Sustainable Investment
A study on performance differences between sustainable and
conventional portfolios
Abstract The past decades sustainable investing has become more mainstream among investors, which raises the question whether socially responsible investments lead to a significant difference in portfolio return. This research focuses on long term abnormal return differences between sustainable and non-sustainable portfolios in the US between 2005 and 2015. Based on MSCI ESG scores, two equity portfolios are constructed. The high-ranked 10-year sustainable portfolio performed significantly worse than the low-rated conventional portfolio. Portfolio comparison over a 5-year period does not show significant abnormal return differences. Considering the look-ahead bias, caused by assuming stability of sustainability scores, the evidence on equal performance seems most likely. Name: Vinnie Vermeulen Student number: 10554955 Supervisor: Drs. P.V. Trietsch, M, Phil Faculty: Economics and Business Specialization: Economics and Finance Date: June 29, 2016Statement of Originality This document is written by Vinnie Vermeulen who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
TABLE OF CONTENTS 1. Introduction 3 2. Literature 6 2.1 What is sustainable investing? 6 2.2 How do investors rate sustainability? 7 2.2.1 Screening 7 2.2.2 MSCI ESG 8 2.3 What are the reasons for over- and underperformance of SRI funds? 9 2.3.1 Underperformance 9 2.3.2 Outperformance 10 2.4 How are abnormal returns measured? 11 2.4.1 Fama-French Four-Factor Model 12 2.5 Does previous empirical research indicate performance differences in the US? 13 2.5.1 No significant performance difference 15 2.5.2 Significant performance difference 16 3. Empirical Research 17 3.1 Portfolio Construction 17 3.2 Method 18 4. Results 19 5. Conclusion 21 6. References 23 7. Appendix 26 1. Introduction
Incorporating sustainability into the investment decision has grown rapidly over the past decades (Renneboog et al., 2008, p. 1723). According to the Social Investment Forum (2012) the sustainable assets under management in the US have increased from 639 billion in 1995 to over 3,7 trillion in 2012. This constant growth raises the question whether sustainable investments perform as least as good as conventional ones. If this is not the case, sustainable investors care less about financial performance than conventional ones and are compensated due to higher non-financial utilities. (Renneboog et al., 2008, p. 1723). Sustainable versus conventional stock performance differences have been examined multiple times over the period between 1995 and 2005. However little is know about performance differences in the last decade. Therefore the research question is as follows: ‘Is there a long term performance difference between sustainable stocks and non-sustainable stocks in the US between 2005 and 2015?’ Statman (2000, p. 38) compared the performance of mutual funds that incorporate sustainability screens with conventional funds and found out that sustainable (SRI) funds outperformed the conventional funds in the period between 1990 and 1998. However this difference was not significant. In the same period Bauer et al (2005, p 1766) found evidence of both out- and underperformance when doing research on sustainable mutual funds in the US, UK and Germany. Hong and Kacperczyk (2009, p. 35) investigated the companies that are excluded from the portfolio due to negative screening. They found out that these so-called ‘sin’ stocks have significant higher returns. Before continuing our empirical analysis, the following questions will be discussed to get an overview of the functioning of sustainable investment: - What is sustainable investing? - How do investors rate sustainability? - What are the reasons for over- and underperformance of SRI funds? - How are abnormal returns measured? - Does previous empirical research indicate performance differences in the US? To validate whether the prior researches are still useful today the time period between January 2005 and December 2015 in the US is being examined. To make a better comparison we retrieve our data from databases that were used (Kempf & Osthoff, 2007; Galema et al, 2008) investigating this subject namely:
- Social responsibility ratings are retrieved from MSCI ESG Research Inc. (former KLD Research & Analytics Inc.). - Stock prices and S&P 500 performance are measured monthly and obtained form Datastream. - One Month Treasury Bill rates, Small-Minus-Big (SMB), High-Minus-Low (HML) and Momentum (MOM) factors used in our regression are retrieved from Wharton WRDS. Using the social responsibility ratings, 30 high-rated companies are added to a sustainable portfolio and 30 are added to a portfolio with the lowest social responsibility ratings. To evaluate the portfolios performance we calculated its abnormal returns using the Fama and French (1993) Three-Factor model expanded with a correction for momentum, which Carhart added in 1997. We repeat the process of portfolio construction based on updated ratings to check whether renewing the portfolio, and holding it for the period between 2010-2015, changes this result. Both portfolios above normal returns are compared to the S&P 500 performance by alpha, which is the constant measured in the Four-Factor model. The efficient market hypotheses states that it is impossible to outperform the market in the long run and therefore we expect the alpha to be not significantly different from zero. A two-sided T-test will be used to compare the difference between both portfolios. Since we expected the constant in both portfolios to be zero, we do not expect abnormal returns between both portfolios to be significantly different from each other. However to further investigate the relationship between sustainability and returns a linear regression of a companies total sustainability scores on its average excess returns will be made to provide more information, which can be useful while forming a final conclusion. Before the hypotheses are tested, the current literature is described and previous empirical studies on performance differences will be discussed. Thereafter the method of portfolio construction and empirical test will be given. This leads to a presentation of results and a final conclusion.
