Socially Responsible Investments
The effect of SRI on financial performance of mutual
funds over time
Master Thesis by Emma N. Penthum (10545344) University of Amsterdam – Amsterdam Business School
MSc Finance – Asset Management Supervisor: mw. dr. E. Zhivotova
June 2018
Abstract
Socially responsible investment has been increasing rapidly over the last decade. A rising interest of individuals in SRI is pushing companies towards being more socially responsible. This paper examines the relative financial performance of SRI mutual funds with respect to conventional mutual funds. Moreover, this paper investigates if the relative performance has changed over time. To investigate this, first the Sharpe ratio is computed, and second, a broad spectrum of models is estimated to investigate the abnormal returns. Using the t-statistic, this study determines if the financial performance between SRI funds and conventional funds differs, and if the performance has changed over time. Moreover, this study also investigates the influence of four fund characteristics on financial performance. These fund characteristics are the expense ratio, size, age, and the screening intensity. Concluding, this study is not able to find significant differences in financial performance between SRI funds and conventional funds. However, weak evidence suggests that the financial performance of SRI funds and conventional funds is converging. Moreover, weak evidence is found that funds with a high expense ratio perform worse than funds with a low expense ratio, and older funds perform worse than young funds. Stronger evidence was found that suggests that screening intensity positively affects fund performance. Lastly, this study
Statement of Originality
This document is written by Emma Penthum who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the
Table of content
1. Introduction ... 4
2. Literature review ... 7
2.1 Socially responsible investments ... 7
2.1.1. What is SRI? ... 7
2.1.2. Interest in socially responsible investments ... 8
2.1.3. Screening criteria ... 9
2.1.4. Implementation strategies of SRI ... 9
2.2 Effect of SRI on financial performance ... 10
2.2.1. Explanations for the underperformance of SRI funds ... 11
2.2.2. Explanations for the outperformance of SRI funds ... 12
2.2.3. Similarities between SRI funds and conventional funds ... 14
2.2.4. Fund characteristics ... 14 2.3 Hypotheses ... 15 3. Methodology... 17 3.1. Fund set-up ... 17 3.2. Financial performance ... 17 3.2.1. Sharpe ratio... 17 3.2.2. Regression models ... 18 3.2.3. Fund characteristics ... 19 3.3. Robustness checks ... 20 4. Data ... 23 4.1 Data collection... 23 4.1.1 Fund statistics ... 23
4.1.2. Risk-free rate and market return benchmark ... 24
4.1.3. Factor benchmarks ... 24
4.1.4. Screening data and fund set-up ... 24
4.2. Descriptive statistics ... 25 4.2.1. Mutual funds... 25 4.2.2. Returns... 27 4.2.3. Screening data ... 28 5. Results... 32 5.1 Financial performance ... 32 5.1.1. Sharpe ratio... 32 5.1.2. Regression models ... 33 5.1.3. Fund characteristics ... 36 5.2 Robustness checks ... 39 6. Conclusions ... 46
6.1 Summary and discussion ... 46
6.2 Limitations ... 47
6.3 Recommendations ... 48
1. Introduction
Over the last decades, investors have shown an increasing interest in socially responsible investments. According to the Report on Socially Responsible Investing Trends in the United States, the amount invested in socially responsible funds grew from 2.71 trillion dollars in 2010 to 8.72 trillion dollars in 2016. Moreover, these investments represented 11% and 21.6%, respectively, of the total amount of assets under management. These numbers show that the amount of socially responsible investments as part of the total assets under management almost doubled. This signifies the increased importance of understanding socially responsible investments.
Socially responsible investment is rising rapidly, and not only the amount invested, but also the existing time of some socially responsible mutual funds has doubled since the first empirical papers on their performance were published. Moreover, previous papers used shorter datasets and often did not evaluate different time periods. Additionally, conflicting results have been presented in the past. Therefore, it is questionable whether the financial performance of these funds has changed over time. The main question of this study will be: “Is there a significant change in the financial performance of socially responsible mutual funds?”.
Socially responsible mutual funds (“SRI funds”) aim to providing investors with high performance, while selecting their stocks based on certain screening criteria. These screening criteria can often be categorized within environment, social or governance criteria. There are many different strategies to implement social responsibility within a fund, which will be discussed later. In addition, there are different motives for investing in SRI funds. First, Koedijk et al. (2005) identified the value-based investor. This investor tries to align his or her beliefs with its portfolio holdings. Furthermore, they identify the value-seeking investor. This investor, in contrast to the value based investor, only targets financial rewards. Therefore, a frequently asked question on this topic, is if these socially responsible mutual funds are able to outperform the conventional mutual funds. Or are investors willing to give up on financial performance in order to invest in these more ethically correct funds?
Several papers, using different methodologies, have investigated this problem. For example, Bauer et al. (2005) did not find a significant difference between SRI funds and conventional funds, even when controlling for common factors. Similar results were found by Bello
(2015) both focused on the financial crisis and found SRI funds outperform conventional funds during this time period.
As Bauer et al. (2005) explained, the origins of SRI are based on ethical investing. Nowadays, the rising interest is due to social awareness in issues such as civil rights, the environment and nuclear energy. As well, individuals are encouraged by the public opinion to invest in socially responsible firms. This public opinion is, therefore, also challenging companies to behave in a more responsible manner in order to attract and keep investors. The increase in both social awareness and the number of socially responsible investments, is leading a growing number academics towards this topic. Broader insights into this topic for interested individuals, investors and funds are particularly important for the future development of SRI funds.
This research will contribute to the current literature as tries to help solve the puzzle on financial performance of socially responsible mutual funds. It is one of the first to assess the differences in financial performance of SRI funds in different time periods. Additionally, the length of the dataset is longer than in all the previous papers. Furthermore, it will assess how portfolio characteristics influence the financial performance of these SRI funds. It will be useful for investors to understand both how fund characteristics affect financial performance and how these effects are developing over time. This study will also compare differences in return between SRI funds and conventional funds over time, to assess if the performance of these two types of funds are converging. Moreover, a spectrum of models will be used to study the differences to give broader insights into the results.
To study the research question, U.S. socially responsible equity mutual funds as defined by the United States Forum for Responsible and Sustainable Investment will be used. Moreover, a sample of U.S. conventional equity mutual funds obtained from the CRSP database will be assessed. The Sharpe ratios (1994) of both socially responsible funds and conventional funds will be compared in different time periods. As well, Jensen’s alpha (1969) will be estimated using four different models to compare the abnormal returns between funds and over time. Finally, the effect of the expense ratio, fund size, fund age and the screening intensity on financial performance is evaluated.
The remainder of this paper is as follows. First, the theoretical framework explains what socially responsible investments are, why investors are interested, what screening criteria can be used and how their financial performance is affected. The third section gives an extensive
description of the methodology. Thereafter, section 4 shows how the data is obtained and presents the descriptive statistics. The results and robustness checks are presented in section 5. The final part of this study gives the conclusions, limitations and suggestions for further research on this topic.
