The impact of screening and portfolio ethicality on socially responsible investment fund performance

Transcription

1 The impact of screening and portfolio ethicality on socially responsible investment fund performance Abstract This paper investigates the relation between the ethicality of portfolios and the fund performance of both socially responsible investment (SRI) and conventional mutual funds. Specifically, we find that the increase in environmental, social and governance scores leads to higher subsequent financial performance in both SRI and conventional funds while the exclusion of sin stocks from a portfolio results in lower subsequent period returns. The results of the impact of ethical screens on the ethicality of SRI funds show that the screens employed by SRI funds positively affect the ethicality of SRI funds. Keywords: Socially responsible investment; SRI; ESG; Mutual funds; Screening intensity; Ethicality of portfolio JEL classification: G11; G23 1

2 1. Introduction Over the past decade, socially responsible investment (SRI) has experienced significant growth and has attracted considerable interest around the world. The US Social Investment Forum (SIF) defines SRI as an investment process that considers the social and environmental consequences of investments, both positive and negative, within the context of rigorous financial analysis. Advocates of SRI believe that the only way for businesses to achieve sustainable growth is to consider all facets of operations, including environmental, social and governance (ESG) factors. Given that an increasing number of firms are integrating ethical considerations into investment decision making, SRI is gradually becoming a mainstream investment vehicle in many developed countries. In the US, professionally managed SRI assets expanded from $3.74 trillion at the start of 2012 to $6.57 trillion at the start of 2014, an increase of 76% (SIF, 2014). Increasing demand has resulted in the growth of SRI mutual funds (SRI funds hereafter). In, Kim, Park, Kim, and Kim (2014) show that the increase in competition due the recent growth in the number of SRI funds has had a positive impact on the performance of SRI funds. One implication of this finding is that SRI funds are continuously attracting new investors by differentiating themselves from conventional funds. The most distinctive feature of SRI funds compared to conventional funds is that SRI funds integrate investors ESG concerns as well as personal values into their investment decisions. SRI funds are largely classified into three different groups, based on how securities are selected, namely, by shareholder advocacy, community investments, and investments with screening criteria. Among the three different types, the screening approach has been the most popular so far in the US. 1 Many studies define an SRI fund as a fund that incorporates screening in their stock selection. For example, Renneboog, Horst, and Zhang (2011) consider SRI funds if their names are related to the terms ethical, socially responsible, ecology, Christian value, or Islamic. The authors also use the definition of Standard & Poor s, which classifies an SRI fund as one whose prospectus specifies social, environmental, and ethical investment goals. Nofsinger and Varma (2014) employ a similar method and find that SRI funds use 1 According to SIF (2014), investments with a screening process were recorded as having assets under management of $2.51 trillion, followed by shareholder advocacy and community investing, with approximately $1.5 trillion and $41 billion, respectively. In addition, overlapping portfolios exist that use a combination of ESG incorporation, shareholder advocacy, and community investing, with assets under management of around $981 billion. 2