2. Literature This chapter starts with a definition of sustainable investing followed by the motivations investors have for choosing a sustainable investment strategy. The second paragraph looks into the development of sustainable screening, which is an important part in sustainable stock selection. In section 2.3 previous theory is regarded to find reasons that could explain both positive and negative performance differences. After discussing the theory, the next section looks at the methods for measuring abnormal performance. Finally section 2.5 investigates whether previous research finds evidence of significant abnormal return differences between sustainable and conventional investing. 2.1 What is sustainable investing? According to Schueth, SRI can be defined as “the process of integrating personal values and societal concerns into investment decision-making.” This definition shows the dual objective of social investors, which is: making money, while making a difference. He mentions that the investors who choose a sustainable investment strategy can have two motivations that are often complimentary (2003, p. 190). The first group wants to put their money to work in a manner that is more closely aligned with their personal preferences. This group is often referred to as ‘feel good’ investors. The second group feels the need to invest in projects that improve the quality of life. They tend to be more interested in a social change of society. In the classical view (e.g. Adam Smith’s) on the economy with complete and competitive markets, everyone maximizes their own profits until the market is Pareto efficient. This means that within the classical approach social welfare is also maximized. (Reneboog et al., 2008, p. 1728). However a more modern view states that due to existence of externalities, profit maximization does not maximize social welfare (Jensen, 2001, p. 11). Externalities arise when the cost of a company are negatively affected by the actions of another company (Reneboog et al., 2008, p. 1728). For example a fishery’s catch is polluted by an upstream chemical plant increasing its production. Amy Domini states that: "Socially responsible investing can be a tool for dialogue between corporations and society." (Burton, 1998b, p. 48) SRI should eventually lead to higher social welfare.
2.2 How do investors rate sustainability? Section 2.2.1 discusses how investors select sustainable stocks using both positive and negative screens. Paragraph 2.2.2 gives more information on how MSCI ESG rates sustainability. 2.2.1 Screening In order to make a sustainable investment decision various types of screens are needed, to help select the stock that meets the investors needs. The Social Investment Forum (2003) reports that over 64% of SRI mutual funds in the US apply more than five different screens. These categories can be classified into two groups namely: negative screens and positive screens. The oldest form of making a sustainable investment decision is based on eliminating undesirable companies from the investment portfolio and is known as negative ethical screening (Bilbao et al., 2012, p. 10904). This concept can be dated back to the 17th century where the Religious Society of Friends prohibited its members from investing in slavery (Schueth, 2003, p. 198). The origin of modern day social investing is linked to the political climate in the 1960’s, where issues like civil rights and nuclear energy increased social awareness (Bauer et al., 2005, p. 1752). The negative screening method excludes companies that perform poor on social, environmental and ethical criteria such as gambling, animal testing and violation of human rights (Renneboog et al., 2008, p. 1728). After a negative screening process, investors make investment decisions based on financial and quantitative selection of the remaining stocks. Positive screening is based on including firms with excellent performance on social criteria (Reneboog et al., 2011, p. 564). Excellent performance is rather a difficult measure because some industries are more sustainable than others by definition e.g. the oil industry vs. financial sector. That is why investors often use a ‘best in class’ approach and rank firms within each industry. Firms can only be selected when a benchmark is passed (Renneboog et al., 2008, p. 1728). The third generation of social screening is referred to a combination of both positive and negative screens (Bilbao et al., 2012, p. 10904). The fourth generation is related to mutual funds and combines both positive and negative screens with shareholder activism (Reneboog et al., 2008, p. 1730). Scheuth mentions that screening decisions are never black and white, because there are no perfect companies. However
screening does help in selecting the companies that best meet the investors social criteria (2003, p. 190). 2.2.2 MSCI ESG Institutions collect and provide data to investors to get better insights in social performance. MSCI ESG Research Inc. is one of these companies and according to the most recent P&I ranking is used by 98 of the top 100 largest money managers. They provide an annual dataset of both positive and negative performance indicators. The seven categories MSCI investigates are: - Corporate governance - Community - Employee relations - Human rights - Environment - Diversity - Product Each category is divided into subcategories measuring either a strength or concern using a binary scoring method. If a company meets the benchmark criteria, the indicator scores a “1”, if it does not meet the criteria it receives a “0”, if no information is available it receives the signal “NR”. To give an insight on how benchmarks work we have a look at how two subcategories are scored. ‘Limited Compensation’ is part the strengths in the governance category and praises companies with low levels of compensations to its top management. Receiving a “1” score means the CEO earns less than $500,000 per year or $30,000 for outside directors. The counterpart of this category is ‘High Compensation’ and is considered a concern. The benchmark in this category is a total compensation of more than $10 million for the CEO or $100,000 for outside directors. A companies total sustainability score is the sum of its scores in each subcategory, where a strength adds one point and a concern subtracts one.