2. Literature review
Section 2.1 explains the most important theoretical elements of socially responsible investments. Section 2.2 gives an in-depth description of the financial performance of SRI funds, relative to conventional funds, and summarizes the current literature on the influence of fund characteristics on fund performance.
2.1 Socially responsible investments
To be able to answer the research question, it is important to define what SRI is. Moreover, this sub-section explains why investors are interested in the topic. Concluding, as part of the fundamental literature, it defines screening and its implementation strategies.
2.1.1. What is SRI?
The first question raised is what socially responsible investment is. However, there is no single definition to it. Moreover, there is not a single term to describe it. Depending on the emphasis, other terms are used such as “impact investing”, “green investing”, “sustainable investing”, “value-based investing”, or “community investing”.
Different from more conventional types of investment, in socially responsible investment, investors apply several types of screens before investing. Examples of these screens are environmental, social, governmental or ethical screens (Renneboog et al, 2008). A definition for socially responsible investment that is used by Berry and Junkus (2013) is “Integrating personal values and societal concerns with investment decisions”. Benson et al. (2006) explain SRI with the underlying philosophy as a starting point. They state that socially responsible investors do not only take into account financial considerations. These investments also consider non-financial considerations, such as environmental or social matters.
The way how socially responsible investments are used can differ among investors. In his work, Kinder (2005) describes three different types of SRI, that can help explain why SRI is used. According to Kinder, the three categories of SRI are value-based SRI, value-seeking SRI, and value-enhancing SRI. In value-based SRI, the investor tries to align his or her beliefs with its portfolio holdings. The second category is value-seeking SRI. This approach seeks to evaluate firms based on social and environmental criteria which possibly affects the firm’s financial performance and, as a consequence, its share price. The third and final category is value-enhancing
SRI. This category differs in one important aspect: value-enhancing investors reject that they are SRI investors. They adopt different strategies such as “engagement” and “shareholder activism”. These strategies will be explained in section 2.1.4. In other words, these investors aim to improve financial performance by trying to make companies of their interest meet different criteria with respect to either social or ethical aspects. They are often most focused on issues surrounding corporate governance.
2.1.2. Interest in socially responsible investments
There is no single motivation for pursuing SRI. The two main arguments in the current theory in favor of SRI point towards the drive by personal values and goals, whereas the other argument is related to portfolio risk and performance.
A possible explanation for the current increase in socially responsible investment comes from the rising social awareness. According to Renneboog et al. (2008), if certain investors would derive utility from investing in socially responsible stocks, they would be willing to give up more financial performance, than other investors would. On the other hand, Ortlitzky et al. (2003) argue that socially responsible indices are highly correlated with corporate financial performance. This indicates that investors might not give up on financial performance. Also, Benson et al. (2006) mention that the interest in socially responsible funds could arise when these funds target certain criteria that match the individual’s values and beliefs, as well as a possible interest in SRI to obtain abnormal returns. Lastly, Bauer et al. (2005) emphasize that awareness in issues such as civil rights, the environment and nuclear energy can be a cause of the rising interest in SRI.
A different explanation is given by Ballestero et al. (2012) in their book. This book discusses how reality deviates from the classical economic theory as proposed by Adam Smith in 1776. Among these deviations, he explains that the classical model does not incorporate the negative impact economic growth can have on ecosystems, public health and the structure of planetary life. Therefore, according to this book, SRI arose spontaneous, to some extent, when trying to correct for these deviations.
Besides the interest in SRI on an individual basis, individuals and the public opinion push firms to be more aware of their practices with respect to environmental, social and governance
As a result, a rising interest in socially responsible investing of mutual funds can be the outcome of changing demands from individuals and institutions.
2.1.3. Screening criteria
Barnett et al. (2006) defined screening as follows. According to them, screening is the process of either including or excluding certain stocks in an investment portfolio based on certain criteria, such as environmental, social or governance criteria. This way, investors are able to select firms that meet their values and beliefs into their investment portfolios. In the case of SRI funds, the funds select stocks that, based on screening, meet certain predetermined criteria. Section 4.2.3. elaborates on the screening criteria that are used in this study.
2.1.4. Implementation strategies of SRI
There are several strategies to implement socially responsible investment. As well, the strategy definitions differ among researchers. The most often defined strategies are negative screening, positive screening, community investment, and shareholder activism.
Negative screening implies avoiding investing in companies that harm individuals, communities or the environment by their products or business practices. Stocks that meet such criteria are often referred to as ‘sin’ stocks. Examples of sectors where sin stocks are active include alcohol, tobacco, gambling, weapon manufacturers, and sex-related industries. Most of the studies done so far are inconclusive about differences between normal and sin stocks (Trinks and Scholtens, 2017). Even though some researchers, such as Sparkes and Cowton (2004), argue that there is a positive relationship between SRI and stock performance, Hong and Kacperczyk (2009) found that portfolios investing in ‘sin’ stocks outperform comparable portfolios with normal stocks. These results are contradicting, and the latter suggests that screening and excluding ‘sin’ stock does not positively affect performance.
On the other hand, positive screening selects companies that make a positive contribution to society or environment with their products or business practices. Third, community investment helps to bring capital from investors and lenders to communities that are unable to get help from traditional financial institutions. The fourth, and final, strategy explained is shareholder activism. This involves the investors, who can take an active corporate role. One of their efforts in this role
is to take on conversations with companies on social, environmental or governance concerns (Ballestero et al., 2012).
Screening data from the United States Forum for Responsible and Sustainable Investment is used throughout this study. Slightly different types of strategies are defined by the U.S. SIF. Table 1 from the Appendix defines these strategies.
2.2 Effect of SRI on financial performance
Many studies have been done to investigate the effect of investing in a socially responsible manner on financial performance of mutual funds. While some studies investigate socially responsible stocks, others compare SRI funds with conventional mutual funds. Also, a broad range of different models are used.
In the early study by Hamilton et al. (1993), an investigation of American socially responsible funds and conventional funds did not lead to conclusive and significant evidence that the performance of these two differ. They concluded this after investigating Jensen’s alpha using the one factor model.
In their study, Bauer et al. (2005) use an international sample of German, UK and US mutual funds to investigate the performance of ethical funds with respect to non-ethical funds. When using the Carhart multi-factor model and controlling for investment style, they were not able to find significant differences in the period from 1990 to 2001. They do find that older funds do perform better than younger funds. This can indicate that there is a learning effect in SRI.
Kreander et al. (2005) investigated European funds to see if there is a difference in financial performance between ethical funds and non-ethical funds. However, they were unable to find a difference between the two. They did find a significant result that the management fee is an explanatory factor for computing the return.