3 different screenings including product-related ESG screens. 2 The screening process results in different levels of ethicality between SRI and conventional mutual funds, as measured by the ESG score. 3 Kempf and Osthoff (2008) report that SRI funds have significantly higher ESG scores and conclude SRI funds are not conventional funds in disguise. This distinct feature of SRI funds leads studies to examine the difference in performance and risk between SRI and conventional funds. Although not conclusive, the general finding is either the underperformance of SRI funds or a lack of difference in performance between SRI and conventional funds in the US market. 4 From a risk perspective, Rudd (1979) argues that restrictions on an investment opportunity set could lead to inefficient diversification and increase the non-systematic risk of an investment portfolio. However, Bello (2005) finds that SRI funds show no difference in terms of the characteristics of assets held, portfolio diversification, and the impact of diversification on performance. Given the backdrop, the purpose of this study is to measure the level of ethicality of SRI and conventional funds, and examine the relation between ethicality and fund performance. Accordingly, this study has two main contributions. First, our study utilizes two measures of fund ethicality, namely, ESG scores and controversial business involvement (CBI) scores, to capture the different dimensions of fund portfolios ethicality. Specifically, we begin by calculating the ESG and CBI scores of the individual firms in each fund s portfolio. We then aggregate these two measures of ethicality at the fund level by calculating the value-weighted average of ESG and CBI scores for the fund portfolios. This approach enables us to gauge the actual ethicality of fund portfolios in the two dimensions (ESG and CBI) for both SRI and conventional funds which can be directly compared. The division of a fund s ethicality into these two dimensions is crucial in explicating the disagreement on SRI fund performance since the two measures have different impacts on fund performance. Empirical studies at the individual firm level suggest that while firms 2 Nofsinger and Varma (2014) define product-related screens as restricting investment in firms that produce certain products related to alcohol, tobacco, gambling, weapons, nuclear technology, pornography, abortion or animal testing. 3 The ESG score is constructed as the value-weighted average of a portfolio s strength and concern scores in ESG issues; the mean value of the concern score is then subtracted from the strength score. The precise definition of the strength and concern scores, plus details of the calculation of the ESG score are provided in Section 4. 4 No performance difference is found by Hamilton, Jo, and Statman (1993), Goldreyer and Diltz (1999), Statman (2000), Bauer, Koedijk, and Otten (2005), and Bello (2005) amongst others. Girard, Rahman, and Stone (2007) find evidence of SRI fund underperformance. 3

4 with higher ESG scores produce superior abnormal returns (see, e.g., Derwall, Guenster, Bauer, & Koedijk, 2005; Kempf & Osthoff, 2008; Statman & Glushkov, 2009), firms classified as sin stocks (i.e., firms with higher CBI scores) also provide higher expected returns (Hong & Kacperczyk, 2009). Since ESG-related screens are aimed at increasing ESG scores while exclusionary screens are designed to decrease CBI scores, these findings indicate that ESG-related screens and exclusionary screens have opposite effects on fund performance. ESG-related screens refer to environmental, social and governance related screens, either negative or positive. Exclusionary screens refer to the process of excluding securities from the investment opportunity set based on certain criteria, such as excluding sin industries. Theoretically, the implementation of ESG-related screens should increase the ESG score of the fund while the use of exclusionary screens should decrease the CBI score of the fund. Prior studies have also utilized ESG scores to measure the ethicality of SRI and conventional funds. For example, Kempf and Osthoff (2008) calculate ethical rankings by first averaging the ratings of subcategories of the MSCI STATS database and then calculating the aggregate ranking for each fund. The difference between our calculation of ethical rankings and Kempf and Osthoff (2008) is that the latter normalizes the portfolio weights to sum up to one. Since data on reported portfolio holdings are not always complete, simply normalizing the portfolio weight might distort the actual ethicality of funds. For example, if a fund has a total reported weight of 30%, the normalization to 100% might not reflect the true ethicality of this fund. Hence, in our study, we use funds that have less than 10% difference between the total equity position of each fund and the total weight reported. In this way, we assume that cash and bond positions do not contribute to the ethicality of funds. Kempf and Osthoff s (2008) approach also gives more weight to the social category of the total ESG score than the environmental and governance categories since the MSCI STATS database has five subcategories under the social category. For comparison purposes, Kempf and Osthoff (2008) rank the SRI and conventional funds according to their aggregate rankings. However, the comparison of rankings between the two fund groups might not be appropriate for drawing a conclusion on the actual difference in the ethicality of portfolios since the rankings do not take into account the magnitude of ESG scores. For example, a ranking difference between two funds in the first quartile could be different from that in the third quartile. 4

