Socially Responsible Investing
June 2020
BSc Economics and Business Economics - Finance Major
6 EC
“The effect of screening intensity on the risk-adjusted return of socially
responsi-ble investment funds: A study based on the MSCI All Country World Index”
Dave Grünewald – 11584793
Supervisor: Drs. P.V.Trietsch, M.Phil.
Abstract
This research paper analyses the influence of screening intensity on socially responsible
invest-ment funds, while holding all variables except screening intensity constant. The CAPM, Sharpe
ratio and graphical analysis, are divided over two different datasets, with one dating back to the
introduction day of each fund and the other one starting on the first day when all of the funds
coexisted. The analysis does not find any significant evidence to refute a curvilinear
relation-ship between screening intensity and risk-adjusted returns and shows a trend towards
overper-formance of funds that exclude certain controversial sectors and the most social funds, which
might have been a result of the booming sector of socially responsible investing.
Statement of Originality
This document is written by Dave Grünewald, 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
super-vision of the completion of the work, not for the contents.
Table of content
1. Introduction ... 1
2. Literature review ... 3
2.1 The definition of SRI and its relevance... 3
2.2 Social responsibility and financial performance ... 4
2.3 Screening intensity and fund performance ... 6
2.4 Measuring screening intensity ... 8
2.4.1 Number of screens ... 8
2.4.2 Type of screens ... 9
2.4.3 SRI Ratings ... 9
2.5 Methodologies of past studies ... 9
2.6 Hypotheses ... 12 3. Data ... 13 3.1 Sampling ... 14 3.2 Categorization ... 16 4. Methodology ... 17 4.1 Sharpe ratio ... 18
4.2 CAPM and Jensen’s alpha ... 18
4.3 Graphical analysis ... 19
5. Empirical Results ... 19
5.1 Empirical results Dataset 1 ... 19
5.2 Empirical results Dataset 2 ... 20
5.3 Graphical analysis ... 22 6. Discussion... 23 7. Conclusion... 25 8. References ... 27 9. Appendix ... 30 9.1 Descriptive statistics ... 30 9.2 ESG Ratings ... 31
1
1. Introduction
Socially Responsible Investing (SRI) is a type of investing in which return performance is not the sole factor for the decision-making process in what assets to invest. The factor that also play a role for this type of investing, is directly or indirectly contributing to society, which can be done in two ways: first, in a way that excludes certain “controversial” assets or industries from the portfolio, which is done by using so-called sectoral screens, and second, by including certain assets or industries that match the own values and believes, which is done by using transversal screens (Capelle-Blancard and Monjon, 2014). The relevance of the topic is indicated by the increasing amount of invested money in socially respon-sible assets. Only in the US, every fourth invested dollar is invested into a fund that is related to a social cause (USSIF, 2018).
Socially responsible investments can normally be divided into five categories: environmental, social, economic, stakeholder and voluntariness (Dahlsrud, 2008). While some mutual funds take all categories into account, others only focus on a selection of them. Also, within their selection of categories, mutual funds differ in the way that they include or exclude companies and industries, depending on the method and number of filters that they use. What is interesting for researchers from the field of finance is, whether investments into socially responsible investment funds offer returns that are similar to the re-turns from conventional investing. For this research paper, the rere-turns of SRI funds with different screen-ing intensities are analysed to see if a shift from conventional investscreen-ing towards the most social forms of investing makes a significant difference regarding the returns that the investor can expect.
The two theories that most researchers in the past have used as the basis for their research, is first of all the overperformance theory and second the underperformance theory. The overperformance theory in-dicates that higher screening intensity of SRI funds leads to an overperformance over conventional funds in the long-term, because effective asset selection leads to a portfolio of the best-managed companies, leading to abnormal positive returns. The underperformance theory implies that higher screening inten-sity leads to underperformance of SRI funds compared to their conventional peers, which can be ex-plained by the fact that the investor has a smaller selection of assets to choose from, to set up a diversified portfolio (Renneboog et al., 2008b). These properties of socially responsible investing have also led to a combination of the two theories, which implies that low screening intensity as well as high screening intensity have benefits on return performance for the reasons mentioned before. Moderate screening intensity is seen as the worst choice, as none of the advantages are captured (Barnett and Salomon, 2006). This leads us to the research question of this paper:
“Does the intensity of screening used in the composition of a fund have an impact on the fund’s risk-adjusted return?”.
To answer this research question, we set up a measurement for screening intensity, that precisely cap-tures the differences from fund to fund. For this attempt, simply measuring screening intensity by the number of screens is insufficient. The reason for this is, that a comparison of funds that invest in different categories of socially responsible investing and apply different methods regarding inclusion or exclusion can easily distort the outcome, as the number of screens has to be measured at a common ground. There-fore, there will be a comparison between the SRI funds with each other as well as with their benchmark to answer the research question. The motivation for writing this paper, is the lack of up-to-date research in the field, while the relevance for socially responsible investing has been booming in the past decade. The 13 funds that are used, are directly derived from their parent index, the MSCI All Country World Index (ACWI), meaning that they cover a portfolio of firms from all around the world and all variables, except for the screening intensity, are kept constant. The attempt to use homogenous funds of this degree might give a deeper insight into which direction SRI might go in the future and sets up a basis for future research into the possibilities of growth and increasing need for socially responsible investing.
The methodology used in this paper is first of all the calculation of the Sharpe ratios and afterwards a CAPM model regression of each SRI fund, with the MSCI ACWI as the benchmark. The intention of this is to look for the significance of the alphas that can be isolated by running the regression. These abnormal returns will give a deeper insight into the over- or underperformance of the different funds. The testing is done for two datasets, with the first one reaching from the starting day of each fund up until the 31st of December 2019 with monthly data and the second one including daily data from the 1st of December 2015 up until the 31st of December 2019. The reason for using two datasets for this re-search paper is the fact that they complement each other. The longer time horizon in the first dataset gives a general insight into the relationship between risk-adjusted return and screening intensity over the whole life of each fund, while the shorter time span of the second dataset allows for precise compar-ison without any bias from coincidental events in the life of each fund.
The following section of this paper goes deeper into past research about this topic and will show where research has already taken us in this field. Afterwards comes a review into the data, the methodology that is used in this paper and the results that they lead to. Finally, the results will be discussed, including a comparison to other research, a conclusion on how to make use of the results that were found and how future research should look onto the topic.
3
2. Literature review
The literature review will give an impression on where research has headed in the past, how other re-searchers in the field set up and tested their hypothesis, and finally the results that they could find. By first defining SRI and showing its financial relevance, we set up a basis for a deeper analysis of the past literature, which has gone into the general analysis of socially responsible investing and its risk-adjusted return characteristics.
For this purpose, different ways of measurement and methodologies are analysed and compared. From there on, this research paper is heading towards a look into the research that has specifically dealt with screening intensity and risk-adjusted returns and the two generally accepted theories in the field, the overperformance theory and the underperformance theory. This will finally lead to the establishment of the hypotheses that will be tested in this paper.
