summary data for the sector industrials

According to MSCI the sector industrials is a pro-cyclical sector. In that light the excess stock return of 0.586, which is lower than the market return of 0.643, again with the knowledge that a big part of the data of this research is from a period of a financial crisis is in correspondence with the theory that pro-cyclical firms are more sensitive to market changes than non-cyclical firms (Berk and Demarzo, 2017).

The casualty occurs here that all of the industrial firms own green buildings. This makes that the mean is equal to zero. 73.33 percent of the firms are involved in human right policies and the boards of directors of the firms are, on average, made up of 27.28 percent women.

DIV -0.0000 -0.0000 -0.0000 0.0000 0.2647 0.3246 1.0000 HR -0.0000 -0.0000 0.0000 0.0000 0.3780 1.0000 GBD 0.0000 -0.0000 0.0000 0.0000 1.0000 MOM -0.2837 -0.0697 -0.4521 1.0000 HML 0.3605 0.1686 1.0000 SMB 0.3434 1.0000 rm_rf 1.0000 rm_rf SMB HML MOM GBD HR DIV DIV 1,440 27.28067 6.710957 16.67 36.36 HR 1,440 .7333333 .4423703 0 1 GBD 1,440 1 0 1 1 MOM 1,440 .0064583 4.424197 -25.17 11.17 HML 1,440 -.1438542 2.78253 -11.25 7.85 SMB 1,440 .1753125 2.309791 -4.29 6.11 rm_rf 1,440 .6429167 4.829178 -17.23 11.35 excessreturn 1,440 .5860116 10.95336 -44.0343 106.0172 Variable Obs Mean Std. Dev. Min Max

17

Table 8: correlation between the explanatory variables for the sector industrials

In table 8 the dummy variable GBD had a value 1 for every industrial company, so there couldn’t be a correlation computed for it with respect to the other explanatory variables. Next to that, again there is a moderate correlation between HML and MOM, -0.4521.

Table 9: summary data for the social responsible proxy market

In table 9 all the summary data for the social responsible market proxy can be examined. As could have been expected the excess return for the social responsible market proxy comes most close to the excess stock return of the market. Though, the excess return is a little bit lower than the excess market return. After investigating the coefficient later in chapter 4 it can be determined whether difference is significant or not.

Furthermore, 75 percent of all the companies own green buildings and 63.33 percent of the companies are in one way or another involved in human right policies. With a maximum of 41.67 percent and a minimum of zero percent on average there are 23.94 percent women in the board of directors for the social responsible market proxy.

DIV -0.0000 -0.0000 0.0000 -0.0000 . 0.0846 1.0000 HR 0.0000 -0.0000 -0.0000 -0.0000 . 1.0000 GBD . . . . . MOM -0.2837 -0.0697 -0.4521 1.0000 HML 0.3605 0.1686 1.0000 SMB 0.3434 1.0000 rm_rf 1.0000 rm_rf SMB HML MOM GBD HR DIV DIV 5,760 23.936 8.565684 0 41.67 HR 5,760 .6333333 .4819362 0 1 GBD 5,760 .75 .4330503 0 1 MOM 5,760 .0064583 4.423045 -25.17 11.17 HML 5,760 -.1438542 2.781805 -11.25 7.85 SMB 5,760 .1753125 2.30919 -4.29 6.11 rm_rf 5,760 .6429167 4.82792 -17.23 11.35 excessreturn 5,760 .6016127 12.23156 -72.60282 139.6126 Variable Obs Mean Std. Dev. Min Max

18

Table 10: correlation between the explanatory variables for the social responsible proxy market

In table 10 it can be seen that the values found for the correlations are more or less the same as in the specific sectors, as could been expected of course. As this correlation table shows an average for all the sectors taken together all the moderate correlations are flattened out by the lower correlation values. This then results in the fact that there are no moderate correlations anymore.