5 Our second contribution is that by appropriately calculating the ESG and CBI scores of SRI and conventional funds, we are able to examine the effect of screening on ethicality and determine how the value-relevant information contained within the two measures of ethicality impacts on fund performance. Despite SRI funds lack of performance difference (or even underperformance) relative to conventional funds, the SRI market continues to grow while the conventional fund market has shrunk. Since SRI fund managers incorporate ESG and product-related concerns into their investment decisions, the ethicality of SRI portfolios is expected to be higher than that of conventional fund portfolios. One of the critical assumptions in the studies that examine the relation between screening intensity and fund performance is that the screening process directly impacts on fund performance. The logic behind this assumption is that an increase in screening intensity leads to a decrease in the investment opportunity set which deteriorates potential diversification effects. In our study, we show that it is the value-relevant information contained in the ethicality of fund portfolios that has a direct impact on fund performance. This issue is empirically important because one of the reasons SRI investors invest in SRI stocks is to obtain utility from non-financial factors. Thus, if the ethicality of SRI funds does not differ from that of conventional funds, then SRI s slogan of doing well by doing good would be somewhat tarnished and could be regarded as another type of green-washing in the context of fund management. 5 The remainder of this paper is organized as follows. Sections 2 and 3 describe the theoretical background and hypothesis development, respectively. Section 4 discusses the sample data and the measures of portfolio ethicality. Section 5 presents the methodologies used in this paper and Section 6 provides our empirical findings. Section 7 concludes the paper. 2. Theoretical background 2.1. Related literature Although no consensus has been formed on the performance of SRI funds relative to that of conventional funds, a few studies have examined the relation between screening 5 Walker and Wan (2012) define green-washing as the discrepancy between the substantive actions on environmental issues and the symbolic actions. 5

6 intensity and fund performance. Barnett and Salomon (2006) investigate how screening intensity affects the financial performance of SRI funds and find a U-shaped relation between screening intensity and performance. Renneboog, Horst, and Zhang (2008) and Lee, Humphrey, Benson, and Ahn (2010) argue that screening intensity negatively affects performance because non-financial screens restrict investment opportunities, reduces diversification efficiency, and thereby adversely affects performance. However, these two studies do not find a U-shaped relation. Humphrey and Lee (2011) use negative and positive screens as well as overall screening intensity as regressors and find no impact of screening intensity on fund performance in Australian SRI funds. One of the critical assumptions of the above studies is that the diminished investment opportunity set due to the implementation of social screens is directly linked to fund performance. The logic behind this argument is that an increase in screening intensity leads to a decrease in the investment opportunity set which in turn results in the deterioration of potential diversification effects. Consequently, SRI portfolios are systematically subject to idiosyncratic risk which worsens the risk-adjusted performance of SRI funds. On the other hand, stakeholder theory suggests that the screening process might be able to identify stocks with superior returns and help avoid stocks that have poor stakeholder relations (Cornell & Shapiro, 1987; Fombrun, Gardberg, & Barnett, 2000). However, a constrained investment opportunity set does not necessarily imply that the actual portfolios of SRI funds are highly ethical. There is also no guarantee that the intense use of screens for any SRI fund would effectively produce a highly ethical investment opportunity set or actual portfolio. Figure 1 illustrates this argument. Hence, it is important to investigate whether SRI fund screening processes increase the degree of ethicality of their portfolios since SRI fund investors not only require financial returns but also have an interest in ethical investment. [Insert Figure 1 here] There has been little attempt to examine the actual role of screens in socially responsible investments. The purpose of screens can be understood in two ways. The first purpose is to promote ethical behaviour and to invest in more socially responsible companies so that the company could have better finance resources or lower cost of capital (Ghoul, Guedhami, Kwok, & Mishra, 2011). The other purpose is that SRI fund managers might 6