2.1 The definition of SRI and its relevance
The literature review of this research paper starts with an explanation of socially responsible investing, shows how socially responsible investing has grown in the past and finally the theory that has in the past contradicted the existence of this type of investing, which is now questioned by researchers from the academic fields of finance and humanities.
Socially responsible investing is a form of investment in which the investor has the intention to contrib-ute to society, while also being able to expect a positive return on the initial investment. Due to the development of our society, in which mass media uncovered numerous bad business practices in the past, companies and investment banks became more conscious about the consequences of economic misbehavior, which did not only often lead to financial damage but also heavily impacted society or the environment (Olmedo et al., 2010).
The exponentially growing demand for SRI can be shown by the development of the percentage of money that is invested into socially responsible assets, as a share of the total amount of invested money. While socially responsible investing accounted for about 11% of the professionally managed assets in the US in the year 2010, this rose to around 25% in the year 2018. In nominal values, initially at 2.71 trillion dollars in 2010, the assets invested under socially responsible principles almost quintupled to around 12 trillion dollars in 2018 (USSIF, 2018). Nevertheless, this numeric comparison has limited meaningfulness in the way that it is doubtable that people have just become more social over time. What this means, is that other factors have also had an impact on this increasing demand for socially respon-sible investing. One of those factors is the trend of green-washing, which implies that companies engage
in contributions to society to improve their social image. Social responsibility is in this case used for marketing purposes rather than an altruistic social motivation (Laufer, 2003).
The mass media has given people from all around the world the chance to see the political and social situation in almost every country. This has made people become more aware of the fact that companies as well as governments do not always act in a way that is beneficial to society and environment, which has directly lead to an increase in demand for ways to positively contribute to society. This has not only led to an increase in donations but also in socially responsible investing, which can be seen by its grow-ing capitalization (National Philanthropic Trust, 2019; USSIF, 2018). The growth in capitalization of socially responsible investing has also directly led to an increasing supply of SRI mutual funds, making the topic relevant for investment theory. Also, other academic fields have been attracted by the broad nature of the topic, which has led to research papers coming from several academic directions like eco-nomics, psychology and even linguistics.
Research about the financial characteristics of socially responsible investing has gone into different dimensions. The research mostly differs regarding the time horizon that is used, their covered geo-graphic area or their scope into a certain factor like the method of screening or the use of certain events like crises. Socially responsible investing has made researchers from the fields of economics and finance question conventional economic theory, as the theory of the “Homo Economicus” describes human be-havior to be selfish and only dependent on self-interest (Persky, 1995). Under the rules of this theory, socially responsible investing cannot be explained, leading to a required extension of the utility function and the need for researchers to look deeper into the topic (Statman, 2004).
2.2 Social responsibility and financial performance
The question that socially responsible investing raises, is the question about its financial performance and by how far it can be compared to the returns that one can expect from conventional investing. The following section goes deeper into this topic and also shows how investor motivations are deviating from the sole desire of maximized returns.
Economic theory that already dates back to Adam Smith’s “An Inquiry into the Nature and Causes of the Wealth of Nations” (1776), suggests that human behavior is led by the desire to maximize their own self-interest, but the fact that SRI funds exist makes researchers doubt this superficial view (Renneboog et al., 2008b). The utility function of humans is also positively influenced by the utility of the people around them, making social responsibility an important factor for economic behavior (Bollen, 2006). What this implies, is that humans might accept a lower rate of return, when having the feeling of “doing good”.
5
The implementation of social responsibility has the potential to act as a signal to investors, as it proves that the company is actively engaged in building an advantageous reputation (Fombrun and Shanley, 1990). Costs that might arise from irresponsible business activities, like nuclear disasters, are prevented by not engaging in them, leading to fewer legal problems and fines (Nofsinger and Varma, 2014). The implementation of social responsibility leads to the attraction and retention of motivated workers, that work for an altruistic cause, which also directly implies a lower risk of bankruptcy (Renneboog et al., 2008a; Verwijmeren and Derwall, 2010).A disadvantage for companies that integrate some form of social responsibility into their business oper-ations, is the occurrence of operating costs that their conventional competition does not incur, leading to a worse positioning in the market and making them rely on governmental subsidies (Chang et al., 2012). Another problem that SRI funds run into is their dependence on the political situation, as some governments might be more supportive of social responsibility than others, giving companies the need to consider this additional risk factor.
This assumption might be offset by the fact that socially responsible firms do in many cases have good relations to governments, due to their effort to contribute to society. This allows the company to be in a better position in political decision making when compared to companies that only focus on financial performance (Nofsinger and Varma, 2014). Western governments have started initiatives to support socially responsible investing, while this is not the case for countries that already have difficulties with serving more basic needs (Renneboog et al., 2008a).
The aim of a company is normally the maximization of shareholder value, which can often not be per-fectly aligned with the maximization of stakeholder value. The reason for this are negative externalities, which are a result of certain business activities. Socially responsible investing aims at maximizing stake-holder value and research into the topic analyses whether this goes hand in hand with a value shift from the shareholders to other stakeholders, leading to higher operational costs that directly imply lower fi-nancial returns for the shareholders.
The results of different research papers differ regarding their conclusions to the underperformance of socially responsible investment funds. While some suggest that they underperform their conventional peers (Girard et al., 2007), others do not find a significant difference in risk-adjusted returns when com-paring SRI funds to their conventional peers (Hamilton et al., 1993; Goldreyer et al., 1999; Statman, 2000; Shank et al., 2005; Chang et al., 2012). The reason for this can be found in the fact that the advantages and disadvantages of socially responsible investing might offset each other, leading to in-significant differences between the two types of investing. Demand for socially responsible investments
has never been as high as nowadays, which is most likely a result of a different motivation than only the desire for the highest return performance as possible.
Table 1: The advantages and disadvantages of the integration of social responsibility into corporate structures
Moskowitz (2000) and Glode (2011) move away from regarding socially responsible investing as being an alternative to conventional investing, and focus on regarding SRI as a complement. While SRI funds underperform their conventional counterparts in a good market state, SRI funds become the superior choice in bad market states, as general uncertainty affects them less (Shefrin and Statman, 1993). This has an impact on investor behavior, as economic psychology suggests that investors are more affected by losses than gains of the same altitude, meaning that investments into SRI funds might be seen as a good hedge for downside risk (Kahneman and Tversky, 1979). What this implies, is that socially re-sponsible investing does also have an impact on investors that are indifferent about their way of invest-ing and might lead them to positively contribute to society as a desirable side effect.
Available SRI funds differ from each other in the way in which they are composed and range from funds that only exclude a few companies to funds that only include companies with the best and most mean-ingful social contributions. This range of funds with differing screening intensities allows investors to pick the funds that match their personal values and with which they can best identify. Nevertheless, investments into funds are made to profit from their financial return, which might significantly differ in altitude when comparing funds of differing levels of screening strictness with each other and their bench-mark(Renneboog et al., 2008b; Capelle-Blancard and Monjon, 2014).