2.3 What are the reasons for over- and underperformance of SRI funds? This paragraph will look at previous studies that investigate whether shifting a business strategy towards more sustainability is related to over- or underperformance and how it will affect the investor.. At first the underperformance of sustainable investing will be investigated, thereafter the outperformance will be discussed. 2.3.1 Underperformance Friedman states that: “Few trends could so thoroughly undermine the very foundations of our free society as the acceptance by corporate offices of a social responsibility other than making as much money as possible” (1962, p. 133). According to Friedman corporate investment is based on maximizing profits while minimizing risk and to make this happen, companies should only care for the interest of their shareholders. Opponents of sustainable investing claim that implementing sustainability into a company leads to higher ‘extra’ cost and thereby neglect the responsibility of acting to the financial interest of shareholders (McWilliams & Siegel 2001, p. 123). Langbein and Posner argue that by using a negative screening process, financially attractive companies are excluded from the portfolio (1980, p. 73). Reneboog et al. confirm this assumption and state that excluding firms limits diversification opportunities and shifts the mean-variance frontier towards unfavourable risk-return trade-offs (2008, p. 1734). They illustrate the underperformance argument of SRI using the example shown in table 1. Table 1
Companies Positive NPV Negative NPV
Positive CSR (A) SRI and Conventional (C) SRI
Negative CSR (B) Conventional (D) Neither
The logic behind this example is that the investment opportunity is limited to four options. Option A generates a positive NPV to the investor and has positive effects on for example environment or community. Both sustainable as conventional investors choose this option because of the positive NPV, thereby sustainable investors also value the positive CSR. Option D generates a negative NPV and is therefore not adopted by
conventional investors. Sustainable investors do not include this option because of negative externalities. Performance differences arise because only conventional investors are willing to pick option B, which generates a positive return, but negatively influences CSR. On the other hand only sustainable investors choose the negative NPV-option, which generates positive externalities but is associated with financial underperformance. We need to nuance this example, because of the assumption that investment opportunities are limited. Sustainable investors prefer positive CSR, however it does not necessarily mean they are willing to choose a negative investment opportunity. Given the assumption that there is a broader selection of investment opportunities, investors pick the stock that maximizes return given variance and minimize the variance given expected return (Fama & French, 2004, p. 26). Only if the supply of sustainable investment opportunities is limited, the investor makes a trade-off between positive CRS and negative expected returns. Depending on both financial and non-financial utilities, it is most likely he chooses an investment that maximizes CSR while minimizing losses. 2.3.2 Outperformance According to Friedman investors only care about profits. If this is true, how come that sustainability is high on the business and political agendas (Lewis & Juravle, 2010, p. 483). Lazlow identifies two trends that help explain the interest in sustainability. At first society has higher expectations of companies in terms of social well-being, health and the environment. Becoming more energy efficient is not enough and reducing carbon emissions is already entering the mainstream. Secondly, new legislation like the California’s AB32 bill1 requires companies to reduce pollution and therefore become more sustainable (2008, p. 78). If we have a look at performance, companies may want to become more sustainable because event studies show that statements on the use of toxic chemicals negatively influence a firms stock price (Hamilton, 1995, p. 112). Besides punishment Klassen and McLaughlin (1996, p. 1212) showed that after receiving environmental performance awards, firms realize significant above normal returns. 1 California’s AB32 bill requires a 25% reduction in carbon dioxide emissions by 2020.
Statman argues that sustainable investors can raise the cost of capital of a non-sustainable company if its capital supply function is less than perfectly elastic. This will be the case if there is a lack of conventional investors willing to provide capital at the same cost (2000, p. 36). The efficient market hypothesis argues that it is impossible for sustainable stocks to outperform conventional ones and vice versa. Reneboog et al. bring forward two arguments on outperformance by SRI based on market inefficiencies. They suggest that SRI screening brings forward valuable information to the investor, which would not have been available otherwise. This information could improve picking the best stocks (2008, p. 1734). A second argument this research provides is built around the idea that stock markets misprice information on sustainability in the short run. Reneboog et al. argue whether there is a causal relationship between net present value and social responsibility. Regarding table 1, option B loses shareholder value in the long run by incurring cost to e.g. litigation, whereas on the long term, company C is likely to gain value because it is relatively under-priced compared to the negative CSR-options (2008, p. 1735). Lazlow supports this argument by stating that companies, delivering profits while destroying value for other stakeholders, have hidden liabilities and therefore higher future costs. On the other hand companies that find solutions to societal problems discover long run profit opportunities (2008, p. 80). Regarding different points of view discussed in previous theory, no answer has been found on whether a sustainable investing always preforms significantly better or worse than using a conventional investment strategy. 2.4 How are abnormal returns measured? Already having a look at table 2 shows that two models are frequently used investigating abnormal stock return differences between sustainable and conventional strategies. This paragraph answers how abnormal returns are measured and also has a look the difference between both models, which could explain the shift from using a single factor model to the multifactor model.