3.2 Methodology

To research whether social responsible companies have an abnormal stock return in the period between 2008 to 2015 in comparison to non-social responsible companies in the US, first there has to be determined which social responsible companies will be chosen to research this question. This will be done by making use of the search engine called ASSET4. ASSET4 is a tool developed by Thomson Reuters to identify which companies perform the best and the worst when they are being examined on different environmental, social, economic and corporate governance performances. In the database of ASSET4 are over 3200 companies that

are globally listed. There are four main pillars that ASSET46 _{rates, namely: economic}

performance, environmental performance, social performance and corporate governance performance. Each of these four pillars are then split up in different categories. These categories then have more than 250 indicators which in turn exist of more than 750 data points plus there

is a link to public data sources. ASSET47 _{gives the option to pick the specific benchmarks as one}

prefers.

It is estimated that investors representing more than €2.5 trillion in assets under management use the ASSET4 data, including prominent investment houses such as BlackRock (Cheng, Ioannou and Serafeim (2014).

This paper selects the fifteen best scoring social responsible companies for each sector. These fifteen best performing companies are chosen on taking the top five for the large capitalization companies, the top five mid capitalization companies and the top five small

6_{https://customers.reuters.com/community/fixedincome/material/ASSET4ESGSCORES.pdf} 7 https://www.thomsonreuters.com/content/dam/openweb/documents/pdf/tr-com-financial/report/starmine-quant-research-note-on-asset4-data.pdf DIV -0.0000 0.0000 0.0000 0.0000 0.0707 0.0865 1.0000 HR 0.0000 0.0000 -0.0000 0.0000 0.1997 1.0000 GBD -0.0000 0.0000 -0.0000 -0.0000 1.0000 MOM -0.2837 -0.0697 -0.4521 1.0000 HML 0.3605 0.1686 1.0000 SMB 0.3434 1.0000 rm_rf 1.0000 rm_rf SMB HML MOM GBD HR DIV

19

capitalization companies. All the companies are ranked for the amount of points they score on environmental, social, economic and corporate governance performances. When there are two or more companies in a sector that score the same amount of points. The decision has been made to always go for the company with the highest market capitalization. Some companies are still in the ASSET4 database, but have been delisted before the 01/01/2016. When that is the case those companies are excluded from the portfolio. Also, there were companies taken out of the dataset, because there was insufficient data on them. This could be due to the fact that for example they got stock enlisted after 2008.

This paper researches the next sectors: Healthcare, industrials, energy and technology. This is done to avoid industry effects (Schmalensee, 1985). These sectors have been chosen on the fact that they do not have a lot of overlap and this way the results of the regression will not have any sector bias.

The 4 different sectors have been chosen to, if all taken together, represent the social responsible proxy market, rather than just taking the best performing companies of all the social responsible companies, because in that way there could be a bias towards one specific sector and perhaps only large capitalization companies would by randomness be chosen that way. To avoid all of these insecurities the decision was made to choose four sectors and select the top five companies for the large capitalization companies, the mid capitalization companies and the small capitalization companies.

The excess return of the four sectors will be regressed on the independent variables of the Carhart four-factor model. This allows us to examine how the sector portfolios behaved compared to the proxy market portfolio. The four independent variables are directly linked to the overall market, because the portfolios that Fama and French constructed to compute the independent variables are made up from companies in the overall market and are chosen carefully, so that no bias would occur. The formula that Fama and French used can be seen later in equations three, four and five. This all allows us to see how the excess return of the social responsible portfolio behaves against the proxy market given the independent variables. The theory of Fama And French (1997) tells us that the alpha of the model, the constant, explains if the excess returns found for a company or a portfolio are abnormal or not. If the constant is significant and positive there is an abnormal return, reversely if the constant is negative and significant there is also an abnormal return, but then negative of course. If, for example, we would have a value for the constant equal to one and it would be significant, then we could conclude that as the Carhart four-factor model represents the market proxy that then there would be an abnormal return for that portfolio.

To be able to calculate the monthly returns of the chosen companies the adjusted closing prices were needed. The Thomson Reuters database provides us with the adjusted closing prices of the stocks. By using the following formula, the monthly returns could

be calculated: 𝑁𝑒𝑤 𝑝𝑟𝑖𝑐𝑒−𝑜𝑙𝑑 𝑝𝑟𝑖𝑐𝑒

𝑜𝑙𝑑 𝑝𝑟𝑖𝑐𝑒 ∗ 100%.