2.3 Screening intensity and fund performance
Socially responsible investing is not a homogenous investment method, but can rather be done with different degrees of intensity, which means that the different findings of researchers are likely to also
7
depend on the strictness of the analysed SRI funds. The following section goes deeper into the relation-ship of screening intensity and risk-adjusted returns by comparing the different findings of previous research papers.Existing literature that analyses the impact of screening intensity on mutual fund performance is in most cases based on at least one out of these two underlying theories: First of all, the underperformance theory, which implies that high screening intensity leads to a smaller horizon of available assets to invest in, leading to a lower risk-adjusted return due to diversification costs. (Rudd, 1981; Renneboog et al., 2008b). Second, the overperformance theory, which implies that high screening intensity eventually leads to a small selection of the assets that underlie the best managers, and that these will eventually converge to positive abnormal returns in the long run. This theory is based on the assumption that man-agers who are able to integrate social responsibility, while not limiting regular business by an excessive degree, are superior to managers that are not able to do so.
Some research papers suggest a combination of the two theories, in which little screening intensity might be good, because it allows for more effective diversification, while very high screening intensity leads to a choice of only the best-managed companies. The problem here is rather seen in not choosing for one of the strategies, which will lead to diversification costs as well as a selection that does not only consist of the very best companies (Barnett and Salomon, 2006). This leads to a curvilinear relationship between screening intensity and risk-adjusted return, which has a U-shape.
Table 2: The effect of screening intensity on risk-adjusted returns
An underlying shortcoming of SRI funds is, that while a certain share of the included companies might have positive alphas, other companies that have positive alphas might be excluded from the fund, leading
to the absence of abnormal returns on average (Nofsinger and Varma, 2014). Lee et al. (2010) discov-ered, that higher screening intensity leads to lower return performance but also lower systematic risk, which also brings Laurel (2011) to a curvilinear effect on risk if moving from low screening intensity towards high screening intensity. Some research also suggests that there is no correlation between screening intensity and fund performance, but this research is mostly focused on a single country or a small group of countries (Hamilton et al., 1993; Goldreyer et al., 1997).
The results of each study depend on the method that is used to measure screening intensity. These dif-ferences do not only lead to different results but also complicate comparisons between different studies, as the underlying samples were treated differently. This means that different methods on the same sam-ple might lead to comsam-pletely different results, making an analysis of the different measurements im-portant for this research paper.
2.4 Measuring screening intensity
Generalizing the different methods of measuring screening intensity leads to inaccurate results, as chances are high that certain screens have a different impact on return performance than others. This raises the need to look into the way in which research papers up until now have been measuring screen-ing intensity. The used methods range from simply measurscreen-ing the number of screens (Barnett and Salo-mon, 2006) over only measuring the screening intensity by their type (Renneboog et al., 2011) to using specially designed SRI measures that are provided from different institutions (Capelle-Blancard and Monjon, 2014).
2.4.1 Number of screens
Using the number of screens is a common method, as it is easy to measure and enables a clear distinction between different SRI funds. Barnett and Salomon (2006) used in this case a range from 1 to 12 screens, which fitted their sample, but this number can deviate between different studies. This method does not capture two important elements: First of all, certain individual screens have a higher impact in general, for example, the exclusion of the tobacco industry or the exclusion of the entire coal industry might make a substantial difference. This method does not capture the effect of different screens but only the nominal amount of screens that are used for the composition of a fund.
Second, even within an individual screen there might be severe differences as some funds might com-pletely exclude the tobacco industry, while others might only exclude companies that derive a certain percentage of their revenue from tobacco-related business. These two shortcomings make the sole reli-ance on the number of applied screens inadequate.
9
2.4.2 Type of screens
Measuring screening intensity by the type of screens that are applied to create a particular fund allows for a clear distinction between the nature of funds. The reason for this is, that the nature of screens is taken into account, which leads to a less biased comparison than for the number of screens. In practice, it is difficult to analyse the correlation between fund performance and screening intensity, as regressions can only be run on quantitative measures and not qualitative ones.
Turning the type of screens into a quantitative measurement can become arbitrary, because it partially relies on a subjective valuation of screening intensity. In addition, small differences between the types of screens that are used already raise a comparison problem for the whole sample, which also makes this approach have its flaws (Chang et al., 2012).
2.4.3 SRI Ratings
The increasing demand for socially responsible investing has led to the appearance of companies that rate investment opportunities and their social engagement. Social responsibility is always to a certain degree subjective and cannot always be measured with numbers. SRI ratings are measurements, that were designed by rating agencies, to give funds a rating for their effort of investing in a socially respon-sible way. The advantage of using SRI ratings for the measurement of screening intensity is, that it is composed of a combination of the number and type of screens, put into a single measure. Due to the fact, that there are numerous providers of SRI ratings, quality comparisons can be made between them, to filter out the most precise ones. Capelle-Blancard and Monjon (2014) use the SRI ratings from Novethic to rank the screening intensity of their funds.
Drempetic et al. (2019) found a positive correlation between firm size and SRI ratings, which shows that certain ratings are subject to biases. This bias makes the objectiveness of SRI ratings from external providers doubtable. SRI rating agencies are often paid by the companies to rate their financial product. This raises the question whether the results might be biased, as no company would pay for bad ratings. Another problem is, that SRI ratings from different providers are hard to compare with each other, as the difficulty to reach a certain high-tier rating can significantly differ from provider to provider (Schol-tens, 2005).
2.5 Methodologies of past studies
After the choice of a screening intensity measurement, it is important to choose a method for the meas-urement of the dependent variable, the risk-adjusted return. Researchers have used different approaches to analyse the correlation between screening intensity and risk-adjusted return, while three approaches
have been used in the majority of research papers. The first measurement of risk-adjusted return is the Sharpe ratio, which allows for a precise comparison between funds (Renneboog et al., 2008a; Capelle-Blancard and Monjon, 2014). This measure uses the excess return of the fund over the risk-free rate divided by its standard deviation, which is its risk.
Sharpe ratio =
!"#$%&!'("#$%
Where 𝑟'*+, is the return of the mutual fund, 𝑟𝑓 is the risk-free rate and 𝜎'*+, is the standard devia-tion of the fund.
A disadvantage of using the Sharpe ratio, is its relative nature, which means that the numerical value of the Sharpe ratio is only meaningful in direct comparison to the Sharpe ratios of other funds. Neverthe-less, this property is not very relevant to this research paper due to the fact that direct comparison is intended in this approach. The second measure that most of the research papers into this topic use is the comparison of Jensen’s alpha, which is an absolute measure and shows the abnormal returns of the different funds (Renneboog et al., 2008a; Renneboog et al., 2008b; Barnett and Salomon, 2006). Some of the research papers gave this measure a new name: Risk-adjusted performance (RAP), but basically mean the exact same thing as the commonly used Jensen’s alpha (Barnett and Salomon, 2006; Capelle-Blancard and Monjon, 2014). This measurement is used by first running a regression of the CAPM model on the individual SRI funds and afterwards isolating the alpha, which can be used in direct cop-parison as well as an absolute measurement of abnormal returns.