7 use the screening process to maximize the wealth of investors. If managers believe that ESG information is value-relevant, they would exert an effort to construct portfolios that have high ESG scores while balancing the costs of sacrificing diversification. Thus, the degree of ethicality of fund portfolios is a result of the implementation of time-invariant screens. In our study, we focus on the relation between the ethicality of fund portfolios with fund performance. Since the measures of ethicality employed in this paper (ESG and CBI scores) can be constructed for both SRI and conventional funds, we are able to compare how the ethicality of both types of funds impacts on fund performance. Studies that examine the relation between screening intensity and fund performance do not consider the different impacts of exclusionary screens and ESG-related screens on fund performance. Hong and Kacperczyk (2009) report that a portfolio of sin stocks earns 3.5% more abnormal return than a portfolio with comparable stocks in the US. This result implies that an increase in the number of exclusionary screens would have a negative impact on fund performance since it would reduce the number of sin stocks in the fund portfolio. On the other hand, there is evidence that ESG scores bear value-relevant information since recent studies have shown that ESG scores have a positive impact on fund performance. Derwall et al. (2005) find that portfolios with high eco-efficiency scores outperform those with lower scores by 6% per annum. Similarly, Kempf and Osthoff (2007) report that a strategy of buying responsible stocks and selling irresponsible stocks produces abnormal returns of up to 8.7% per annum. Nofsinger and Varma (2014) propose that SRI portfolios perform better during financial crisis periods at the expense of underperformance during non-crisis periods, in which investors who demand downside protection would find merit. Statman and Glushkov (2009) study the returns of stocks rated on social responsibility during They find that the expected returns of stocks of socially responsible companies are higher than those of conventional companies. Similarly, Edmans (2011) analyzes the relation between employee satisfaction and long-run stock returns and finds that firms with high employee satisfaction earn significantly higher abnormal returns than industry benchmarks, even after controlling for firm characteristics and industries. We introduce two measure of ethicality: ESG and CBI scores. Although these two dimensions of portfolio ethicality are not mutually exclusive, they are the result of different screening processes. ESG scores are constructed from ESG-related screens and CBI scores 7

8 are constructed from exclusionary screens. SRI funds that apply ESG-related screens aim to either invest in stocks with high ESG performance (positive ESG screens) or not invest in firms with poor ESG performance (negative ESG screens). Theoretically, the use of ESG-related screens should increase the ESG scores. Exclusionary screens are those that exclude securities from the investment opportunity set based on certain criteria. Although negative ESG screens and exclusionary screens may overlap to some extent, they are not identical. For example, many socially responsible investors filter out tobacco-related businesses in their investment pool (an exclusionary screen), but this screen is not related to ESG screens. To reduce the chance of overlap, during the construction of CBI scores, we restrict our focus to businesses related to tobacco, alcohol, gambling, firearms, and military or nuclear operations. These industries are less likely to be related to ESG screens Predictions from the corporate social responsibility literature The mean variance efficient portfolio theory introduced by Markowitz (1959) suggests that a portfolio s risk-return relation can be described by an asset pricing model. The cornerstone of modern asset pricing theory is that the value of an asset is equal to the expected discounted payoff under the assumption that individuals aim to maximize economic utility. Fama and French (2007) point out that the assumptions of standard asset pricing models are unrealistic and show how disagreement and tastes for assets as consumption goods can alter asset prices. In Fama and French s (2007) framework, socially responsible investors can be regarded as misinformed (or as those who have tastes for assets). Their model predicts that the impact of misinformed investors on asset prices is large if (1) the amount of invested wealth is substantial; (2) they are misinformed or have a taste for many assets; (3) their portfolios differ from the market portfolio; and (4) the returns of assets demanded by misinformed investors are not highly correlated with those of the assets they underweight. More specifically, if socially responsible investors have sizable assets under management, the expected returns of socially responsible stocks will be lower than that of the assets the investors underweight (i.e., irresponsible stocks). This taste-based hypothesis has been empirically examined by Hong and Kacperczyk (2009). The authors find that sin stocks are underpriced and produce positive abnormal returns after controlling for traditional 8