Nofsinger and Varma (2013) found that socially responsible mutual funds outperform conventional mutual funds during periods of market crises. They, therefore, argue that these SRI funds protect investors against downside risk. Moreover, they find that both funds do not differ with respect to fund characteristics. They conclude this by investigating abnormal returns during normal periods and crises periods using several factor models.
studies. This error arises due to two problems. The first problem is that when computing the financial performance often is controlled for systematic risk, whereas this systematic risk, empirically measured, captures part of the trade-off. The second problem is that collective measures of SRI could mix up other relationships between SRI and returns.
Most of the papers above found that there seems to be no significant difference between SRI fund and conventional funds. However, some papers find contradicting results. Therefore, current literature does not answer which of the two types performs better. The following sub-sections try to explain how under- and outperformances can be achieved.
2.2.1. Explanations for the underperformance of SRI funds
According to Walley and Whitehead (1994), a tradeoff between socially responsibility and financial performance exists. Higher costs will be incurred when companies want to avoid harmful practices, for example, reduce the amount of pollution they produce. These higher operating costs can be an important cause for SRI funds to obtain lower returns.
Darwall et al. (2011) attempt to explain why SRI funds perform different from conventional funds by using the shunned-stock hypothesis. The first assumption that this hypothesis makes is that social investors are value-driven, and care more about non-financial rewards than financial rewards. Moreover, it assumes that the group of investors is large enough to affect stock prices. The shunned-stock hypothesis predicts that when investors shun certain non-ethical stocks, the prices of these investments will deteriorate. As a result, the returns of these stocks go up. This conclusion is based on the theory of Merton (1977) that states that such shun stocks trade at a discount as they have fewer investors, indicating limited risk sharing.
Besides to the shunned-stock hypothesis, socially responsible investment also has additional costs compared to conventional investment. First of all, it is expensive to conduct research in this field and select the most socially responsible firms. Moreover, this data has to be updated occasionally to remain accurate. However, as screening information is more readily available, this cost might not be as relevant as it has been before (Becchetti et al., 2015). Also, it is costly to engage in socially responsible practices, such as environmental improvements or community care. As a result, one can conclude that firms that are operating in a socially responsible manner have a competitive disadvantage with respect to their counterparties that do not incur costs related to socially responsible behavior (Friedman, 1970). When implementing SRI by selecting
stocks that meet certain screening criteria, funds are likely to select firms with above-average high operating costs.
Another disadvantage is that certain screens and implementation strategies exclude entire industries, when they do not meet specific screening criteria. Examples are the gambling industry and the tobacco industry. According to portfolio theory, an investor wants to diversify its portfolio in order to decrease the unsystematic risk. Excluding firms by screening can lead to an inefficient portfolio (Barnett and Salomon, 2006). This can have a negative effect on the financial performance of this portfolio, especially with respect to risk. This implication also holds when just looking at the exclusion of certain stocks. Excluding these stocks limits the diversification benefits and can potentially diminish the portfolio returns as well (Becchetti et al., 2015).
A third disadvantage that is named by Bechetti et al. (2015) is that when stocks are no longer seen as socially responsible, based on the predetermined screening criteria and rules that mutual funds hold, they must be sold. In some cases, these stocks are well performing stocks that are a good addition to the fund’s portfolio. This leads to a disadvantage for the fund with respect to both performance and risk. Furthermore, it increases the fund’s transaction costs if they have to sell frequently.
2.2.2. Explanations for the outperformance of SRI funds
Opposite to the shunned-stock hypothesis by Darwell et al. (2011), the same paper also proposed the errors-in-expectations hypothesis. This hypothesis states that the market is not able to value socially responsible practices correctly. The first reason that Darwell et al. (2011) give for this is that it is a subjective concept, and investors are not able to price it correctly. The second reason is that accounting standard are not adapted to socially responsible practices and, as much of the value created by it is intangible, these businesses are not priced correctly. The final reason suggests that negative social practices get more attention than positive social practices, and therefore, are not valued enough. Concluding, the errors-in-expectations hypothesis predicts that SRI funds are able to provide investors with superior returns as the market does not value socially responsible according to its actual value.
others. As a result, these firms can obtain a better financial performance. Moreover, social relationships tend to influence performance as well. Proponents of socially responsible investments, therefore, argue that socially responsibility can be seen as an investment for future financial performance.
As previously explained, socially responsibility often implies incurring higher costs. Walley and Whitehead (1994) argue that a trade-off exists between socially responsibility and financial performance. On the other hand, the paper by Stanwick and Stanwick (1998) suggests that reverse causality can exist in this situation. Companies that obtain higher returns, often have more money on hands and might be more willing to spend it on improving their business and reducing harmful practices. Where, on the other hand, businesses without this money are not as able to pursue socially responsibility. Moreover, Porter and van der Linde (1995) propose that due to the different approach and possibilities for innovation, new technologies can be developed leading to a comparative advantage for SRI. This implies a positive relationship between socially responsibility and financial performance. However, as the paper is dated, it is possible that the comparative advantage has disappeared.
Furthermore, Fombrun and Shanely (1990) explain that a high performance with respect to environment can lead to an improved reputation. As a result, this can improve the firm’s cash flows as it has easier access to cash. Using a similar argumentation, doing well with respect to society and governance could also benefit the firm’s reputation and, in effect, improve financial performance. A similar argumentation comes from Renneboog et al. (2008) who emphasized that a good environmental and social performance signals stability and good management. This in return can lead to improve financial performance. According to Graves and Waddock (1994), an outperformance can also be due to the fact that socially responsible mutual funds are often more “innovative and growth minded”. This can have a positive effect on financial performance. The effect of socially responsible investment depends on whether the shunning-stock hypothesis or the errors-in-expectation hypothesis holds (Darwell et al., 2011). Moreover, it matters if the other costs outweigh the benefits or the opposite way. This thesis will investigate the financial performance of portfolios based on screening more elaborately. However, utility as a benefit of socially responsible investment is not considered here. Only objective financial performance is evaluated during this study.
2.2.3. Similarities between SRI funds and conventional funds
The amount of SRI funds has been rapidly increasing over time. One of the reasons for this is the increased interest in socially responsibility, and the public opinion that is pushing funds towards increased socially responsibility. Laurel (2011) claims that socially responsibility moves towards being the norm, rather than a choice by the fund. This suggests that SRI funds and conventional funds will converge towards each other over time. Crifo and Mottis (2010) are supporters of this suggestion. They hypothesize that screening will become a normality.
Empirical evidence of the similarities between SRI funds and conventional funds is found by Bello (2005). As one argues that SRI can be a disadvantage due to the loss of diversification benefits, Bello found that SRI funds and conventional funds can diversify equally well. Therefore, investing in a socially responsible manner does not lead to significant loss of diversification benefits according to this paper. Moreover, according to Nofsinger and Varma (2013), SRI funds and conventional funds do not differ with respect to fund characteristics. These theories suggest that SRI funds and conventional funds are similar and, resulting, these funds will obtain similar financial performance.