Another question this paper wants to address is whether it is beneficial for a company to become even more social responsible and what that will do to the excess stock returns of those

20

companies. Capelle-Blancard and Monjon (2012) found some evidence for this to be true. This question will be tested by adding the new benchmarks (variables) to the model. To be able to compare the four different sectors with each other the benchmarks should be something that they all have in common. For example, the CO2 emission in the energy sector is much higher than in the healthcare sector. To avoid scenarios like this more general benchmarks have been chosen. These are the three determinants: green buildings (GBD), human rights policy (HR) and board diversity (DIV). The determinants green buildings and human rights policy are dummy variables. Which will be one if the companies have green buildings and are involved in human rights policy respectively and conversely it will be zero when companies do not have green buildings and are not involved in human rights policy respectively. The determinant board diversity represents the percentage of what part of the company’s board is represented by women.

There will be done a second regression on the selected stocks and this regression will reveal whether one of the determinants added has a significant effect on the return of the stock. With this it can be concluded that there might be certain determinants that are more important to investors than others. The new determinants have been chosen on the basis of ecological, social and corporate governance criteria, which is according to Renneboog et al. (2008) what social responsible investors apply when choosing the companies they want to invest in.

The model that will be used to examine the main question of this paper is the Carhart four-factor model. The Fama-French three-factor model is widely used to examine the abnormal return of a social responsible stock or fund. This model, though, is not widely used to examine the abnormal return of social responsible companies, but is was used for example by Kemp and Osthoff (2007) and Galema et al. (2008). Kemp and Osthoff (2007) found that social responsible companies in some cases did have an abnormal return.

The Carhart four-factor model is a model expended from the Fama-French three-factor model. Carhart added the variable momentum (MOM) to the model which he found to be significant important to test whether the returns were abnormal or not (Carhart, 1997). The Fama-French three-factor model is derived from the CAPM model (Fama and French, 1997). The difference is that the model of Fama and French controls for the impact of the size factor and the book-to-market factor. The size factor variable is called SMB, which stands for small market capitalization minus big market capitalization and the book-to-market factor is called HML which stands for High book-to-market ratio minus low book-to-market ratio (Fama and French, 1997). In 2005 Bauer et al. found that socially responsible mutual funds have different loading factors with respect to the conventional mutual finds. To control for such differences the next regressions are estimated:

Rt,i− RFt = αi + β1i (RMt − RFt) + β2i SMBt + β3i HMLt + β4i MOMt + εi,t (1)

Rt,i− RFt = αi + β1i (RMt − RFt) + β2i SMBt + β3i HMLt + β4i MOMt + β5i GBD + β6i HR + β7i DIV

21

The dependent variable is the monthly return of portfolio i in month t in excess of the risk-free rate. The independent determinants are the returns of four zero-investment factor portfolios. Rmt−Rft denotes the excess return of the market portfolio over the risk-free rate. The market portfolio is the CRSP value-weighted index. SMBt denotes the return difference between a small and a large capitalization portfolio in month t. HMLt denotes the return difference between a high and a low book-to-market portfolio in month t. MOMt denotes the return difference between portfolios of stocks with high and low returns over the past twelve months. Alpha denotes the abnormal return of the portfolio i.

In the second regression GBD stands for the dummy variable green buildings if the company has green buildings the dummy variable becomes a one, if not then it will become a zero. DIV stands for the dummy variable board diversity. If the board is diverse with also women, this variable will give the percentage of the diversity. HR stands for human right policy, if the company is active in the human right policy then the dummy variable will become a one, if not it will become a zero.

To test whether social responsible companies can outperform conventional companies the following hypotheses will be tested at a level of five percent significance.