9 risk factors. They also note that this effect is driven by institutional investors, who are restricted from investing in sin stocks by social norms. Heinkel, Kraus, and Zechner (2001) develop a calibrated equilibrium model in which exclusionary investment strategies can reduce risk-sharing opportunities for polluting firms, which leads to a share price drop. This limited risk-sharing hypothesis also predicts that the expected returns of irresponsible stocks are higher than those of responsible stocks to compensate for the limited risk-sharing opportunities of irresponsible stocks. However, the implicit assumptions of the hypothesis, that the share of socially responsible investors is sufficiently large and that the socially responsible investors are homogeneous in exclusionary investment strategies, need further investigation. In contrast to the aforementioned two hypotheses, Derwall, Guenster, Bauer, and Koedijk (2011) present a competing hypothesis called the errors-in-expectations hypothesis. The idea is that corporate social responsibility (CSR) contains value-relevant information and financial markets do not completely reflect the aspects that could create opportunities for socially responsible investors to generate abnormal returns. Derwall et al. (2011) explain why the market might not fully understand the value of CSR. First, since CSR is a multidimensional and partially subjective concept and its appropriate measurement is difficult for investors, it is challenging to examine the relation between CSR and firm fundamental value. Second, no accounting standards have formally used CSR, so there are no sound evaluation tools to measure the value added of CSR to firm value. Consequently, socially responsible stocks have higher risk-adjusted returns because the market is slow to recognize the positive impact that strong CSR practices have on companies expected future cash flows. 3. Hypotheses development Since SRI funds incorporate social and environmental factors when making their investment decision, the ethicality of SRI funds should differ from that of conventional funds. However, the empirical evidence in the literature is mixed. Kempf and Osthoff (2008) provide evidence that SRI funds are not analogous to conventional funds since SRI funds exhibit higher ESG rankings than their counterparts do. On the contrary, Utz, Wimmer, Hirschberger, and Steuer (2014) find that SRI and conventional funds do not differ in terms 9

10 of mean and maximum ESG scores, although SRI funds have slightly larger minimum scores. If a difference in ethicality exists between the two groups of funds, then it should be attributable to the screening process imposed by the SRI funds. If no difference exists, it could be because either SRI funds screening processes do not help construct socially responsible portfolios or conventional funds also consider ESG factors in their investment process. For example, conventional fund managers could exclude firms with low ESG scores from their opportunity set since these firms could be subject to higher litigation risk and capital costs. Therefore, it is important to determine whether the screening processes used by SRI funds actually improve their ESG scores. Since exclusionary screens ban industries or companies involved in products related to tobacco, alcohol, gambling, firearms, and military and nuclear operations from the investment universe, the use of more exclusionary screens is expected to decrease portfolios involvement in controversial businesses. Similarly, the use of ESG-related screens should increase SRI funds ESG scores since SRI fund managers are inclined to invest in stocks with high ESG score (positive screens) and avoid stocks with low ESG scores (negative screens). Based on the aforementioned discussion, our hypotheses are as follows: Hypothesis 1a: SRI funds have higher ESG scores than conventional funds. Hypothesis 1b: Exclusionary screens have a negative impact on SRI funds CBI scores. Hypothesis 1c: ESG-related screens have a positive impact on SRI funds ESG scores. While the impact of CSR on firm value has been examined theoretically and empirically at the individual firm level, there has been little attempt to investigate such a relation at the fund portfolio level. Theoretical models (Fama & French, 2007; Heinkel et al., 2001) predict that the use of exclusionary screens will lead to a decrease in fund portfolios CBI scores which in turn reduces fund performance. Other empirical studies argue that fund portfolios with higher ESG scores outperform those with lower ESG scores (Kempf & Osthoff, 2007). In reality, SRI funds may combine different investment screening strategies, including the use of exclusionary and ESG-related screens, to cope with various investor demands. This procedure results in different ESG and CBI scores across SRI funds. Accordingly, whether the ethicality of fund portfolios leads to a performance difference 10