2.4.1 Fama-French Four-Factor Model Fama and French state that the capital asset pricing model is still frequently used in asset pricing due to its simplicity (2004, p. 25). However they introduced the Three- Factor model after finding empirical evidence on the inefficiency of the CAPM-framework (Derwall et al., 2004 ,p. 56). The SMB-factor corrects for differences between small-cap stocks versus large-cap stocks and the HML-factor measures the return difference between a value and a growth portfolio. The higher explanatory power of the multifactor model can be a motivation for abandoning the use of a single-factor model in previous research. Carhart expanded the Three-Factor model in 1997 with a momentum factor, because it failed to explain the momentum strategy discovered by Jegadeesh and Timan (1993). The monthly excess returns of the portfolios are calculated using the Four-Factor model as below. 𝑅!− 𝑅!" = 𝛼!+ 𝛽! 𝑅!" − 𝑅!" + 𝛽!𝑆𝑀𝐵!+ 𝛽!𝐻𝑀𝐿!+ 𝛽! 𝑀𝑂𝑀!+ 𝜀! where, 𝑅!= 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑖 𝑖𝑛 𝑚𝑜𝑛𝑡ℎ 𝑡 𝑅!" = 𝑜𝑛𝑒 − 𝑚𝑜𝑛𝑡ℎ 𝑈. 𝑆. 𝑇 − 𝐵𝑖𝑙𝑙 𝑟𝑎𝑡𝑒 𝑎𝑡 𝑡 𝑅!" = 𝑟𝑒𝑡𝑢𝑟𝑛 𝑜𝑛 𝑎 𝑣𝑎𝑙𝑢𝑒 − 𝑤𝑒𝑖𝑔ℎ𝑡𝑒𝑑 𝑚𝑎𝑟𝑘𝑒𝑡 𝑝𝑟𝑜𝑥𝑦 𝑖𝑛 𝑚𝑜𝑛𝑡ℎ 𝑡 𝑆𝑀𝐵!= 𝑟𝑒𝑡𝑢𝑟𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑎 𝑠𝑚𝑎𝑙𝑙 𝑎𝑛𝑑 𝑙𝑎𝑟𝑔𝑒 − 𝑐𝑎𝑝 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝐻𝑀𝐿! = 𝑟𝑒𝑡𝑢𝑟𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑎 𝑣𝑎𝑙𝑢𝑒 𝑎𝑛𝑑 𝑔𝑟𝑜𝑤𝑡ℎ 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑀𝑂𝑀! = 𝑟𝑒𝑡𝑢𝑟𝑛 𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑜𝑛 𝑎 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑜𝑓 𝑝𝑎𝑠𝑡 𝑜𝑛𝑒 − 𝑦𝑒𝑎𝑟 𝑤𝑖𝑛𝑛𝑒𝑟𝑠 𝑎𝑛𝑑 𝑎 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜 𝑜𝑓 𝑝𝑎𝑠𝑡 𝑜𝑛𝑒 − 𝑦𝑒𝑎𝑟 𝑙𝑜𝑠𝑒𝑟𝑠 𝜀! = 𝑒𝑟𝑟𝑜𝑟 𝑡𝑒𝑟𝑚 Β1 is interpreted as the portfolio’s exposure to market risk. The constant in this model measures the average abnormal return in excess of the return on the market proxy. This has our particular interest in testing whether sustainable stocks preform differently than conventional ones.
2.5 Does previous empirical research indicate performance differences in the US? Various empirical studies have analysed performance differences between a sustainable investment strategy versus a conventional one in the US. First the studies finding no significant performance difference are discussed. Secondly we have a look at previous studies investigating the same period and how they where able to find significant measures of out- and underperformance. The most important studies on the US are summarized and presented in Table 2.