H0 → The constant has a coefficient equal to zero H1 → The constant has a coefficient unequal to zero

The Fama/French factors are constructed using the 6 value-weight portfolios formed on size and book-to-market. SMB (Small Minus Big) is the average return on the three small portfolios minus the average return on the three big portfolios. The formula for this is:

SMB = 1

3(𝑠𝑚𝑎𝑙𝑙 𝑣𝑎𝑙𝑢𝑒 + 𝑠𝑚𝑎𝑙𝑙 𝑛𝑒𝑢𝑡𝑟𝑎𝑙 + 𝑠𝑚𝑎𝑙𝑙 𝑔𝑟𝑜𝑤𝑡ℎ) −

1

3 (𝑏𝑖𝑔 𝑣𝑎𝑙𝑢𝑒 + 𝑏𝑖𝑔 𝑛𝑒𝑢𝑡𝑟𝑎𝑙 +

𝑏𝑖𝑔 𝑔𝑟𝑜𝑤𝑡ℎ) (3)

HML (High Minus Low) is the average return on the two value portfolios minus the average return on the two growth portfolios. The formula for this is:

HML = 1

2 (𝑠𝑚𝑎𝑙𝑙 𝑣𝑎𝑙𝑢𝑒 + 𝑏𝑖𝑔 𝑣𝑎𝑙𝑢𝑒) −

1

2 (𝑠𝑚𝑎𝑙𝑙 𝑔𝑟𝑜𝑤𝑡ℎ + 𝑏𝑖𝑔 𝑔𝑟𝑜𝑤𝑡ℎ) (4)

Mom is the average return on the two high prior return portfolios minus the average return on the two low prior return portfolios. The formula for this is:

MOM = 1

2 (𝑠𝑚𝑎𝑙𝑙 ℎ𝑖𝑔ℎ + 𝑏𝑖𝑔 ℎ𝑖𝑔ℎ) −

1

2 (𝑠𝑚𝑎𝑙𝑙 𝑙𝑜𝑤 + 𝑏𝑖𝑔 𝑙𝑜𝑤) (5)

Although the variable SMB is in the regression which should control for the impact of small and big market capitalization companies, the fact that this paper divides the companies on their market capitalization does not mean that that bias could not still arise.

22

This paper will have a closer look on environmental criteria’s, social criteria’s and corporate governance criteria. With this closer look this study tries to determine which criteria’s have a significant result on the stock returns on the companies and the portfolios. ‘What is the effect of additional benchmarks on the performance of a stock?’.

23

Chapter 4

As has been discussed before, many different researches have been conducted on the question whether SRIs not only pay off to achieve the goal of making the world more

sustainable, but also to achieve financial success. In this chapter we will examine what the results are for our specific research. We will have a look at how the social responsible market proxy has performed against the conventional market proxy. Next to that the results of the specific sectors will be discussed as well. The results are based on the data of the chosen companies given by setting the desired benchmarks in ASSET4 from 2008 to 2015. Again, some companies have been taken out of the list of chosen companies. This is due to either not having enough data for these companies or the companies have been taken out, because in the ranking they were ranked at the same place as another country and as said before, the decision was made to then go for the country with the highest market capitalization.

The results will be examined by first looking at the social responsible market proxy and afterwards at the specific sectors.

Table 11: Determinants of the excess return for the social responsible market proxy

Table 11 gives us the results of regressions one and two. Given these results the next events can be concluded: when the market return (Rm_Rf) rises with one percent then the excess return will rise with 1.196 percent. When Small Minus Big (SMB) rises with one percent then the excess return will rise with 0.525 percent. For these two variables, it can be stated that this is correct as they are both highly significant. For the variable High Minus Low (HML) though it can be seen that a one percent increase the variable will lead to a decrease in excess return of 0.117 percent. As this coefficient is not significant this result is not a correct conclusion. Going on for the variable Momentum (MOM) it can be concluded that a one percent rise in

24

momentum will lead to a decrease in the excess return of 0.248 percent. Until now the variables Rm_Rf, SMB, HML and MOM had the same coefficients in both the regressions.