Concluding, there are many different theories and papers on the differences and similarities in performance, as well in characteristics, between SRI funds and conventional funds. However, they all investigate a different sample over a different period of time. Therefore, this study will investigate if there are performance differences over time. This can be a reason why results differ when investigating different time periods.
2.2.4. Fund characteristics
This research aims to explain the effect and the development of fund characteristics on fund performance of socially responsible mutual funds over time. The fund characteristics that will be investigated during this thesis are the expense ratio, fund size, fund age, and the screening intensity. As mentioned above, Elton et al. (2012) suggest that when fund size increases, the expense ratio goes down. This suggests that these two are negatively related to each other. This also indicates that it is possible that when fund size increases and the expense ratio goes down, firms perform better. However, on the other hand, Berk and Green (2004) argue that when a fund is
According to Elton et al. (2012), fund size should have a positive effect on financial performance. A first argument for this, is that when the fund size increases, the expense ratio goes down. Moreover, they argue that large funds have lower transaction costs and have excess to better traders than smaller funds. This can positively affect financial performance. On the other hand, Chou and Hardin (2014) find that an increase in fund size has a negative impact on the financial returns of real estate mutual funds. Lastly, Huang and Mahieu (2012) did find significant effect that large pension funds outperform smaller pension funds. Therefore, the results are not in accordance.
Barnett and Salamon (2006) find that firms that comply with more screening criteria and are selected based on those criteria, tend to be more stable firms. They are also better managed. This could indicate that high screening intensity could positively affect financial performance. On the other hand, Lee et al. (2010) find that screening intensity has no effect on the unadjusted returns. However, it seems that alpha decreases by 70 basis points per screen when using the Carhart model. A possible explanation can be that more screening results in less risk, which in turn results in a lower return. Moreover, Humphrey and Lee (2011) specify their research to Australian socially responsible funds. They find that there is only weak evidence that higher screening intensity provides a better risk-adjusted return.
In their study, Bauer et al. (2005) found that older funds outperformed younger fund. This indicates that age might be important in determining fund performance, and that there could be a learning effect. A different research by Cowling et al. (2018) suggests that during the crisis, young business kept on growing more than old businesses. They also recovered more quickly. This reveals that, possibly, even when old funds perform better, young funds might still perform better during crisis periods.
2.3 Hypotheses
Based on the literature, this sub-section derives the hypotheses. As a first step, the financial performance will be assessed. The current stance of literature does not give a decisive answer on whether SRI funds outperform conventional funds, or the other way around. This leads to the following hypothesis:
Additionally, this discussion moves further as Laurel (2011) and Crifo and Mottis (2010) claim that SRI funds and conventional funds will converge towards each other. This is due to the fact that they argue that socially responsibility will become a normality. The next hypothesis can be derived on that statement:
H2: The difference in financial performance between SRI funds and conventional funds has
become smaller over time.
These two hypotheses are the main hypotheses of this study and will be examined in order to draw conclusions on the relative performance between SRI funds and conventional funds. Moreover, as a last part of this thesis, the fund characteristics are studied. The current literature gives explanations for both a high expense ratio being related to high performance as a low expense ratio being related to high performance. Therefore, the following hypothesis is tested:
H3: Funds with a high expense ratio have a different financial performance than funds with
a low expense ratio.
The second characteristic studied is size. Literature points towards large mutual funds outperforming small mutual funds (e.g. Elton et al., 2012). Therefore, the following hypothesis is tested:
H4: Large SRI mutual funds outperform small SRI mutual funds.
A similar argumentation can be used for age. Current literature suggests that a learning effect (Bauer et al., 2005) exists, and that older funds perform better than younger fund. This lead to the following hypothesis:
H5: Older SRI funds outperform younger SRI funds.
The final theory tested is on screening intensity. Highly screened firms tend to be more stable, according to Barnett and Salomon (2006). This suggests that more screening can improve fund performance. The following hypothesis is tested:
H6: Highly screened SRI funds outperform low screening SRI funds.
3. Methodology
This section will extensively describe the methodology of this study. Various tests will be performed in order to answer the research question. The first part explains the fund set-up for this analysis. The second part of the methodology elaborates on performance tests for the financial performance of the SRI fund and the conventional fund. The third part of the methodology explains how the fund characteristics will be investigated.
3.1. Fund set-up
First, the socially responsible and the conventional fund have to be set up. As adopted from other works, such as Becchetti et al. (2015), an equally weighted ‘superfund’ of the SRI funds is created. In a similar process, an equally weighted superfund of conventional funds it created. The monthly returns of both these funds are calculated by taking the equally weighted average of all the monthly returns of the funds included in the superfunds (socially responsible mutual funds or conventional mutual funds, respectively). The fund is generated as follows:
ri,tsuperfund = 1 Nt∑
N j=1 rj,t
The monthly returns per firm j at time t are summed up and divided by the total number of observations at every time t. This generates an average return per month. This process is done separately for the SRI funds and the conventional fund. As a result, two superfunds are generated.
3.2. Financial performance
The first way to measure financial performance in this study is the Sharpe ratio. Moreover, a spectrum of models is regressed in order to evaluate the abnormal returns. The following paragraphs explain how these tests are performed.
3.2.1. Sharpe ratio
A way to measure the risk-adjusted performance of the portfolios is by computing the Sharpe ratio (Sharpe, 1994). This ratio can determine the attractiveness of an investment by accounting for the risk-free rate and the standard deviation of the excess returns. The Sharpe ratio is given by the following formula:
S = Rp− rf
σp
In words, the Sharpe ratio is computed by taking the return of the fund minus the risk-free rate, and dividing this excess return by the standard deviation of the excess return. In this study, the Sharpe ratio will be computed over several time periods for both the SRI fund and the conventional fund to investigate how the risk-adjusted performance developed over time.
3.2.2. Regression models
Secondly, for both the superfunds (SRI superfund and the conventional superfund) the CAPM (Markowitz, 1952), the Fama and French three-factor model (1993), the Carhart (1997) four-factor model, and the five-factor model by Fama and French (2015) are estimated. The equations for the time-series analyses are as follows:
(1) CAPM: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + εi,t
(2) Three-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + εi,t
(3) Four-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + εi,t
(4) Five-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5irtRMW + β6irtCMA +εi,t
Here, ri,t is the return of the superfund and rf,t will be the monthly risk-free rate. rmt – rf,t is defined
as the excess market return. rtSMB, where SMB is short for Small Minus Big, presents the spread
between returns from firms with a low market capitalization and high market capitalization. rtHML,
where HML is short for High Minus Low, presents the spread between returns of firms with a high book-to-market ratio and a low book-to-market ratio. rtMOM, where MOM is short for momentum,
presents the spread between stocks that had a low return and stock that had a high return over the past twelve months. rtRMW, where RMW stands for Robust Minus Weak, presents the spread
between returns for portfolios with robust operating profitability and portfolios with weak operating profitability. rtCMA, where CMA is short for Conservative Minus Aggressive, is the
average return of a conservative portfolio minus the average return of an aggressive portfolio. The alpha, ⍺i, presents the abnormal return of the fund often termed as Jensen’s alpha (1969).