Table 2
Study Period Country Method Sig. difference
Reneboog et al. (2011) 1992-2003 17 countries, including US 4-factor No Galema et al. (2008)
1992-2006 US 4-factor Community
portfolio significantly outperforms Kempf & Osthoff (2007)
1992-2004 US 4-factor Top rated
sustainable companies outperform Bauer et al. (2005) 1990-2001 US, UK, Germany 4-factor US (domestic) SRI funds underperform. No significant international difference Derwall et al. (2004)
1997-2003 US 4-factor Top rated
sustainable companies outperform Schroder (2004) 1990-2002 US, Germany, Switzerland 2-factor No Statman (2000) 1990-1998 US CAPM No Goldreyer et al. (1999) 1981-1997 US CAPM No Hamilton et al. (1993) 1981-1985 1986-1990 US CAPM No
2.5.1 No significant performance difference Hamilton et al. (1993) presented the first research that focussed on the US market. They compared the performance of 32 SRI funds to that of 320 randomly selected conventional ones. Jensen’s alpha is measured using the NYSE Index as a benchmark. Because the number of sustainable mutual funds increased from six in 1981 to 32 in a period of nine years, this research decided to divide the investigated period into two sub-periods. They concluded that the 17 SRI funds, being established before 1985, realized an abnormal return of -0.06% per month and outperformed their conventional competitors, which gained an alpha of -0.14%. However sustainable funds established after 1985 underperformed compared to non-SRI funds (-0.28% versus -0.04%). Take note that for both periods the abnormal excess returns were not significantly different from each other. Goldreyer et al. (1999) compared the performance of 49 SRI funds with a reference group of 180 non-ethical funds, matched by size and market beta. The research did not find significant differences between the performance of both investment strategies. They did however find significant evidence that SRI funds using positive screens (α=-0.11%) outperform the ones without (α=-0.81%). This evidence is based on a sample of 29 funds but supports the hypothesis that more screens improve the performance of sustainable investing. Barnett and Salomon also investigated the relation between the intensity of screening and returns. Based on 67 SRI funds, they found a curvilinear relationship, which declines at low intensity screening (1-4 screens) but increases until it reaches a maximum of 12 screens (2006, p. 1117). Barnett and Salomon hereby confirm the previous findings by Goldreyer et al. in 1999. Statman (2000) conducted a research for the period of 1990-1998, comparing the performance of 31 SRI funds with 62 non-SRI funds. Jensen’s alpha is measured against the S&P 500. The difference in average monthly alpha of SRI funds (-0.42%) and conventional ones (-0.62%) is not significant. Using the Domini 400 Social Index2 as a benchmark does not change this result. Schroder (2004) preformed an international SRI performance study that included 30 sustainable funds from the US, 16 from Germany and one from Switzerland. Abnormal performance is measured using a two-factor model and a combination of two 2 The DSI 400 is an index consisting of social responsible stocks using MSCI ratings.
indexes as a benchmark namely: a blue-chip index3 and a small-cap index. Alphas are ranging between -2.06% and 0.87%, with only 4 out of 46 being significantly negative at the 5% level. Hereby he confirms that SRI funds do not significantly differ from the benchmark. Another international study by Bauer et al. (2005) used the multifactor model, presented by Carhart in 1997, to control for size, book-to-market differences and momentum. He concluded that US domestic SRI funds preform significantly worse than conventional US domestic funds. US international fund differences however are insignificant. Having a look at other countries investigated, we see that alphas between sustainable and conventional funds in Germany do not significantly differ and in the UK both domestic and international sustainable funds outperform non-SRI funds. Also regarding the other countries, this research does not provide us with an answer explaining performance differences. 2.5.2 Significant performance difference Research by Derwall et al. (2004) investigated the relationship between environmental screens and stock returns. Using the most recent ‘eco-efficiency’ scores from Innovest two portfolios are created. Both the high- and low-rated portfolio are rebalanced at the end of June with renewed ratings. The portfolio consisting of high-rated firms significantly outperformed the low-rated environmental portfolio, which is in conflict with the efficient market hypothesis. A potential explanation states that environmental information is undervalued by financial markets (Reneboog et al., 2008, p. 1732). Kempf and Osthoff preformed a similar research in 2007. Using sustainability scores provided by KLD Research and Analytics two portfolios are created, one consisting of high-rated companies and the second of low-rated companies. Portfolios are restructured yearly using the most recent ratings. The alpha of the highest rated portfolio remains significantly higher even after taking into account for transaction costs. Galema et al (2008) investigated abnormal performance differences using the Fama-French Four-Factor Model, with and the S&P 500 as a benchmark, for the period between 1992 and 2006. Six strength and six concern portfolios based on the six categories investigated by KLD at that time, which are: 3 The Blue-Chip Index tracks the stock of top-preforming, publically traded companies and tends to have similar movements as the market.