Now only the variables of regression two will be evaluated. The dummy variable Green Buildings (GBD) has a coefficient of 0.260 as this coefficient is not significant it does not influence the excess return. Would this coefficient have been significant then a value of one for GBD would have let to a 26 percent change in the excess return. The same goes for the dummy variable Human Rights Policy (HR). The p-value of the coefficient is not significant. Would it have been significant then the excess return would decrease with 25.7 percent. The variable Board Diversity (DIV) has a coefficient that is also not significant. As this variable is not a dummy variable a one unit increase in the variable would have let, if significant, to a 3.339 percent increase in the excess return.

Looking at the p-values of the constants of both the regressions it can be concluded that they are both significant. Nevertheless, they are both negative, this thus results in the conclusion that both of the models used find that social responsible companies do have an abnormal return, but those abnormal returns are negative. Looking at the values for the constants it can be seen that the constant of the second regression has a much bigger, negative, impact on the excess return than the constant of the first regression does. This could mean that adding the variables GBD, DIV and HR actually lowers the excess return of the social responsible market proxy. For this to be concluded a t-test has to be performed on the two found values of the constants.

Looking at the adjusted R-squared result it can be seen that the R-squared is the same for both regressions. This means that both of the regressions explain the variance in the dependent return equally well. With an adjusted R-squared value of 0.257 it means that 25.7 percent of the variance in the independent variable is explained by both of the regressions.

25

Looking at table 12 it can be seen that for both regressions the variables market return and SMB are both significant. As both of the variables have a positive sign it means that they both have a positive effect on the dependent variable. The variables HML and MOM are also significant in both of the regressions, but they have a negative effect on the dependent variable.

The results show that both of the constants in the two regressions are negative, though only the constant in the first regression is significant. These numbers mean that after the variables GBD, HR and DIV are added to the regression these variables cannot change the numbers of the excess return in such a way that there is an abnormal return to be found. Whereas after omitting these extra variables there is ground for stating that there is an abnormal return, though it is negative.

Looking at the adjusted R-squared of both the regressions almost the same result as with the social responsible market proxy can be seen. Both of the regressions almost explain the same percentage of the variance in the dependent variable. There is only a difference of 0.1 percent. The first regression does a slightly better job to explain the variance in the dependent variable than the second regression. With this result it can be concluded that adding the extra variables to the second regression does not have any influence on explaining the excess return, in fact it does a worse job in explaining the excess return than the regression without the extra variables.

Table 13: Determinants of the excess return for the healthcare sector

In table 13 it is shown that only the variables excess market return and SMB are significant. They both have a positive effect on the excess return. The constant of the first regression is positive, while not significant, so there is no abnormal return in this sector. The constant of the second regression is negative, but also not significant and thus again there is no abnormal return. The difference in the signs for the constants does suggest that in this sector the added variables in regression two have a negative influence on the excess return. This is

26

somewhat strange as two out of the three added variables have a positive sign and combined they outperform the variable with the negative sign. Though, this cannot be taken as a certain fact as none of the added variables are significant. The adjusted R-squared of both of the models is 0.233 which means that 23.3 percent of the variance in the dependent variables can be explained by the variance in the explanatory variables.

Table 14: Determinants of the excess return for the industrials sector

In table 14 it can be seen that in the sector industrials, the variables excess market return, SMB and MOM are all significant. The variables excess market return and SMB have a positive effect on the dependent variable excess return, while the variable MOM has a negative effect. In this model the dummy variable GBD is omitted due to the fact that all the companies in this portfolio owned a green building, and thus had a value of one for this variable. The constant in this sector is for both of the models negative, though not significant. Hence for this sector the excess return of the portfolio is no different than that for the market proxy. The adjusted R-squared of the first model tells us that 35.8 percent of the excess return can be explained by the explanatory variables. For the second model this percentage is slightly lower with a percentage of 35.7 percent. This means that just as for the energy sector the first regression does a better job to explain the variance in the dependent variables than does the second regression.

27

Table 15: Determinants of the excess return for the technology sector

In table 15 the results for the technology sector show that variables market return, SMB and MOM are significant and therefor do explain the dependent variable. The constant of both regressions are negative and not significant. Also, both of the regression have an adjusted R-squared result of 0.318 which means that they both explain the variance in the dependent variable equally well with a percentage of 31.8 percent.