11 between SRI and conventional funds is an empirical question. Although conventional funds do not explicitly employ a screening process, their portfolios still have an inherent degree of ethicality which can be captured by ESG and CBI scores if these measures contain value-relevant information. Prior studies indicate that SRI funds do not underperform conventional funds (Goldreyer & Diltz, 1999; Hamilton et al., 1993; Statman, 2000), except for Girard et al. (2007), who do find evidence of underperformance. However, these studies do not directly measure the relation between fund ethicality and performance. By measuring the ethicality of both SRI and conventional funds, our study aims to fill this gap in the literature. Accordingly, we state the following related hypotheses: Hypothesis 2a: There is a positive relation between a fund s CBI score and subsequent performance. Hypothesis 2b: There is a positive relation between a fund s ESG score and subsequent performance. Several event studies have shown that stock prices react differently to positive and negative CSR news. Most recently, Krüger (2015) shows that the market response to environmental news is asymmetric. The author finds that the positive impact on stock prices followed by positive news is smaller than the negative effect of bad news. The author attributes the decrease in firm value in response to bad environmental news to the legal penalties imposed by the government. The author also finds a statistically insignificant relation between positive news and firm stock return. Based on these empirical findings, Derwall et al. (2011) conclude that while investors may be fully aware of the negative impact of low CSR on future cash flows, they may not be aware of the positive impact of high CSR. When we decompose the future cash flows of a firm into normal cash flows and cash flows related to good and bad CSR practices, the stock price at time zero can be described as follows: P 0 = t=0 Cashflow t + CSR good t CSR bad t (1 + r) t, (1) where P 0 is the stock price at time zero, Cashflow t is a firm s cash flow unrelated to its CSR practices, and CSR good t and CSR bad t are a firm s cash flow associated with good and 11

12 bad CSR practices, respectively. Given the empirical evidence that CSR good t is generally insignificant and has a small impact, if the stock market does not account for value-relevant CSR information to determine stock prices, it could overestimate a stock s future cash flows which would lead to lower future expected returns on unethical portfolios. When fund managers apply sophisticated screening processes to minimize negative cash flows from bad CSR practices and maximize cash flows from good CSR activities, their stock prices will be higher than expected by the market which results in better performance of ethical portfolios. Hence, increases in the ESG scores of portfolios increases the probability of CSR adding positive value to fund portfolios. Therefore, if a fund manager uses effective screening processes to select firms with good CSR practices (measured by the strength of ESG scores) and rule out those with bad CSR practices (measured by the concern of ESG scores), then it is likely that that good CSR practices are capitalized and creates a positive impact on fund performance, while poor CSR practices may not be priced which could lead to a smaller impact on fund performance. Thus, our hypotheses are stated as follows: Hypothesis 3a: There is a positive relation between the strength of ESG scores and fund performance. Hypothesis 3b: There is no relation between the concern of ESG scores and fund performance. 4. Data 4.1. SRI funds To obtain a list of US SRI mutual funds, we use multiple sources including Morningstar, US SIF reports from 1999 to 2010, and the SRI World Group. 6 A fund is classified as an SRI fund if it is listed by at least one of these sources. Following Nofsinger and Varma (2014), we also hand-collect some missing funds by searching keywords that frequently appear in the names of SRI funds and verify whether the missing funds were, in fact, SRI 6 The US SIF reports are available at and the list of SRI funds provided by the SRI World Group is available at The list has been used by other SRI researchers, including Nofsinger and Varma (2014). 12

13 funds, using their prospectuses and websites. 7 We then exclude balanced, bond, money market, stock index, and international equity funds and focus our analysis on domestic US equity funds, whose holding information we obtain from the Center for Research in Security Prices (CRSP) database. After checking each individual fund s prospectus, we selected all mutual funds that did not explicitly specify the incorporation of ethical screens as a reference group. Our total sample consists of 300 SRI funds. The return data were obtained from the CRSP Survivor-Bias-Free US Mutual Fund Database Ethicality measures To evaluate the degree of portfolio ethicality, we first construct firm-level ESG scores using data from MSCI STATS, formally known as the KLD database, which provides the annual ESG ratings of over three thousand publicly traded firms in the US. The database consists of three major ESG categories (environment, social, and governance) as well as six controversial business involvement indicators. 8 The social category has five subcategories including community, human rights, employee relations, diversity, and customers. Except for the business involvement indicators, each measure has strength (positive) and concern (negative) ratings. For example, when a firm is environmentally friendly (environmentally unfriendly), the strength (concern) measure is given a value of one. There are a number of different ways to construct a firm-level ESG score, but we employ an approach analogous to that of Deng, Kang, and Low (2013). The MSCI STATS database itself provides the aggregate ratings in seven categories by simple summation, however, this approach has a drawback. 9 Since the number of strength and concerns in each category varies over time, a comparison of aggregate ratings across years and dimensions is not meaningful. Similar to Deng et al. (2013), we construct a measure of ethicality for individual stocks by dividing the strength and concern measures of each category by the respective number of strength and concern indicators within that category. We then take the difference between the adjusted total strength score and the adjusted total concern score. Finally, the overall ESG score of the firm is calculated by taking the 7 We use the same keywords as Nofsinger and Varma (2014): Social, socially, environment, green, sustainability, sustainable, ethics, ethical, faith, religion, Christian, Islam, Baptist and Lutheran. 8 This includes alcohol, gambling, firearms, military, nuclear power, and tobacco. These businesses are typically excluded from the investment opportunity set for socially responsible investors by exclusionary screens. 9 These categories refer to the environmental, governance and five social subcategories. 13