- Corporate governance - Community - Employee relations - Human rights - Environment - Diversity Comparing each categories strength portfolio with its corresponding concern portfolio showed that only the community strength portfolio significantly outperformed its counterpart. However Hong and Kacperczyk (2009) investigated whether companies involved in alcohol and tobacco production outperform companies that are not excluded from the portfolio due to negative screening. They find that these ‘sin’ stocks are less held by pension funds, are relatively under-priced and have significant higher excess returns than the other stocks. An overall conclusion regarding previous research does not provide us with an answer on whether sustainable stocks have significant higher or lower abnormal returns. Different outcomes in previous research can be due to the use of different models or bias that is caused by fund managers wrongly applying screening methods. 3. Empirical Research Chapter 3 start with the information on how sustainable and conventional portfolios are constructed. This is followed by paragraph 3.2, where our research method is discussed and the hypothesis are given. 3.1 Portfolio Construction To investigate whether sustainable and conventional stocks perform differently, two portfolios are constructed using a third generation screening process on all S&P 500 stocks. Using the MSCI ESG data, explained in section 2.2.2, each company is rated on all seven categories and receives a total sustainability score based on the strengths and concerns in each subcategory. We do not emphasize categories in particular and therefore each is weighted equal. The sustainable portfolio consists of 30 companies with the highest sustainability score. 30 companies with the lowest sustainability score are added to the conventional
portfolio. The construction of both 10-year portfolios is based on the end of year ratings from 2004. Regarding the research by Derwall et al. (2004, p. 54), we assume that total sustainability scores are stable over time. Our own findings on stability of total scores are listed in Appendix 7.2. These data show that most scores are stable or slightly increasing. The increase could suggest that both high- and low-rated companies are becoming more sustainable over time. We repeat the process and construct two similar 5-year portfolios using the 2009 ratings and check whether changing the portfolio using updated ratings influences the outcome. These shorter termed portfolios should be less biased by the stability assumption. 3.2 Method To investigate whether the sustainable portfolio has different abnormal excess returns than a conventional portfolio the Fama-French Four-Factor model, discussed in section 2.5.1, is used. The higher explanatory power is preferred over the simplicity of use of the single-factor model. Our portfolios consist of stocks listed on the S&P 500, therefore this index is used as a benchmark. The one-month T-bill rate is used as the risk-free rate to compute excess returns. Using this information the following hypotheses will be tested. Hypothesis 1 𝐻0: 𝛼! = 𝛼! ‘𝐵𝑜𝑡ℎ 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜𝑠 𝑑𝑜 𝑛𝑜𝑡 𝑑𝑖𝑓𝑓𝑒𝑟 𝑖𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2005 − 2015’ 𝐻1: 𝛼! ≠ 𝛼! 𝐵𝑜𝑡ℎ 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜𝑠 𝑑𝑜 𝑑𝑖𝑓𝑓𝑒𝑟 𝑖𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2005 − 2015’ Hypothesis 2 𝐻0: 𝛼! = 𝛼! ‘𝐵𝑜𝑡ℎ 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜𝑠 𝑑𝑜 𝑛𝑜𝑡 𝑑𝑖𝑓𝑓𝑒𝑟 𝑖𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2010 − 2015’ 𝐻1: 𝛼! ≠ 𝛼! 𝐵𝑜𝑡ℎ 𝑝𝑜𝑟𝑡𝑓𝑜𝑙𝑖𝑜𝑠 𝑑𝑜 𝑑𝑖𝑓𝑓𝑒𝑟 𝑖𝑛 𝑎𝑏𝑛𝑜𝑟𝑚𝑎𝑙 𝑟𝑒𝑡𝑢𝑟𝑛𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 2010 − 2015’ From our hypothesis it is followed that we do not expect any difference in abnormal returns between sustainable and conventional portfolio. The relationship
between stock returns and sustainability scores could give more insight if 0-hypothesis are rejected and sustainable portfolios do differ in abnormal returns. A linear regression of sustainability scores on average excess returns is computed to show how a companies average excess return is influenced by the height of its sustainability score. To test this relationship the average excess returns are calculated from the 60 companies used to make our 10-year portfolios. These companies are used because they have the longest series of returns. Regarding hypotheses 1 and 2, we expected zero outperformance between sustainable and conventional portfolios in both periods. A possible explanation could be that sustainability scores have no effect on excess return. Therefore our third hypothesis is as follows. Hypothesis 3 𝐻0: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑙𝑖𝑛𝑒𝑎𝑟 𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛𝑠ℎ𝑖𝑝 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑒𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛 𝑎𝑛𝑑 𝑡𝑜𝑡𝑎𝑙 𝑠𝑢𝑠𝑡𝑎𝑖𝑛𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑠𝑐𝑜𝑟𝑒 𝐻1: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑙𝑖𝑛𝑒𝑎𝑟 𝑟𝑒𝑙𝑎𝑡𝑖𝑜𝑛𝑠ℎ𝑖𝑝 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑒𝑥𝑐𝑒𝑠𝑠 𝑟𝑒𝑡𝑢𝑟𝑛 𝑎𝑛𝑑 𝑡𝑜𝑡𝑎𝑙 𝑠𝑢𝑠𝑡𝑎𝑖𝑛𝑎𝑏𝑖𝑙𝑖𝑡𝑦 𝑠𝑐𝑜𝑟𝑒 4. Results Performance estimates obtained from our Four-Factor model are summarized in Table 3. First, all R-squared’s are above 0.88 and confirm the high explanatory power of the model. Secondly, none of the alphas is significantly different from zero. Hereby we can conclude that none of the portfolios is gaining above normal excess returns, which is in line with the efficient market hypothesis. The EMH says that, on average, it is impossible to beat the market by gaining above normal returns. The significant coefficients on SMB indicate that these three portfolios have a bias towards small-cap stocks and the significant HML coefficient indicates that these three portfolios are growth-stock orientated.