Capital Asset Pricing Model (CAPM) and Jensen’s alpha as a measurement for excess perfor-mance over the expected perforperfor-mance given by the CAPM
𝑟
'*+,− 𝑟𝑓 = 𝑎 + ß ∗ (𝑟
6− 𝑟𝑓) + 𝑒
𝐽𝑒𝑛𝑠𝑒𝑛
=𝑠 𝐴𝑙𝑝ℎ𝑎 = B𝑟
'*+,
− 𝑟𝑓C − ß ∗ (𝑟
6− 𝑟𝑓)
Where 𝑟'*+, is the return of the particular fund, 𝑟𝑓 the risk-free rate, 𝑎 is Jensen’s alpha, ß is the fund’s CAPM beta, 𝑟6D!EFG is the return of the market.
The third model, that is commonly used in previous literature, is the multi-factor model. This model includes additional factors to measure the abnormal returns and is intended to describe the returns of funds and stocks in a way that is more precise than the CAPM model. Two commonly used models are the Fama-French three-factor model and the Carhart four-factor model. These models include factors that give a deeper insight into the structure of the market and allow for an analysis that goes beyond the analysis of the CAPM model.
Fama-French three-factor model
11
Carhart four-factor model
𝑟
'*+,= 𝑟𝑓 + ß ∗ (𝑟
6− 𝑟𝑓) + ß
HIJ∗ 𝑆𝑀𝐵 + ß
NIO∗ 𝐻𝑀𝐿 + ß
RIS∗ 𝑈𝑀𝐷 + 𝑎 + 𝑒
Where rWXYZ is the return of the particular fund, rf the risk-free rate, a is the abnormal return, ß is the fund’s market beta, ß]^_ is the coefficient for the SMB factor, ß`^a is the coefficient for the HML factor, ßb^c is the coefficient for the UMD factor, SMB stands for Small-minus-big, HML stands for High-minus-Low, UMD stands for the momentum factor and rdefghi is the return of the market. The results that were found using these three measurements differ between the several research papers, which can be explained by the same reasoning as the differences in the findings of the papers, that analyse the general performance of SRI funds. The papers do not only differ in the countries, funds and benchmarks that are used, but also in the way in which they measure screening intensity.
Table 3: Results of past studies regarding the over- and underperformance of SRI funds
The funds that are used in the research papers either have a national or an international scope, which differs between research papers. Hamilton et al. (1993) and Goldreyer et al (1997) use a single-country approach which allows for a more individual and specific comparison. Schröder (2004), Bauer et al. (2005) and Kreander et al. (2005) use a multi-country approach to analyse the fund performance of SRI funds, which has the advantage that it limits the presence of country-specific biases.
Barnett and Salomon (2006), as well as Capelle-Blancard and Monjon (2014), use a graphical scheme in which they show the correlation between risk-adjusted returns and screening intensity, which clearly shows their findings of a curvilinear effect for the sample that they use. This method allows for a clear and distinctive overview of their results and serves as a precise summary for their purpose, which is similar to this research paper.
Graph 1: The curvilinear relationship between risk-adjusted return and screening intensity retrieved from Barnett and Salomon (2006)
The risk-adjusted return, which is the dependent variable is plotted on the different degrees of screen-ing intensity, which is the independent variable. The advantage of this graph is that the linearity of the relationship is directly visible and allows for direct inclusion and exclusion of the existence of certain patterns. Nevertheless, this method does not make a numerical approach obsolete, as it only serves as a summarizing presentation of the results, and does not serve the purpose for exact numerical analysis. The combination of different screening intensity measurements and risk-adjusted return measurements makes almost all of the research papers in the past unique and explains their different outcomes. Never-theless, most researchers settled on the use of at least one of these measures, which at least allows for some comparison between the different research papers.
2.6 Hypotheses
The hypotheses in this research paper have the intention to analyse the different dimensions of the im-pact of screening intensity on risk-adjusted performance. The first dataset, which includes monthly data from each fund, starting on their day of introduction, allows us to precisely analyse the fund’s perfor-mance compared to its benchmark. The fact that low screening intensity as well as high screening in-tensity have advantages over the choice of moderate screening inin-tensity leads to the first hypothesis.
Hypothesis 1: Starting at of the day of their introduction, funds with low and high screening intensity do not significantly underperform their benchmark, while funds with moderate screening intensity do underperform.
The advantage of including a hypothesis that deals with the historical performance dating back to the day of introduction of the funds is, that it shows the performance over different market states, which is
13
not the case for an analysis that only includes a particular time frame. The results from testing this hypothesis are as unbiased from any exceptional situations or market states as possible. Nevertheless, the results can only be used to compare the different SRI funds to their benchmark, which raises the need for the second hypothesis that deals with the comparison of the funds with each other.Hypothesis 2: Using the same time frame, funds with low and high screening intensity have a higher risk-adjusted return than funds with moderate screening intensity.
To test the second hypothesis, the second dataset is used, which includes the daily returns of the funds in a time frame in which all of the SRI funds existed. The results from this hypothesis test are intended to show whether any abnormal returns can be related to the screening intensity of the funds, without any bias that might have arisen in the first dataset due to different founding dates. The third hypothesis has the purpose of summarizing the results and should give an overview of the curvilinear effect of screening intensity on risk-adjusted returns.
Hypothesis 3: Regardless of the time frame, the relationship between screening intensity and risk-ad-justed returns is curvilinear and has a U-shape.
The insights that are gained from testing this hypothesis are intended to give an impression on the under- and overperformance of the benchmark as well as when comparing the funds directly with each other. This research paper has the intention to add a study to the existing literature, in which homogenous SRI funds are used and compared to their underlying parent index. This has the advantage, that all var-iables are kept constant, while only changing the screening intensity by adding more and more appli-cable filters onto the investment universe of the very broad parent index. By conducting research on this basis, it is possible to effectively isolate the effect of a changing screening intensity on the fund’s risk-adjusted return, while preventing any falsification by selection biases.
3. Data
This section deals with the data that is used for this research paper. The section starts with a descrip-tion of the benchmark, the MSCI ACWI and how the different SRI funds are derived from this parent index. Afterwards, the different SRI funds are categorized into different levels of screening intensity to make the comparison valuable for other samples as well. The description of the SRI funds shows that the number of included companies from the parent index is a reliable measurement for screening in-tensity and combines the approaches of previous research papers.
3.1 Sampling
Similar to the approach of Schröder (2004), Bauer et al. (2005) and Kreander et al. (2005), this research paper uses a multi-country approach, by using the MSCI ACWI as the benchmark. The data that is analysed in this study is directly retrieved from the website of MSCI. With the coverage of 23 developed countries, 26 developing countries and in total more than 3000 businesses in the large and mid-cap sector, the MSCI ACWI is one of the broadest indexes available to investors.