After the results have been obtained, the alphas of the different funds are compared to see if there are significant differences either between the funds. This can be done by performing a t-test on the abnormal returns that are found. The t-statistic is calculated in the following way:
t-statistic = SE(ɑɑSRI− ɑCONV SRI−ɑCONV)
SE(ɑSRI - ɑCONV) = √SE(ɑSRI)2+ SE(ɑCONV)2
The alphas, ɑSRI and ɑCONV respectively, represent the abnormal returns of the SRI superfund and
the conventional superfund. SE(ɑSRI - ɑCONV) presents the combined standard error of both funds.
The abnormal returns are not only computed for the entire sample period. The sample period will also be split into two sub-periods: the first sub-period runs from January 1st 1988 to December
31st 2002, and the second sub-period is from January 1st 2003 to December 31st 2017. For both
sub-periods and both superfunds, the abnormal returns will be estimated using model (1), (2), (3) and (4). Using the same formula to compute the t-statistic as is explained above, this study will determine if the abnormal returns differ significantly over time.
Besides running these regressions separately for different time periods and for both SRI superfund and the conventional superfund, the difference is return will also be assessed. The following regressions will be performed:
(5) rtSRI - rtCONV = ⍺i + β1i(rmt – rf,t) + εi,t
(6) rtSRI - rtCONV = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + εi,t
(7) rtSRI - rtCONV = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + εi,t
(8) rtSRI - rtCONV = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5irtRMW + β6irtCMA + εi,t
Again, the abnormal returns, as presented by the alpha, will be investigated. This step is performed to assess the relative differences between the SRI fund and the conventional fund. Moreover, it provides the data needed to investigate if SRI fund and conventional funds have become more similar over time or not.
3.2.3. Fund characteristics
value are included in the group “High” and values below the 25% percentile value in the group “Low”. This is done separately for each year to balance the samples over the entire time period. The observations in both groups are put in an equally weighted superfund, which creates the funds “High” and “Low”. For every characteristic, the following three regressions, based on the Carhart (1997) four-factor model, are performed:
(9) rt,I,HIGH = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + εi,t
(10) rt,I,LOW = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + εi,t
(11) rt,i,HIGH – rt,I,LOW = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + εi,t
These regressions are performed for the entire sample from January 1st 1988 to December 31st
2017, and for the two sub-periods from January 1st 1988 to December 31st 2002 and from January
1st 2003 to December 31st 2017. The abnormal returns are investigated across time and across
groups. The fund characteristics that will be assessed are the expense ratio, fund size, fund age and screening intensity.
3.3. Robustness checks
The first part of the robustness checks is included in the methodology on the financial performance. As this study will investigate the abnormal returns using four different models, the robustness of the results will be tested at the same time. If the results among the four models are similar, the results on the financial performance are robust.
The fund characteristics are investigated by using only one model. Therefore, as a robustness check, this will be repeated by using the CAPM model, the three-factor model and the five-factor model as well. Likewise, if the results do not differ much among models, we can argue that the results found are robust.
As a third part of the robustness checks, a socially responsible index is included in the regressions. Renneboog et al. (2007) state that underperformance of socially responsible funds can be driven by the exclusion of an ethical risk factor. Therefore, potentially, the abnormal returns can be improved when accounting for ethical risk. This can be investigated by adding an index to proxy
(12) CAPM: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtethical + εi
(13) Three-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtethical + εi
(14) Four-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5irtethical + εi
(15) Five-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5irtRMW + β6irtCMA +
β7irtethical + εi
Here, the variable rtethical presents the monthly return of the ethical index, in this case, MSCI KLD
400 Social Index. This index is chosen as it is the longest running ethical index available1.
Another robustness check that will be performed is based on the study by Bauer et al. (2005). This robustness check investigates if the management fee affects the relative performance between the SRI fund and the conventional fund. This is interesting as it is often claimed that SRI funds are more expense than conventional funds. Moreover, Kreander et al. (2005) suggest that the management fee is a significant explanatory factor for computing return. After adding back management fees2 to the return, the difference portfolio is generated. The results of the portfolio
without management fees and with management fees added back are compared to see if this influences the results. Regressions (5), (6), (7), and (8) will be regressed for both the fund without management fees added back and with management fees added back.
Finally, a robustness check based on the market timing ability by Treynor and Mazuy (1996) is performed. As explained by Chen and Stockum (1986), one assumption that has to hold in order to be able to time the market, is that the market beta is non-stationary. Therefore, including the market timing factor can determine if beta is stationary or not. This is tested by including the market timing ability in the four regression models that are used throughout this study:
(16) CAPM: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2i(rmt – rf,t)2 + εi
(17) Three-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4i(rmt – rf,t)2 + εi
(18) Four-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5i(rmt – rf,t)2 + εi
(19) Four-factors: ri,t - rf,t = ⍺i + β1i(rmt – rf,t) + β2irtSMB + β3irtHML + β4irtMOM + β5irtRMW + β6irtCMA +
β7i(rmt – rf,t)2 + εi
1 The MSCI KLD 400 Social Index was formerly known as the Domini 400 Social Index and was incepted on May
These regressions are performed for both the SRI fund and the conventional fund. The market timing factor is presented as (rmt – rf,t)2. If the beta for this factor is significant, the market timing
ability of the fund manager significantly affects the returns. In that case, the market beta is non-stationary and we cannot assume that the beta in our regression is the same for the entire estimation period.
4. Data
This thesis will focus on both SRI funds and conventional funds in a period from January 1st 1988
to December 31st 2017. Therefore, all data is obtained for this time period. The databases used are
the United States Forum for Sustainable and Responsible Investment (US SIF), Center for Research of Security Prices (CRSP) via Wharton Research Data Services (WRDS), the Kenneth R. French Data Library and Datastream. The following parts explain how the data is collected and presents the descriptive statistics.
4.1 Data collection
This section describes how the monthly returns, fund summaries, market return benchmark, factor benchmarks, the risk-free rate benchmark and the screening data is obtained and what database is used access the data.
4.1.1 Fund statistics
Using the CRPS Survivor Bias-Free U.S. Mutual Funds database, accessed via WRDS, the monthly returns of all U.S. mutual funds are obtained. An often found problem in datasets is the survivorship bias, where mutual funds that perform bad cease to exist. This can bias the results. However, this database is survivor bias-free. Here, no difference is made between returns from SRI funds and conventional funds, yet. The returns are readily available and are computed by CRSP using the following formula:
rt = NAVNAVt∗cumfact
t−1 – 1
Here, NAV presents the net asset value and the cumfact is a cumulative factor taking into account an adjustment factor based on distributions and reinvestments or the split ratio and the cumfact of the previous period.