14 average across the three major ESG categories. To calculate the firm-level CBI score, we simply take an equal-weighted average of the controversial business involvement indicators. After computing the firm-level ESG scores, we match the MSCI STATS data with the CRSP s fund holding information. We then construct equal-weighted and value-weighted ESG and CBI scores for both SRI and conventional funds. The value-weighted approach is based on the weights of securities in a fund s portfolio at the time when the ESG and CBI scores are calculated. When a portfolio has higher ESG (CBI) scores than other funds, the portfolio is considered more ethical (unethical). Note that the CRSP mutual fund holding information is largely incomplete from 2001 to 2002 and some funds do not have complete stock holding data, even after the first two years. 10 Hence, to ensure that the ethicality of fund portfolios is calculated accurately, we only include funds that have at least 90% of stock holding information available for the fund s equity position. After calculating the fund-level ESG and CBI scores, we match fund characteristics from the CRSP database with the fund-level ESG and CBI scores which includes net fund flow (F low), total asset value (T NA), family total asset value (F amily T NA), fund age (Age), expense ratio (Expense), and turnover ratio (F.T urnover). In addition, we calculate fund portfolio characteristics as the value-weighted averages of individual stock characteristics in a mutual fund s portfolio. Specifically, we calculate bookto-market ratios (BM), firm size (Cap), leverage (Leverage), dividend yields (DivY ield), return on assets (ROA), cash flow volatility (CF V olt), return volatility (RetV olt), share turnover (T urnover) and the Amihud s (2002) illiquidity ratio (Illiquidity). The bookto-market ratio is the book value of equity divided by the market capitalization. Similar to Fama and French (1993), we calculate book value at the end of the previous fiscal year as the sum of common stockholder s equity, deferred taxes and investment credits. Firm size is the market capitalization measured in billions of dollars. Leverage is represented by the debt-to-equity ratio. Dividend yield refers to the annual percentage dividend yield calculated as a stock s annual dividends (split-adjusted) divided by price (split-adjusted). Return on asset is the net income divided by average total assets. Similar to Zhang (2006), cash flow volatility is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum three years), scaled by the 10 Our sample starts from the year 2000, but due to the lack of portfolio holding information, our analysis covers the period from 2003 to

15 average total assets. Return volatility is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. Share turnover is calculated as the trading volume divided by shares outstanding over the last calendar year. A higher turnover potentially indicates greater liquidity. Following Amihud (2002), the illiquidity ratio is calculated as the daily absolute return divided by its trading volume, and multiplied by a thousand for scaling purposes. A higher value for the illiquidity ratio indicates low liquidity. 5. Methodologies We first examine the difference in ESG scores between SRI and conventional funds (hypothesis 1a). To do this, we begin by calculating the average ESG scores across SRI and conventional funds. The mean and median ESG scores are then compared using Wilcoxon/Mann-Whitney tests. Next, we run multivariate regressions to investigate whether the exclusionary and ESG-related screens used by SRI funds affect the CBI and the ESG scores, respectively (hypothesis 1b and 1c). Accordingly, we estimate the following models: CBI i,t = α + β 1 Exclusionary i + β 2 F und characteristics i,t + β 3 T ime dummies + ε i,t, (2) ESG i,t = α + β 1 ES i + β 2 SS i + β 3 GS i + β 4 F und characteristics i,t + β 5 T ime dummies + ε i,t, (3) where CBI i,t is the CBI score of the SRI fund; Exclusionary i is the intensity of exclusionary screens used; and ESG i,t is the total ESG score of the SRI fund. ES i, SS i and GS i are dummy variables that have a value of one if an SRI fund employs environmental, social and governance screens, respectively, and zero otherwise. 11 T ime dummies are included to control for time variation of the dependent variable. F und characteristics i,t include a fund s total net asset value, fund family total net asset value, age, expense and turnover ratios. Next, we run pooled OLS regressions to examine the impact of CBI and ESG scores 11 We collect the information on the use of environmental, social and governance screens for SRI funds from their prospectus. 15