Retrieved from the MSCI website
The 13 SRI funds that were chosen for this research paper, all build upon the ACWI and only list certain companies from the parent index, but do not add new ones that are not included in the parent fund. The market capitalization of all companies together is around 39.8 trillion dollars. MSCI uses three different ratings to set up its SRI funds: ESG Rating, ESG Controversies score and business involvement screen-ing research. The use of a SRI ratscreen-ing follows the approach of Capelle-Blancard and Monjon (2014), who used an external rating agency to make the funds comparable. In this research paper, the rating is pro-vided by MSCI itself.
The ESG Rating is a rating that measures the management of ESG-related risks in a company, reaches from CCC to AAA and is basically divided into 7 grades. What is taken into account are the three main categories which are environmental, social and governance factors, and their 37 subcategories. The rat-ing is composed of a combination of the score in every individual category. The ESG Controversies score is a rating that measures the controversies that a particular company is engaged in and includes activities that have a negative impact on environment, society or corporate governance. It is strongly connected to the United Nations Declaration of Human Rights as well as the International Labor
15
Organization. The score reaches from 0 to 10, with 0 being given to the most controversial companies. Business involvement screening measures the involvement of companies in certain industries or busi-ness activities, that might lead to their exclusion from certain funds. This measure is not directly used to rate a company, but rather to evaluate their eligibility for a particular SRI fund.Table 4: Overview of the SRI funds and the benchmark, the MSCI ACWI
The first dataset in Table 4, from now on referred to as Dataset 1, includes the monthly returns of the funds, starting on the day of their first offering. The advantage of this dataset is, that it allows for an analysis of the return performance over the whole life of each fund. The data can efficiently be compared to the ACWI using the same time frame to exclude any biases related to certain market states and other coincidental events.
The second dataset, from now on referred to as Dataset 2, includes the daily returns of each fund between the 1st of December 2015 up until the 31st of December 2019, this defined time frame increases the comparability of the funds. The two datasets are complementary, as they both capture results that would be strongly limited when only using one of them. The reason for this is, that the analysis of Dataset 1 gives important historical insights into the relationship between risk-adjusted returns and screening in-tensity and the over- or underperformance of the benchmark, while Dataset 2 is rather focused on di-rectly comparing funds to each other. The reason for the choice of daily returns instead of monthly returns for Dataset 2 is, that a sample of monthly data would include a number of observations that is too small to make the assumption that the results from the analysis are reliable. The risk-free rate that is used for the CAPM regression and the calculation of the Sharpe ratios is the 1-month Treasury bill, to precisely match the frequency of returns in the samples. It was retrieved from the Wharton Research and Data Services (WRDS) database and includes monthly data, which has been turned into a rate with daily frequencies for Dataset 2 by using a formula that takes daily compounding into account.
3.2 Categorization
For further analysis, the funds are categorized into one of three degrees of screening intensity, catego-rizing them by the amount of companies that they still include from the parent index. This section gives a general overview of the criteria that exclude and include certain companies from the funds, while more in-depth information can be found in the appendix.
The best common ground for the comparison of screening intensity with other research papers, is achieved by an approach that combines the methods of other researchers in the field. This approach is based on categorizing the different SRI funds into one of three degrees of screening intensity, which are low, moderate and high screening intensity. The funds are divided by using the number of companies that are still included from the parent index, the MSCI ACWI. This number is negatively correlated to the number of screens, their strictness and the SRI rating received by MSCI. This approach combines the measurement of the number of screens, used by Barnett and Salomon (2006), the type of screens used by Renneboog et al. (2011), as well as the use of SRI Ratings, which was used by Capelle-Blancard and Monjon (2014).
Funds with the lowest amount of included companies have the highest screening intensity, which is justified by an increasing difficulty to be eligible for the particular fund and the granting of a high SRI rating. This approach ensures good comparability between the funds and prevents biases that would arise from simply categorizing them by the number of screens that they use. Categorizing the funds simplifies the results and makes them more applicable to other samples, as every SRI fund is unique and individual results are not meaningful to the whole population of SRI funds.
The five funds that include the most companies are in the Low screening group, four are in the moderate screening group and another four in the high screening group. The cutoff was done in a way in which the different degrees of screening intensity almost include the same number of funds, and the differences regarding the number of included companies, are sufficiently large between the different degrees of screening intensity.
The low screening intensity funds are mostly based on sectoral screens, which means that they exclude companies from certain sectors, industries or countries. Due to the fact, that the barriers of entry to these funds are low, they include a large share of the companies from their parent index, the MSCI ACWI. These funds aim at attracting investors, that do not want to invest in certain industries or into companies that engage in human right violations, while not excessively limiting their diversification possibilities.
17
Table 5: The 13 SRI funds categorized by their degree of screening intensity, retrieved from MSCImethodology sheets
The moderate screening intensity funds include sectoral as well as transversal screens, which means that they not only exclude firms with certain characteristics but also only include companies that fulfil certain requirements. The companies that are eligible for these funds need a moderate SRI rating, which is directly connected to the requirement for companies to have low negative externalities on the environ-ment and society.
The high screening intensity funds focus on transversal screens and have high barriers to entry. The companies that are eligible for these funds need high SRI ratings and need to actively contribute to society, by implementing high standards of ESG. These funds aim at attracting investors that clearly value social responsibility over diversification. The funds that are used in this research paper offer a broad selection of different screens, which makes the analysis valuable to other samples and allows for comparison to previous studies.
4. Methodology
The measurements and methodologies that are used in this research paper are based on the most com-monly used methods and have the intention to offer a solid ground for comparison to other research papers. This is done by using a combination of the approaches used by other researchers in the field, which are the Sharpe ratio and the Jensen’s alpha. The multi-factor model is not used, as all of the funds in the sample are directly derived from the parent index, which also serves as the benchmark.
The Fama-French three-factor model and the Carhart four-factor model are only applicable to the spe-cific excess market return of the benchmark that they use. This means that these models cannot be
applied on the datasets of this research paper. Using the Sharpe ratio as well as the Jensen’s alpha has the advantage that the results can be used to verify each other. The first and second hypothesis are tested by using the Sharpe ratio and the Jensen’s alpha, while the third hypothesis is tested by using a graphical analysis of the relationship between risk-adjusted return and screening intensity.
4.1 Sharpe ratio
Earlier research papers often included the Sharpe ratio of their funds, which is why this measure will also be calculated in this research paper, as it makes a comparison to the work of other researchers more convenient. The Sharpe ratio can be calculated by using the average excess returns of the funds over the risk-free rate divided by their standard deviation, which is their risk.
Sharpe ratio =
!jkl&!' (jklWhere 𝑟Hmn is the return of the SRI fund, 𝑟𝑓 the risk-free rate and 𝜎Hmn is the standard deviation of the returns of the SRI fund.
The Sharpe ratio is a commonly used measure as it is simple to calculate and compare. Nevertheless, its relative character implies that the measure is not valuable on its own and needs to be compared to a benchmark or other reference.