After evaluating all the generated results of the social responsible market proxy and the four different sectors there are some things that can be concluded. By having, in regression one, the dependent variable regressed on the independent market proxy variables it can be seen how the excess return of the specific sector or social responsible market proxy is influenced by the independent market proxy variables. Due to the fact that in the first regression there are only independent variables added there that are linked to the market proxy the constant will tell us how much the difference is between the excess return of the chosen sectors and the given market proxy. This allows us to decide whether the computed constant is telling us whether the sector does have an abnormal return or not.

First, after examining the p-value of the constants in table 1, it can be concluded that both models show that the abnormal return that has been found is negative. This means that after diversifying your money into different sectors with companies that are social responsible there is a probability of more than 95 percent that the return of the invested money will be lower than when that money would have been invested in non-social responsible companies. This answers the main research question of this paper.

Looking at the GBD, HR and DIV from the second regression it can be concluded that these variables do not have an influence on the excess return of the social responsible proxy

28

market. Thus, in return tells us that these variables individually are not important to investors. If they would be important then adding these variables would have a significant value.

When zooming in into the sectors more specifically it can be seen in table three, four and five that for the sectors healthcare, industrials and technology none of the constants were significant. This means that those sectors did not out -or underperformed the proxy market of conventional markets.

In table two it can be seen that in the first regression the energy sector has a negative abnormal return compared to the given proxy market. This result is not the same when looking at the second regression. There it can be seen that the constant is also negative, but not significant. This means that the added variables GBD, HR and DIV cannot influence the excess return in such an either positive or negative way that the constant becomes significant. As can be seen the standard error of the constant changed a lot after adding the extra variables. This has probably to do with the correlation between the added variables. As can be seen in the correlation matrix there is a moderate correlation between the variables, but the correlation is not high enough to call the variables multicollinear. A high correlation between two variables makes that the standard error of the coefficient rises. This is due to the fact that the error terms of the two (or more) variables are correlated with each other. Due to the higher standard error

the t-value of the coefficient goes down. The formula for the t-value is 𝛽0−0_𝑆𝐸 given that the null

hypothesis for the coefficient is that the coefficient is equal to zero (Stock and Watson, 2012). As can be seen, as the SE rises the t-value will go down and therefore not be significant.

29

Chapter 5

This paper researched whether there could be made an abnormal return when investing in social responsible companies compared to conventional companies. To do so this paper used the alpha coefficient in the Carhart four-factor model in order to determine if there was an abnormal return. The data used for this research was from the beginning of 2008 to the end of 2015. Furthermore, this paper tried to examine whether putting more social responsible restrictions on a firm could increase the abnormal return of those portfolios, as being proposed by Capelle-Blancard and Monjon (2012).

The results show that when all sectors are taken together there can be traced a negative abnormal return with respect to the first model. The alpha is approximately -0.27, this means that for every one point increase in the constant, the excess return will decrease with 0.27. The R-squared in this model tells us that 28.70 percent of the variation in the dependent variable can be explained by the explanatory variables. The value of the R-squared is rather low when seeing that four out of the five coefficients in the model are significant. This tells us that although the variables; Rm_Rf, SMB, MOM and alpha are significant. Still there could be omitted variables in the model. Exposing the model to the Ramsey RESET test does not show that the model suffers from functional form misspecification. The results for the Ramsey RESET test can be found in the appendix in list 4.

When looking at the second model again we find a significant negative alpha score of -1.12. Which means that after adding additional social responsible restrictions on the firms in the portfolio the portfolio did not perform better than the proxy market, it got worse.

These numbers show that investing in the four specific sectors together does not result into an abnormal return. In fact, it leads to a negative abnormal return compared to the market proxy. Several studies have been conducted on this matter, some found evidence of the possibility to get an abnormal return, while others found that the return would be the same, or even negative. In line with these findings, the findings of this research are not odd.