* Significant at the 1% level 2005-2015 N=131 Standard error ** Significant at the 5% level 2010-2015 N=71 in brackets *** Significant at the 10% level After describing the coefficients from our multifactor model we analyse the abnormal return difference between sustainable and conventional portfolios. Regarding the F-test on two variances, described in Appendix 7.3, we assume that for both periods variances do not differ. A T-test with equal variances is used to test the difference between corresponding alphas. Table 4 Period Sustainable – Conventional T-Value 2005-2015 -0.0007 -3.401* 2010-2015 -0.0002 -0.618 Table 4 shows underperformance of both sustainable portfolios compared to the conventional portfolios. This difference is only significant for the 10-year regressions. Both outcomes confirm that sustainable companies do not significantly outperform Portfolio Alpha Market SMB HML MOM R-square d Sustainable 2005-2015 0.0005 (0.001) 1.061* (0.033) 0.189* (0.067) 0.227* (0.068) -0.156* (0.034) 0.929 Conventional 2005-2015 0.001 (0.002) 0.966* (0.038) 0.0404 (0.077) 0.162** (0.078) -0.102* (0.039) 0.883 Sustainable 2010-2015 -0.001 (0.002) 1.027* (0.041) 0.188** (0.079) -0.063 (0.091) -0.161** (0.057) 0.921 Conventional 2010-2015 -0.001 (0.002) 1.144* (0.051) 0.342* (0.098) 0.377* (0.113) -0.190* (0.071) 0.915 Table 3
conventional ones. The outcome does not give an answer on whether high-rated stocks underperform or show no difference at al. Taking into account that the 10-year portfolios are likely to suffer more from look-ahead bias, caused by assuming that ratings are stable over a longer period, the evidence on zero differences seems stronger. The relationship between average excess returns and total sustainability score is shown in table 5. Table 5
Regression output Constant Sustainability Score R-squared Coefficient Standard error 0.0052* (0.0009) 0.00003 (0.0001) 0.0005 N=60 The outcome of our regression shows an R-squared that is close to zero, which means the model has no explanatory power. The table also shows no linear relationship between sustainability score and excess return. The bias caused by using the same sustainability score over a 10-year period probably influences the outcome and therefore more research in this field recommended. Based on this regression output no linear relationship is found. 5. Conclusion The constant growth of SRI raised the question whether sustainable investments performed differently than conventional ones. Using the Fama-French Four-Factor model the abnormal returns of both high- and low-rated portfolios in the US are measured. Regarding the regression output presented in Table 3 we can conclude that neither one of the portfolios performs significantly different than the S&P 500, which is in line with the efficient market hypothesis. The comparison between portfolios, presented in table 4, shows significant underperformance of the 10-year sustainable portfolio at the 1% level. This means we need to reject our first 0-hypothesis. The portfolios, based on the end of 2009 ratings, do not show significant differences between sustainable investing and using a conventional strategy. Thereby we do not reject our second 0-hypothesis. Both outcomes show that a
sustainable strategy does not outperform a conventional one in the long run. This argues the assumption of Reneboog et al (2008) and Lazlow (2008) that in the long run non-sustainable companies destroy shareholder value. It is difficult to say if sustainable portfolios have equal performance compared to conventional portfolios since we found evidence of underperformance of SRI in our 10-year portfolio but no significant difference between the 5-year portfolios. Taking into account that the 5-year portfolios are less influenced by look-ahead bias, the evidence on zero-performance differences in the US is stronger than the underperformance evidence. The zero performance difference is in line with the findings presented in previous empirical research by Hamilton et al. (1993), Goldreyer et al. (1999), Statman (2000), Schroder (2004) and Reneboog et al. (2011). Regarding the regression output of a companies total sustainability scores on its average excess return, discussed in table 5, shows that total sustainability scores have no effect on average excess returns and that the model has zero explanatory power . Based on the information found in this table we do not reject the third 0-hypothesis. This outcome does not help us explaining why the 10-year portfolio did show performance differences between the two strategies. Since no evidence on a linear relationship is found does not mean that sustainability scores do not affect returns. It is recommended that this relationship is investigated more in future research. First of all the sample is relatively small and only consists of high and low rated companies. Secondly the outcome is influenced by the same look-ahead bias, caused by assuming sustainability scores are stable. Updating the scores yearly gets rid of this bias. Using the scores in each MSCI ESG category instead of using a total sustainability score could also improve this regression, because this could show which of the categories have an effect on a companies return and which not. A recommendation on the multifactor model regressions is updating the portfolios yearly based on recent sustainability scores. This way the outcome is not influenced by look-ahead bias. Adding middle rated portfolios could also improve future findings in following research because that data is not investigated in this comparison.