As previously mentioned, the results of Dataset 1 are usable for a direct comparison between the SRI funds and their benchmark, while the results of Dataset 2 are suitable for comparing the different SRI funds to each other. For the first and second hypothesis the Sharpe ratio as well as Jensen’s alpha are used to verify the reliability of the results and increase the overlap with previous literature.
4.2
CAPM and Jensen’s alpha
The Jensen’s alpha allows us to measure the risk-adjusted return as it incorporates the beta of the fund and shows the excess return that cannot be explained by the CAPM. The regression is run on the two datasets, to get a deeper insight into whether SRI fund performance has changed over time and to make the funds properly comparable to each other.
CAPM with ACWI as market index
𝑟
Hmn− 𝑟𝑓 = 𝑎 + ß ∗ (𝑟
opqn− 𝑟𝑓) + 𝑒
The alpha is the excess performance over the expected performance given by the CAPM
19
Where 𝑟Hmn is the return of the SRI fund, 𝑟𝑓 the risk-free rate, 𝑎 is Jensen’s alpha, ß is the fund’s CAPM beta, 𝑟opqn the return of the MSCI ACWI (benchmark) and RAP the risk-adjusted perfor-mance.This paper analyses the cumulative abnormal returns of the individual SRI funds over their parent index, the MSCI ACWI. To account for differences in the structure of risk, the Jensen’s alpha is used. For measuring the alpha, it is necessary to run a regression of the funds returns on the CAPM with the MSCI ACWI to represent the market in this case. The alpha is the abnormal return that is not explained by the market movement or the fair compensation for excess risk and shows whether the particular SRI fund has a significantly different return than what would be expected by its risk profile.
4.3 Graphical analysis
The third hypothesis, that deals with the curvilinear effect of screening intensity on risk-adjusted returns can be answered by a graphical analysis of their relationship. A graphical presentation of the relationship between the risk-adjusted returns and screening intensities of each fund has the advantage of showing a relationship without raising the need to run quantitative tests, which would not be reliable, due to the insufficient number of observations of alpha’s and funds.
As already mentioned in the literature review, a graphical presentation of the results gives deeper in-sights into the relationship of screening intensity and risk-adjusted returns and allows us to make better inferences on Hypothesis 3.
5. Empirical Results
The empirical results in this section are intended to answer the previously designed hypotheses. The two quantitative models are used for both datasets to answer the first and second hypothesis. The graphical analysis at the end of this section is used to give an answer to the third hypothesis and all of the results are finally summarized into a table.
5.1 Empirical results Dataset 1
The analysis of Dataset 1 shows, that 10 out of the 13 SRI funds have a higher Sharpe ratio than their benchmark, the MSCI ACWI. Economic theory suggests, that the market portfolio should have the highest Sharpe ratio (Sharpe, 1965). This property does not hold for the data in Dataset 1, as only 3 of the funds have a lower Sharpe ratio than their benchmark. The differences of the Sharpe ratios between the different SRI funds are not comparable, as the time frame that is used in Dataset 1 is different for almost all of the funds. Nevertheless, all of the differences between the Sharpe ratio of the funds and
their benchmark are relatively low, which means that further testing of the data is necessary to make an inference on the significance of the difference in risk-adjusted returns.
Table 6: Sharpe ratios and CAPM results of the funds and their benchmark, retrieved from Dataset 1 (Note that *,** and ***, indicate the significance levels of 10%, 5% and 1% respectively and the num-bers in brackets are the p-values.)
The deeper analysis of the differences is done by isolating the alphas of the different funds by using the CAPM and analysing their statistical significance. As you can see in Table 6, three of the funds seem to overperform their benchmark, of which one only does so at the 10% significance level, while the other two have significant abnormal returns at the 1% significance level. None of the funds significantly un-derperforms the benchmark. Two of the funds that overperform their benchmark significantly are in the low screening-intensity group, and one of them is in the high screening intensity group. The first im-pression of this would suggest, that there might be a curvilinear pattern for the relationship between risk-adjusted return and screening intensity. The average of abnormal returns is higher for the low and high screening intensity groups, although the difference between the low and moderate group is rela-tively low. The results found in the analysis of Dataset 1 show that it is possible to reject Hypothesis 1, as none of the funds significantly underperforms the benchmark, regardless of the screening intensity.
5.2 Empirical results Dataset 2
The analysis of Dataset 2 gives a slightly different result than that of Dataset 1, as 8 of the funds over-perform the benchmark, and 5 underover-perform, when comparing their Sharpe ratios to the benchmark. The returns in this sample are noticeably smaller than in Dataset 1, which is a result of the shift in data
21
frequency from monthly data to daily data. Also, the risk-free rate in this sample is the same for every fund due to the equal time frame. As one can see in Table 7, the ACWI SRI has the highest Sharpe ratio of all funds, meaning that its risk-adjusted return is the highest when using this measure. The ACWI Sustainable Impact Index has the second highest Sharpe ratio, followed by the ACWI that excludes Fossil fuels.Table 7: Sharpe ratios and CAPM results of the funds and their benchmark, retrieved from Dataset 1 (Note that *,** and ***, indicate the significance levels of 10%, 5% and 1% respectively and the num-bers in brackets are the p-values.)
The funds that overperform the benchmark are coming from all of the screening-intensity groups which shows that there is no pattern for a relationship between risk-adjusted returns and screening intensity. After the analysis of the Sharpe ratios of each fund, the data is again analysed by a regression on the CAPM, which directly shows that only one fund overperformed the benchmark significantly, and also this overperformance is only at a significance level of 10%. Nevertheless, none of the funds underper-formed the benchmark significantly. The average abnormal returns for the low and high screening in-tensity funds are again higher than for the moderate screening inin-tensity group. The results show that Hypothesis 2 can be rejected, as only one fund has significant abnormal returns. This means, that it is not possible to infer that the low and high screening intensity funds have higher risk-adjusted returns than the moderate screening intensity group.
5.3 Graphical analysis
To conduct a graphical analysis of the results, the risk-adjusted returns, in the form of Jensen’s alphas are plotted on the fund’s screening intensity. This measure was chosen over the Sharpe ratios, as it allows for direct comparison between the two datasets, for the reason that the Sharpe ratios in Dataset 1 can only be compared to the Sharpe ratio of the benchmark.
The data in Graph 2 coming from Dataset 1, show a non-linear pattern of the relationship between risk-adjusted return and screening intensity. The overperformance of the benchmark seems in general to be randomly distributed, even though the funds that overperform the most are in the lower and higher screening intensity groups, which might speak in favor of a curvilinear relationship. Nevertheless, an analysis of the pattern in Dataset 2 is needed to exclude the possibility that this result is biased from any special market states or exceptions that might have arisen from the founding date of the particular funds.