Also, using the CRPS database, the fund summaries of all mutual funds are obtained. Variables that are obtained here are the ticker, the company name, expense ratio, management fee, inception date, the monthly total net assets, and the turnover ratio. These annual statistics can be merged into the monthly return file.
4.1.2. Risk-free rate and market return benchmark
To proxy the market return, the average monthly return of the NYSE, NASDAQ and AMEX indices is obtained using the CRPS database via WRDS. For the risk-free rate, the monthly return of a 30-day T-bill will be used.
4.1.3. Factor benchmarks
In order to perform regressions of Fama and French three-factor model (1993), the four-factor model by Carhart (1997), and the five-factor model by Fama and French (2015) proxies for the factors are needed. Here, data from the Kenneth R. French Data Library are used. In this database, benchmarks for the SMB (small minus big), HML (high minus low), the MOM (momentum), RMW (robust minus weak profitability), and CMA (conservative minus aggressive) factor are readily available and can be obtained as monthly returns for the required time period.
One additional factor will be used in order to perform robustness checks. This factor is used to represent ethical risk and will be proxied by the MSCI KLD 400 Social Index. This index, previously known as the Domini Social 400 index, can be obtained using Datastream. This benchmark is obtained for the period from 1990 to 2017, as it was incepted on May 31st 1990.
4.1.4. Screening data and fund set-up
As screening is relatively new, less data is available than for some other statistics. Moreover, not all the databases with ESG screening information are accessible. For example, there is no public access to the Bloomberg Environmental Social and Governance (ESG) data service. Additionally, the U.S. SIF is not able to provide their historical screening information on U.S. mutual funds. However, some data is available. From the U.S. SIF, we are able to retrieve a list of SRI mutual funds as of March 3rd 2018. These include all fund types (equity, bond, balanced, and
international global foreign). As several mutual funds provide the same fund in different share classes, only one share class per fund is remained3. The main reason for this is that including them
all leads to duplicates with respect to screening strategies. Moreover, only equity mutual funds are
3 Some mutual funds provide the same fund in different share classes. They differ with respect to cost structure. For
used in this study. 67 SRI equity mutual funds remain after implementing this strategy. Section 4.2.3. will extensively description the screening data in general.
4.2. Descriptive statistics
This section presents the descriptive statistics of the data used in this study. Additionally, this section aims at giving an understanding of the data that is used in this research. Furthermore, in the data preparation, all variables are inspected on outliers. If necessary, the variable is winsorized at the 1% level.
4.2.1. Mutual funds
First, the average characteristics of all SRI funds and all conventional funds will be given. This is followed up by some notes on the correlation between the variables of these groups.
Table 1: Fund Characteristics
This table presents the monthly return in percentages, the monthly standard deviation in percentages, the average fund age in years, the average management fee in percentages, the average expense ratio in percentages, the average turnover ratio in percentages and the total net assets in millions of dollars over a period of January 1st 1988 to December 31st
2017 of all SRI funds in the sample investigated. Screening presents the average number of screens that the mutual funds use when selecting their stocks per fund type.
SRI Conventional Difference
Monthly return 0.697% 0.5876% 0.104% SD return 4.289% 5.062% -0.773% Fund Age 10.393 7.889 2.504 Management Fee 0.553% 0.561% -0.007% Expense Ratio 1.156% 1.268% -0.112% Turnover Ratio 46.985% 67.834% -20.849%
Total Net Assets 185.822 141.749 44.073
Screening 13.800 - -
Several insights are given when comparing the fund characteristics of the SRI funds and the conventional funds. First of all, the SRI funds seem to provide higher monthly returns than the
have a lower standard deviation. Another observation is that the average fund age of SRI funds exceeds that of conventional funds. This is against expectations, as the SRI industry is young and developing. However, possibly, many new conventional funds have been established, where bad performing ones have ceased to exist. The management fees are very similar, where on the contrary, the expense ratios are less similar. Again, against expectations (Walley and Whitehead, 1994), the average expense ratio of the SRI funds is lower than the average expense ratio of the conventional funds. Moreover, the average turnover ratio of the SRI funds is 46.985%, whereas it is 67.834% for conventional funds. This indicates that conventional funds may sell and buy assets more often than SRI funds. This suggests that SRI funds might not suffer from above average transaction costs, as suggested by Bechetti et al. (2005). Finally, the average total net assets of SRI funds exceeds that of the conventional funds.
The correlation matrices for the important variables for both the SRI funds and the conventional funds can be found in Table 4 and Table 5 of the Appendix. The correlation matrix for the SRI funds provides us with a few interesting findings. First of all, the correlation between the return and the expense ratio is 0.0086. This indicates that a higher expense ratio is related to higher returns. On the other hand, the correlation between return and the expense ratio for conventional funds is -0.0117 and contradicting the correlation of the SRI funds. A similar result is found for the management fee. The correlation between the return and the management fee for SRI funds is -0.0156, where the correlation is 0.0069 for conventional funds. The correlations between returns and size, age and the turnover ratio are in the same direction. Size and age do have a stronger correlation with return for the conventional funds than for the SRI funds. A final important implication is the correlation of -0.0193 between the total number of screens and the return. This is different than what would have been expected, but will be studied further on. As explained in the methodology, this study investigates two superfunds, the first being a SRI fund and the second a conventional fund. In order to establish the SRI superfund, the socially responsible mutual funds as determined by the US SIF are used4. Only equity mutual funds are
used in this study. As a result, 67 mutual funds remained after the selected process. The conventional superfund uses the non-responsible equity funds from the CRSP database. This results
in a sample of 36,7515 conventional mutual funds that are used to generate the conventional super
fund.
The following graph clearly shows how the number of SRI funds has rapidly increased. In 1988, the first year of this study, only four SRI funds matching our criteria existed. This number increased to 67 in 2017.
Graph 1: Increasing number of SRI funds
Graphical representation of the development of the number of SRI funds. The horizontal axis presents the year and the vertical axis presents the number of SRI funds.
4.2.2. Returns
The following graph, Graph 2, presents the returns of the risk-free rate (proxied as the monthly return of a 30-day T-Bill), the market return (proxied as the average return of the NYSE, NASDAQ and AMAX indices), and the SRI return (proxied as an equally weighted average of all SRI funds).
Graph 2: Monthly returns
This graph presents the monthly returns in percentages in the time period from January 1st 1988 to December 31st 2017.
The returns shown are the risk-free rate, proxied by the monthly return on a 30-day T-Bill, the average return of the 67 SRI mutual funds used in this study and the market return, proxied by the average return of the NYSE, NASDAQ, and AMAX indices.