6. References Barnett, M., & Salomon, M. (2006). The curvilinear relationship between social responsibility and financial performance. Strategic Management Journal, 27 (11), 1101-1122 Bauer, R., Koedijk, K., & Otten, R. (2005). International evidence on ethical mutual fund performance and investment style. Journal of Banking & Finance, 29, 1751-1767 Bilbao-Terol, A., Arenas-Parra, M., & Cañal-Fernández, M. (2012). A fuzzy multi objective approach for sustainable investments. Expert Systems with Applications, 39, 10904-10915 Derwall, J., Guenster, N., Bauer, R., & Koedijk, K. (2004). The eco-efficiency premium puzzle. Financial Analyst Journal, 61 (2), 51-63 Eurosif. (2014). European SRI Study 2014. Retrieved from http://www.eurosif.org/our- work/research/sri/european-sri-study-2014/ Fama, E. F., & French, K. R. (2004). The Capital Asset Pricing Model: Theory and Evidence. Journal of Economic Perspectives, 18 (3), 25-46 Fama, E. F., & French, K. R. (2011). Size, value and momentum in international stock returns. Journal of Financial Economics, 105, 457-472 Friedman, M. (1962). Capitalism and Freedom, Chicago: Chicago University Press. Galema, R., Plantinga, A., & Scholtens, B. (2008). The Stocks at Stake: Return and Risk in Socially Responsible Investment. Journal of Banking & Finance, 32 (12), 2646– 2654. Goldreyer, E., Ahmed, P., & Diltz, J. (1999). The performance of socially responsible mutual funds: Incorporating sociopolitical information. Journal of Financial Research, 28, 41-57
Hamilton, J. T. (1995). Pollution as News: Media and Stock Market Reactions to the Toxics Release Inventory Data, Journal of Environmental Economics and Management, 28, 98-113 Hamilton, S., Joe, H., & Statman, M. (1993). Doing well while doing good? The investment performance of socially responsible mutual funds. Financial Analysts Journal, 49 (6), 62-66 Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social norms on markets. Journal of Financial Economics, 93, 15-36 Klassen, R. D., & McLaughlin, C. P. (1996). The Impact of Environmental Management on Firm Performance. Management Science, 42 (8), 1199-1214 Kempf, A., & Osthoff, P. (2007). The Effect of Socailly Responsible Investing on Portfolio Performance. European Financial Management, 13 (5), 908-922 Langbein, J. H., & Posner, R. A. (1980). Social Investing and the Law of Trusts. Faculty Scholarship Series, 79 (72), 72-112 Lazlow, C. (2008). How the worlds leading companies are doing well by doing good, Stanford: Stanford University Press Lewis, A., & Juravle, C. (2010). Morals, Markets and Sustainable Investments: A Qualitative Study of ‘Champions’. Journal of Business Ethics, 93, 483-494 McWilliams, A., & Siegel, D. (2001a). Corporate social responsibility: A theory of the firm perspective. Academy of Management Review, 26, 117-127 Renneboog L., ter Horst J., & Zhang, C. (2008). Socially responsible investments: Institutional aspects, performance, and investor behaviour. Journal of Banking and Finance, 32, 1723–1742.
Renneboog, L., Ter Horst, J., & Zhang, C. (2011). Is ethical money financially smart? Nonfinancial attributes and money flows of socially responsible investment funds. Journal of Financial Intermediation, 20, 562-588 Sachs, J.D., & Reid, W.V. (2006). Investments Toward Sustainable Development. Sciencemag, 312,1002 Schueth, S. (2003). Socially Responsible Investing in the United States. Journal of Business Ethics, 43, 189-194 Social Investment Forum, (2002, 2012). Report on sustainable and responsible investing trends in the US. Retrieved from http://www.ussif.org/content.asp?contentid=82 Statman, M. (2000). Socially Responsible Mutual Funds. Financial Analyst Journal, 56 (3), 30-39
7. Appendix Appendix 7.1.1 to 7.1.4 are the estimates of our multifactor model regressions using STATA. Appendix 7.1.1 High-rated portfolio 2005-2015 Appendix 7.1.2 Low-rated portfolio 2005-2015
Appendix 7.1.3 High-rated portfolio 2010-2015 Appendix 7.1.4 Low-rated portfolio 2010-2015
Appendix 7.2 Yearly total sustainability scores in this table show that the stability assumption of Derwall et al. (2004) is reasonable in our research. Numbers 1 to 5 are the best rated companies and numbers 6 to 10 are the lowest rated companies. Nr. Company 2004 2005 2006 2007 2008 2009 1 Intel 10 10 11 15 14 14 2 HP 9 11 14 12 12 12 3 Xerox 8 7 11 13 13 13 4 Avon Products 7 6 7 7 7 7 5 Southwest Airlines 7 8 6 6 5 5 6 First Energy -10 -9 -8 -8 -6 -6 7 Concophillips -9 -8 -8 -6 -6 -6 8 Dynegy -8 -8 -7 -7 -8 -8 9 Exxon Mobil -7 -9 -10 -9 -10 -10 10 Reynolds American -7 -6 -7 -5 -2 -2
Appendix 7.3 The following F-test is used to check if portfolios differ in variance with 95% certainty: 𝐹 = 𝑠!"#$%&'%()*! 𝑠!"#$%#&'"#()! Appendix 7.4 Linear regression estimates of a companies total sustainability score on its average excess return using EXCEL. Portfolio Standard Deviation Alpha N F-Value Significant at 95% Sustainable 2005-2015 0.0014 131 0.7657 No Conventional 2005-2015 0.0017 Sustainable 2010-2015 0.0017 71 0.6525 No Conventional 2010-2015 0.0021