Graph 2: Jensen’s alpha plotted on the fund’s screening intensity for both datasets
The data from Dataset 2, shows a different pattern, in which all of the funds alpha’s are around 0, except for the ACWI SRI and the ACWI Sustainable Impact Index, which have alphas that are clearly above 0. The ACWI Islamic Index visibly underperforms the benchmark. This pattern shows once more, that a comparison of the fund’s in the same time frame is essential to come to a valuable conclusion when analyzing the relationship between risk-adjusted returns and screening intensity. Nevertheless, it is im-portant to keep in mind, that the only significant deviation from 0 was only at the 10% significance level, which means that it is not unlikely that in reality none of the funds overperforms the benchmark in this time frame.
23
The results from the graphical analysis imply that it is not possible to reject Hypothesis 3, as the sample in Dataset 1 shows higher risk-adjusted returns for the low and high screening intensity funds, which might have been coincidental, but it is not possible to infer that there is no curvilinear effect of screening intensity on risk-adjusted return.Table 8: Summary of hypotheses, measurements and conclusion
6. Discussion
The differences in values of the analysis of the two datasets might be explainable by the following reasons. First of all, the time frame in Dataset 1 is already dating back to the year 2002 for the earliest available fund, which implies significantly different risk-free rates but also a market state that cannot easily be compared to the post-2009-crisis period of 2015 to 2019. Another reason for the differences might be a sector-based bias, as some industries that are included or excluded might have been more or less important in the past than nowadays, which might also lead to significantly different results when directly comparing the two datasets.
The direct comparison in Table 9 shows that some funds that have been under- or overperforming their benchmark since their founding date, have not compulsively overperformed during a more recent time frame. A fund’s return performance is directly influenced by the demand for the fund, as a money inflow leads to rising prices. A trend-based bias would imply, that trends might lead to a rise in de-mand for a particular way of socially responsible investing, which would mean that differences in risk-adjusted return cannot only be traced back to screening intensity, but need to be put in their historical context to see if external factors might have played a role for the development of the price over time.
The two different measures used in this analysis lead to approximately the same results. The analysis of the datasets showed that socially responsible funds do not necessarily underperform their conven-tional benchmark, even though other research found significant evidence for this. Nevertheless, every sample that was used in previous literature differed in all characteristics and none of the results can be compared perfectly. In Dataset 1, two of the funds from the lower screening intensity group significantly overperformed their benchmark. These two funds exclude a similar set of industries, which might be an indication, that these industries have performed poorly in the past. The fossil fuel industry, which gets more and more replaced by renewable energies has already had its booming years, which might be a reason for the overperformance of funds that particularly exclude it.
Table 9: Summary of Jensen’s alpha for the two datasets (Note that *,** and ***, indicate the signifi-cance levels of 10%, 5% and 1% respectively)
The ACWI ESG leaders fund, which overperformed its benchmark in Dataset 1, is a fund that offers investors a position between investing into a social cause and expecting a diversified and well-composed portfolio. The best-in-class approach of this fund and the fact that this fund is one of the flagships of MSCI’s SRI funds, might have led to a substantial money inflow.
In Dataset 2, only one fund significantly overperformed the benchmark, which is the ACWI SRI. This fund is one of the most social funds in the sample, and unlike the Sustainable Impact Index, does not only include companies that made it their mission to actively contribute to society but also funds that are doing their regular business and at the same time ensure to minimize their negative externalities on society and the environment. The overperformance of this fund over its benchmark might be justified by the increasing popularity of socially responsible investing without completely moving away from conventional investing, which directly leads to an increase in money inflow. The relative, as well as absolute growth in the amount of money that is invested in socially responsible assets, has especially boomed in the past few years, making it a logical consequence for funds like this to perform well when analyzing the specific time frame of Dataset 2.
25
The analysis of the datasets and the resulting graphs do not justify the rejection of the hypothesis that a curvilinear relationship between risk-adjusted returns and screening intensity exists. Dataset 1 shows an overperformance of the funds of the low and high screening intensity funds, while Dataset 2 shows no clear pattern. Previous literature had a stronger focus on single countries and specific markets, which entails that certain country-specific factors might have had an impact on the results. The fact that almost every research paper in this field comes to different conclusions, should be seen as an incentive, rather than discouragement, for researchers to continue publishing new research in this academic area, which is one of the most promising of our time.7. Conclusion
The analysis conducted in this research paper does not find significant evidence to infer that Hypothesis 1 and Hypothesis 2 can be confirmed. The presence of a curvilinear relationship between risk-adjusted returns and screening intensity cannot be rejected. The theory which this research paper is built upon suggests that choosing low, as well as high screening intensity leads to advantages over a choice that implies moderate screening intensity. According to theories from past literature, low screening intensity allows for effective diversification while high screening intensity leads to a selection of the most well-managed companies.
The theory that might explain the lack of diversification costs, when choosing the funds with the highest screening intensities, is that a relatively small number, often stated to be around 50, of randomly chosen companies is already sufficient for a well-diversified portfolio (Campbell et al., 2001). Even though the companies, that are included in the funds were not chosen randomly, they include a sufficient amount of companies from different industries to be considered well-diversified. Nevertheless, the lack of ab-normal returns for the funds with high screening intensity can be explained by the fact that well-managed companies might be included, but other companies that would offer abnormal returns are likely also excluded from the funds, leading to normal returns on average. Funds with low screening intensity might be exposed to the opposite effects, which means that they include more companies that are underper-forming and overperunderper-forming as well as being well-diversified through a bigger number of included funds. This research paper adds a new approach to existing literature, as it focusses on funds that invest into companies from all around the world, which allows for an unbiased analysis, that is free from any country- or market-specific biases.
The limitations of this research paper are, first of all, its necessarily purely financial and static view of the relationship between screening intensity on risk-adjusted return. Socially responsible investing is a topic that needs to be analysed from a multidimensional academic approach, as many factors that play
a significant role are nonfinancial. The returns of each SRI fund need to be analysed in their historical context, in which certain social trends might have had an influence on their performance. Second, it is difficult to generalize the results that were found in this research paper to other samples of funds, as all of them have individual factors that might change the outcomes when running similar regressions. The categorization of the funds into low, moderate and high screening intensity is to some extent based on subjective valuation, similar to other research papers in this field. The low number of SRI funds that were used in this research paper limit its meaningfulness and does not allow for statistical tests of the Sharpe ratios. This means, that the results that were found in the analysis of the datasets in this research paper might not be applicable to any other funds or markets.
Nevertheless, this research paper should be seen as an incentive for further research, not only from the academic fields of economics and finance but also from other academic fields, as social responsibility is one of the most multi-faceted research topics in social sciences. Future research should focus on dif-ferent time frames and an approach that is more connected to other academic fields as well. Deeper research into socially responsible investing might lead to a better understanding and new insights into the altruistic causes of social behaviour and the reasons for the shift from pure self-interest towards an individual’s value-maximization that is based on positive contributions to society.
27
8. References
Bauer, R., Koedijk, K. & Otten, R. (2005). International evidence on ethical mutual fund performance and investment style. J. Banking Finance. Vol. 29, 1751-1767.
Barnett, M. & Salomon, M. (2006). The curvilinear relationship between social responsibility and fi-nancial performance. Strategic Management Journal. Vol. 27 (11), 1101-1122.