Over time, the market return and the SRI return have deviated much from the risk-free rate. However, the market return and SRI return remained close and show similar movements. As can be seen from this graph, the average return of the SRI funds tends to be more extreme. The extremes are both higher and lower. This indicates that SRI funds might be less stable than the market over the last thirty years. The risk-free rate remained stable and moved towards 0% over time.
4.2.3. Screening data
As described, the screening data is only available as of March 3rd 2018. The U.S. SIF is not able to
provide historical information. Therefore, we assume that this data has not changed. This however, can lead to a look-back bias where we assume that all data is similar in the past. This is obviously not true. The assumption that it is similar can bias the results. This will be discussed later on in
The US SIF (2018) separates screening policies into different strategies. These are summarized in Table 1 of the Appendix. For every SRI fund, the US SIF screened the fund based on the predetermined screening criteria that are organized in separated categories. The first category is environment, the second category is social, the third category is governance, and the fourth and final category is product. They also determined two additional criteria. Table 2 of the Appendix provides an extensive description of all screening categories. It explains what screens are included in every category and summarizes the literature on these categories. Finally, Table 3 of the Appendix shows the distribution of the strategies over the different screening criteria. The table gives a clear overview of the main strategies that are used per screening criteria. For example, an often used strategy for the screens climate, environment, human rights, and labor is Combination. This strategy combines positive investment with restrictive investments. The screens alcohol, animal, defense, gambling, and tobacco are often linked to the strategy Restricted Investment.
Next, the descriptive statistics from the screening data will be shown. The following table shows how many of the funds use certain screening criteria for their funds. Here, the second column presents the number of funds in the sample that use the screen. The third column shows the percentage of mutual funds using the screen.
Table 2: Screening criteria
This table presents how many of the 95 SRI Funds in our sample screen based on the different categories. The first column presents the screen, the second column how many funds screen their investments based on the selected screen, and the third and final column presents the percentage of funds using the selected screen. Information is based on the screening information provided by the US SIF as of March 1st, 2018.
Number of funds using this screen
Average number of firms using this screen Climate/Clean Tech 58 0.817 Pollution/Toxics 57 0.803 Environment/Other 69 0.972 Environment 0.864 Community Development 65 0.915
Diversity & EEO 66 0.930
Human Rights 65 0.915 Labor Relations 64 0.901 Conflict Risk 37 0.521 Social 0.836 Board Issues 58 0.817 Executive Pay 50 0.704 Governance 0.761 Alcohol 58 0.817 Animal Welfare 53 0.746 Defense/Weapons 67 0.943 Gambling 57 0.803 Tobacco 68 0.958 Products 0.853 Other/Qualitative 24 0.338 Shareholder Engagement 52 0.958 Total 0.802
Based on percentages, products is the most used category, followed by environment, social and governance. Conflict risk and other/qualitative are the least used screens, whereas environment/other, defense/weapons and tobacco are the most often used screens by the mutual funds in this sample. On average, the SRI funds in this study use almost 80% of the screening criteria set up by the US SIF to screen their investments.
include very related screens. For example, the correlation between the screen for climate and the screen for pollution is 0.900. Similarly, the correlation between the screen for human rights and the screen for labor rights is 0.943 and that for the screen board issues and the screen executive pay 0.800. The screens conflict risk and tobacco are negatively related to most of the other screens.
5. Results
This section presents the results. The first sub-section presents the main results with respect to the financial performance. This section is followed up by the robustness checks of this study.
5.1 Financial performance
As the methodology explains, this study investigates the financial performance using different measures. First, the Sharpe ratio is computed. Second, several regression models are estimated and finally, the fund characteristics are investigated.
Graph 3: Moving average of abnormal returns
The following graph presents the average abnormal returns both SRI funds and conventional funds. The abnormal returns are estimated for a moving five-year time period using the Carhart (1997) four-factor model.
As the graph shows, the SRI funds seem to outperform the conventional funds. Moreover, it seems like the abnormal returns have become more similar over time. However, this result does not hold for every time period and this graph only is not sufficient to make any conclusions. Therefore, next, the Sharpe ratios are computed following by the estimation of the regression models.
5.1.1. Sharpe ratio
Table 3: Sharpe ratio
Sharpe ratios per fund over different time periods. The final column presents the difference between the Sharpe ratios of the SRI fund and the conventional fund.
SRI fund Conventional fund Difference
1988-1992 0.1982 0.1945 0.0037 1993-1997 0.2662 0.2784 -0.0122 1998-2002 -0.0300 -0.0690 0.0390 2003-2007 0.2881 0.3600 -0.0719 2008-2012 0.0497 0.0506 -0.0009 2012-2017 0.3537 0.3263 0.0274 Entire sample 0.1384 0.1301 0.0083
As the table indicates, the Sharpe ratios of both funds are close in every sub-period. In three out of six sub-periods, the SRI funds obtained a higher Sharpe ratio. On the other hand, the remaining three sub-periods indicate that the conventional fund was able to obtain a better risk-adjusted performance. Over the entire sample, the Sharpe ratio of the SRI fund exceeds that of the conventional fund. This can be seen by the ratios of 0.1384 and 0.1301, respectively. This indicates that the SRI fund performed better with respect to risk-adjusted return. However, the results show that the SRI fund does not perform better in each sub-period, and the difference in the Sharpe over the entire sample is small.
5.1.2. Regression models
Looking at the Sharpe ratio only is not sufficient to determine how the funds performed relatively. This sub-section presents the results from the regression models, and investigates the abnormal returns.
Table 4: Regression results
This table presents the main regression results for the CAPM, three-factor, four-factor and five-factor model. The
sample period runs from January 1st 1988 to December 31st 2017 and the results for both the SRI superfund and the
conventional superfund are shown. The t-statistics are in the parentheses.
CAPM Three-factor Four-factor Five-factor SRI CONV SRI CONV SRI CONV SRI CONV
Market risk premium 1.0775*** (41.27) 1.0780*** (47.11) 1.0702*** (40.92) 1.0500*** (46.46) 1.0248*** (40.30) 1.0307*** (44.65) 1.0374*** (37.29) 0.9935*** (39.60) SMB 0.1166*** (4.23) 0.1422*** (5.97) 0.1221*** (4.71) 0.1446*** (6.14) 0.1562*** (5.49) 0.1156*** (4.50) HML 0.0868*** (3.06) -0.0117 (-0.48) 0.0409 (1.49) -0.0309 (-1.24) 0.0376 (0.98) 0.0432 (1.25) MOM -0.1206*** (-6.95) -0.0506*** (-3.21) -0.1227*** (-6.98) -0.0398** (-2.51) RMW 0.0952** (2.55) -0.1063*** (-3.15) CMA -0.0473 (-0.89) -0.1056** (-2.19) Constant -0.0008 (-0.96) -0.0013* (-1.80) -0.0011 (-1.37) -0.0014* (-1.91) -0.0001 (-0.19) -0.0010 (-1.33) -0.0005 (-0.57) -0.0003 (-0.41) Observations 360 360 360 360 360 360 360 360 Adjusted R2 0.8258 0.8607 0.8365 0.8729 0.8557 0.8762 0.8581 0.8799
*Significant at 10% level. ** Significant at 5% level. * Significant at 1% level.