With the second model this research tried to see what the effect would be on adding multiple variables that are considered to enhance the social responsibility of a company. After adding these factors the alphas of the coefficient of those variables turned out to be not significant for a p-value of five percent or smaller. The result of this regression was that the constant, the indicator if the sector or all the sectors combined, was significant, but highly negative. Given also that none of the variables GBD, HR and DIV have a significant value it can be concluded that investors do not find these variables to be important.

If we then focus on the individual sectors, it can be seen that the p-value for alpha shot up after adding the variables in model 2. All the rest stayed more or less the same. These much lower significance levels could be due to the noise of the added variables. Another interesting difference is that with the regressions of the second model the coefficients of the MOM and HML variables tented to be less significant. As explained before already, this could be due to the moderate correlation between these variables.

With these results it can be concluded that investing in social responsible companies, although it enhances the sustainability of the earth, it is not a good alternative, compared to the proxy market, to invest in.

30

This study has some limitations. For example, it tries to construct a market portfolio with only four different sectors. The number of sectors could be expanded to get a better overview of the market movement for social responsible companies.

Second, the period that this paper does its research on is partly during the financial crisis. Though an event like a financial crisis to occur is not odd, a follow up study could research whether an abnormal return could have been made given the data would not be partially from a financial crisis. The bad results from this study could be this negative due to the crisis.

A follow-up research could consider not to do a long-term study, but rather a short-term study. This would suite investors who want to be social responsible, but still make an abnormal return on the stock market in the short run.

One important question is, what makes the social responsible companies underperform? Is it the fact that they are social responsible and thus have to comply to strict social rules, or are the companies just not as good in what they do as compared to conventional companies? This is an interesting question that could be tried to be answered in another study.

31

References

Attig, N., El Ghoul, S., Guedhami, O., & Suh, J. (2013). Corporate social responsibility and credit ratings. Journal of Business Ethics, 117(4), 679-694.

Baldauf, B., & Santoni, G. J. (1991). Stock price volatility: some evidence from an ARCH model. Journal of Futures Markets, 11(2), 191-200.

Barber, B. M., & Lyon, J. D. (1997). Detecting long-run abnormal stock returns: The empirical power and specification of test statistics. Journal of financial economics, 43(3), 341-372. Bello, Z. Y. (2005). Socially responsible investing and portfolio diversification. Journal of Financial Research, 28(1), 41-57.

Berk, J., and P. DeMarzo (2017). Corporate Finance (4th Global Edition). Pearson Berry, T. C., & Junkus, J. C. (2013). Socially responsible investing: An investor perspective. Journal of Business Ethics, 112(4), 707-720.

Berry, T. C., & Junkus, J. C. (2013). Socially responsible investing: An investor perspective. Journal of Business Ethics, 112(4), 707-720.

Bilbao-Terol, A., Arenas-Parra, M., Cañal-Fernández, V., & Bilbao-Terol, C. (2016). Multi-criteria decision making for choosing socially responsible investment within a behavioral portfolio theory framework: a new way of investing into a crisis environment. Annals of Operations Research, 247(2), 549-580.

Blanchett, D. (2011). Exploring the “Cost” of Socially Responsible Investing (SRI). Retrieved from http://www.fi360.com/main/pdf/2011_blanchett_presentation.pdf

Capelle‐Blancard, G., & Monjon, S. (2014). The performance of socially responsible funds: does the screening process matter?. European Financial Management, 20(3), 494-520. Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and access to finance. Strategic Management Journal, 35(1), 1-23.

Cheng, B., Ioannou, I., & Serafeim, G. (2014). Corporate social responsibility and access to finance. Strategic Management Journal, 35(1), 1-23.

Dam, L., & Scholtens, B. (2015). Toward a theory of responsible investing: On the economic foundations of corporate social responsibility. Resource and Energy Economics, 41, 103-121. Erhardt, N. L., Werbel, J. D., & Shrader, C. B. (2003). Board of director diversity and firm financial performance. Corporate governance: An international review, 11(2), 102-111. 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. Graham, M. H. (2003). Confronting multicollinearity in ecological multiple