Bollen, N. (2007). Mutual fund attributes and investor behavior. Journal of Financial and Quantitative
Analysis. Vol. 42, 683-708.
Campbell, J., Lettau, M., Malkiel, B. & Xu, Y. (2001). Have individual stocks become more volatile? An empirical exploration of idiosyncratic risk. Journal of Finance. Vol. 56 (1), 1-43.
Capelle-Blancard, G. & Monjon, S. (2014). The performance of socially responsible funds: Does the screening process matter?. European Financial Management. Vol. 20 (3), 494-520.
Chang, C. E., Nelson, W. A., & Doug Witte, H. (2012). Do green mutual funds perform well?.
Man-agement Research Review. Vol. 35 (8), 693-708.
Dahlsrud, A. (2008). How Corporate Social Responsibility is Defined: an Analysis of 37 Definitions.
Corporate Social Responsibility and Environmental Management. Vol. 15, 1-13.
Drempetic, S., Klein, C. & Zwergel, B. (2019). The influence of Firm Size on the ESG Score: Corpo-rate Sustainability Ratings under Review. Journal of Business Ethics.
Fombrun, C. & Shanley, M. (1990). What’s in a name? reputation building and corporate strategy.
Academy of Management Journal. Vol. 33 (2), 233-258.
Girard, E., Rahman, H. & Stone, B. (2007). Socially responsible investments: goody-two-shoes or bad to the bone?. Journal of Investing. Vol. 16 (1), 96-110.
Glode, V. (2011). Why mutual funds “underperform”. Journal of Financial Economics. Vol. 99 (3), 546-559.
Goldreyer, E. F., Ahmed, P. & Diltz, J. D. (1999). The performance of socially responsible mutual funds: incorporating sociopolitical information in portfolio selection. Managerial Finance. Vol. 25 (1), 23-36.
Hamilton, S., Jo, H. & Statman, M. (1993). Doing well while doing good? The investment perfor-mance of socially responsible mutual funds. Financial Analysts Journal. Vol. 49 (6), 62-67.
Kahneman, D. & Tversky, A. (1979). Prospect theory: an analysis of decision under risk.
Economet-rica. Vol. 46, 171-185.
Kreander, N., Gray, R.H., Power, D.M. & Sinclair, C.D. (2002). The financial performance of Euro-pean ethical funds 1996–1998. Journal of Accounting and Finance. Vol. 1, 3-22.
Kreander, N., Gray, R.H., Power, D.M. & Sinclair, C.D. (2005). Evaluating the performance of ethical and non-ethical funds: a matched pair analysis. Journal of Business Finance and Accounting. Vol. 32 (7), 1465-1493.
Laufer, W.S. (2003). Social Accountability and Corporate Greenwashing. Journal of Business Ethics. Vol. 43, 253-261.
Laurel, D. (2011). Socially responsible investments in Europe: The effects of screening on risk and the clusters in the fund space. Available at SSRN 1883427. 1-47.
Lee, D.D., Humphrey, J.E., Benson, K.L. & Ahn, J.Y.K. (2010). Socially responsible investment fund performance: the impact of screening intensity. Journal of Accounting and Finance. Vol. 50 (2), 351-370.
Moskowitz, T. J. (2000). Discussion. Journal of Finance. Vol. 55 (4), 1695-1703.
Nofsinger, J., & Varma, A. (2014). Socially responsible funds and market crises. Journal of Banking
and Finance. Vol. 48, 180-193.
National Philanthropic Trust. (2019).The 2019 Annual Report. https://www.nptrust.org/reports/an-nual-report/ (retrieved 23.06.2020)
Olmedo, E., Muñoz-Torres, M. J. & Fernández-Izquierdo, M. A. (2010). Socially responsible invest-ing: sustainability indices, ESG rating and information provider agencies. International Journal of
Sustainable Economy. Vol. 4 (2), 442-461.
Persky, J. (1995). Retrospectives: The Ethology of Homo Economicus. Journal of Economic
29
Renneboog, L., Ter Horst, J., & Zhang, C. (2008a). Socially responsible investments: Institutional as-pects, performance, and investor behavior. Journal of Banking & Finance. Vol. 32, 1723-1742. Renneboog, L., Ter Horst, J. & Zhang, C. (2008b). The price of ethics and stakeholder governance: evidence from socially responsible mutual funds. Journal of Corporate Finance. Vol. 14 (3), 302-332. Renneboog, L., Ter Horst, J. & Zhang, C. (2011). Is ethical money financially smart? Nonfinancial at-tributes and money flows of socially responsible investment funds. Journal of FinancialIntermedia-tion. Vol. 20 (4), 562-588.
Rudd, A. (1981). Social responsibility and portfolio performance. California Management Review. Vol. 23 (4), 55-61.
Scholtens, B. (2005). Style and Performance of Dutch Socially Responsible Investment Funds. The
Journal of Investing. Vol. 14, 63-72.
Schröder, M. (2004). The performance of socially responsible investments: investment funds and indi-ces. Financial Markets and Portfolio Management. Vol. 18, 122-142.
Shank, T., Manullang, D. & Hill, R. (2005). Doing well while doing good revisited: a study of socially responsible firms’ short-term versus long-term performance. Managerial Finance. Vol. 31 (8), 33-46.
Sharpe, W. (1965). Risk-aversion in the stock market: some empirical evidence. Journal of Finance. Vol. 20 (3), 416-422.
Shefrin, H. & Statman, M. (1993). Ethics, fairness, and efficiency in financial markets. Financial
Ana-lysts Journal. Vol. 49 (6), 21-29.
Smith, A. (1776). An Inquiry into the Nature and Causes of the Wealth of Nations.
Statman, M. (2000). Socially responsible mutual funds. Financial Analyst Journal. Vol. 56 (3), 30-39. Statman, M. (2004). What do investors want? Journal of Portfolio Management 30th Anniversary
Is-sue. 153-161.
USSIF, (2018). Report on US Sustainable, Responsible and Impact Investing Trends. https://www.us-sif.org/files/Trends/Trends%202018%20executive%20summary%20FINAL.pdf (retrieved
03.06.2020).
Verwijmeren, P. & Derwall, J. (2010). Employee well-being, firm leverage, and bankruptcy risk.
9. Appendix
9.1 Descriptive statistics
Appendix Table 1: Descriptive statistics of Dataset 1
Appendix Table 2: Descriptive statistics of Dataset 2
Appendix Table 3: Differences in Sharpe ratios of all the funds; the Sharpe ratios of the funds on the horizontal axis are subtracted by the Sharpe ratios of the funds on the vertical axis
31
9.2 ESG Ratings
MSCI ESG Rating Subcategories retrieved from MSCI fund methodology sheet
9.3 SRI Fund Characteristics
Low Screening intensity
ACWI excluding tobacco Excludes companies that directly produce or re-trieve more than 5% of their revenue from to-bacco-related business
ACWI excluding controversial weapons Producers of controversial weapons, and compa-nies owning shares in the producers are excluded