Table 3 presents the results for the four different regression models. As can be seen, the results are very similar. For both funds and all models, the funds underperform the market. For the conventional fund, this underperformance is significant at the 10% level when looking at both the CAPM model and the three-factor model. For the SRI fund, this underperformed is not significant for any of the models. Table 4 gives an overview of the abnormal returns of the different funds and models over different time periods, in order to compare the financial performance over time.
Table 5: Abnormal returns
Overview of the monthly abnormal returns in percentages. For each model, the abnormal returns for the SRI fund, the conventional fund and the difference fund is presented. The difference is obtained by regressing the difference in return of the SRI fund and the conventional fund (ri,tSRI – ri,tCONV). The entire sample period runs from January 1st 1988 to
December 31st 2017. The last row presents the differences between the period from January 1st 1988 to December 31st
2002 and the period from January 1st 2003 and December 31st 2017.
CAPM Three-factor Four -factor Five-factor
SRI CONV Diff SRI CONV Diff SRI CONV Diff SRI CONV Diff
Entire sample -0.0821 (-0.96) -0.1349 (-1.80) 0.0528 (0.94) -0.1140 (-1.37) -0.1372* (-1.91) 0.0231 (0.42) -0.0151 (-0.19) -0.0957 (-1.33) 0.0805 (1.53) -0.0460 (-0.57) -0.0303 (-0.41) -0.0157 (-0.32) 1988-2002 -0.0022 (-0.02) -0.0965 (-0.92) 0.0935 (0.90) -0.0662 (-0.50) -0.0715 (-0.73) 0.0052 (0.05) 0.1092 (0.85) -0.0462 (-0.45) 0.1554* (1.68) 0.0421 (0.32) 0.000 (0.01) 0.0415 (0.47) 2003-2017 -0.1915* (-1.82) -0.1966* (-1.85) 0.0080 (0.20) -0.1871* (-1.80) -0.1944* (-1.84) 0.0074 (0.18) -0.1623* (-1.67) -0.1751* (-1.72) 0.0128 (0.32) -0.1262 (-1.30) -0.1081 (-1.08) -0.0182 (-0.45) Diff time 0.1894 (1.12) 0.1000 (0.67) -0.0855 (0.77) -0.1208 (0.72) -0.1230 (0.85) 0.0021 (0.02) -0.2715* (1.70) -0.1289 (0.89) -0.1426 (-1.42) -0.1683 (1.04) -0.1081 (-0.75) -0.0597 (-0.64) *Significant at 10% level. ** Significant at 5% level. * Significant at 1% level.
As can be seen from the table, almost all the abnormal returns that are obtained are negative. For the SRI fund, the abnormal returns for the entire sample range from -0.0151% to -0.1140% per month. For the period from 1988 to 2002 this range is from -0.0662% to an outperformance of 0.1092% per month. The final period, 20032017, shows an underperformance ranging from -0.1262% to -0.1915% per month. The underperformance for the CAPM model and the three-factor model is significant in that period.
The conventional fund also presents underperformance in most of the cases. For the entire sample, all abnormal returns are negative and for the three-factor model, a significant underperformance of 0.1372% is found. From 1988-2002, no significant under- or outperformances are obtained. For the period from 2003 to 2017, three out of four models find that the conventional fund underperformed the market significantly.
The third column per model presents the results for the regressed difference. For the CAPM model, these results are all positive. This indicates that the SRI fund performed better than the conventional fund. However, the results are not significant. The same holds for the three-factor model. However, the four-factor model finds that in the period from 1988 to 2002, the SRI fund outperformed the conventional fund significantly by 0.1554%. The five-factor model also finds that the SRI fund outperformed the conventional fund over the sub-period from 1988-2002, by
0.0415%. However, over the entire sample, this model finds that the conventional fund performed better. This result is not significant.
The final row presents the differences over time. The most important results are the differences over time for the difference fund. Even though these results are not significant, they signal that the difference between SRI funds and conventional funds has become smaller. A possible explanation can be that the SRI funds generated above average returns in their development stage and got more stable over time. Furthermore, the results are in line with Laurel (2011) that argued that socially responsibility will be the norm. In that case, SRI funds and conventional fund might become more similar over time.
5.1.3. Fund characteristics
This section presents the results on the relationship between the fund characteristics and the financial performance of SRI mutual funds. The fund characteristics studied are the expense ratio, fund size, fund age, and screening intensity. The results are presented in Table 6.
Table 6: Financial Performance of Fund Characteristics
Overview of the monthly abnormal returns in percentages. For each characteristic, it separates the results for group “High”, group “Low” and the difference group (return group high minus return group low, rtHIGH - rtLOW) in the columns. The rows present different (sub-)periods, where the final row presents the difference over
time between the two sub-periods.
Expense ratio Size Age Screening intensity High Low Diff High Low Diff High Low Diff High Low Diff
Entire sample -0.1491 (-1.38) -0.0232 (-0.28) -0.1259 (-1.09) 0.0407 (0.47) -0.0115 (-0.14) 0.0523** (2.09) 0.0243 (0.24) 0.0256 (0.26) -0.0013 (-0.01) 0.0370 (0.33) -0.1491 (-1.29) 0.1861 (1.41) 1988-2002 0.1042 (0.49) 0.0728 (0.52) 0.0314 (-0.34) 0.2117 (1.46) 0.1153 (0.90) 0.0964** (2.02) 0.1239 (0.72) 0.2736* (1.66) -0.1498 (-0.78) 0.1946 (0.83) 0.00965 (0.04) 0.1850 (0.66) 2003-2017 -0.2713** (-2.54) -0.1326 (-1.50) -0.1379** (-2.17) -0.1487 (-1.50) -0.1576* (-1.69) 0.0089 (0.50) -0.1395 (-1.22) -0.2086** (-1.99) 0.0690 (1.19) -0.0518 (-0.54) -0.2634** (-2.08) 0.2116** (2.15) Diff -0.3755 (-1.57) -0.2054 (-1.24) -0.1693 (-1.40) -0.3604** (-2.05) -0.2729* (-1.70) -0.0849 (-1.72)* -0.2634 (-1.27) -0.4817*** (-2.47) 0.2188 (1.09) -0.2465 (-0.97) -0.2731 (-1.09) 0.0266 (0.09) *Significant at 10% level. ** Significant at 5% level. * Significant at 1% level.