16 on the subsequent period fund performance (hypothesis 2a and 2b). 12 R i,t+1 = α + β 1 ESG i,t + β 2 CBI i,t + β 3 F und characteristics i,t + β 4 P ortfolio characteristics i,t + β 5 T ime dummies + ε i,t, (4) where ESG i,t is a fund portfolio s ESG score; CBI i,t is a fund portfolio s CBI score; F und characteristics i,t include net fund flow, total net asset value, fund family total net asset value, age, expense and turnover ratios for fund i. P ortfolio characteristics i,t are portfolio characteristics control variables which contain the book-to-market ratio, market capitalization, leverage, dividend yield, return on asset, cash flow volatility, return volatility, individual stock turnover and illiquidity. Finally, to examine the relation between the strength and concern of ESG scores on fund performance (hypothesis 3a and 3b), we run the following pooled OLS regression: R i,t+1 = α + β 1 ESG(+) i,t + β 2 ESG( ) i,t + β 3 CBI i,t + β 4 F und characteristics i,t + β 5 P ortfolio characteristics i,t + β 6 T ime dummies + ε i,t, (5) where ESG(+) i,t and ESG( ) i,t are a fund portfolio s strength and concern ESG scores, respectively. To further investigate if there is an asymmetric impact of ESG scores on future fund performance, we consider the quantile estimation of Eq. (4). Quantile regression is introduced by Koenker and Bassett (1978), which can be viewed as an extension of the classical least squares estimation of conditional mean models to the estimation of an ensemble of models for several conditional quantile functions. There are several advantages to using quantile regressions over simple OLS regressions. First, when data are heterogeneous, quantile regressions allow inferences about the influence of regressors conditional on the distribution of the endogenous variable. OLS regression models merely estimate the relation between covariates and the conditional mean of the dependent variable. Quantile regression extends the regression model to conditional 12 We compute standard errors that are clustered by both fund and time in the spirit of Thompson (2011) and Petersen (2009). Thompson (2011) argues that double-clustering is most important when the number of firms and time periods are not too different. The variance estimate for an OLS estimator β is V ( β) = V firm + V time,0 + V white,0, where V firm and V time,0 are the estimated variances that are clustered by firm and time, respectively, and V white,0 is the usual heteroskedasticity-robust OLS variance matrix (White, 1980). 16

17 quantiles of the dependent variable. Because quantile regressions estimate conditional quantile functions, they are appropriate when there is a significant degree of variation in the data. Therefore, quantile regressions can capture information about the slope of the regression line at different quantiles of the endogenous variable (fund performance) given the set of exogenous variables (ESG, CBI, and fund/portfolio characteristics). Second, since no distributional assumption is imposed on the error term, quantile regression estimates exhibit strong model robustness. The conditional quantile regression analysis developed by Koenker and Bassett (1978) and extended by Koenker and Hallock (2001) accounts for the skewed distribution of fund performance, and can be used to draw more appropriate inferences with respect to independent variables across the performance distribution Empirical findings 6.1. Difference in portfolio ethicality between SRI and conventional funds We begin by examining the time series of ESG and CBI scores for SRI and conventional funds. Table 1 shows the number of funds used in our analysis and the time-series of the value-weighted ESG and CBI scores. Note that the number of SRI funds is much less than the number of conventional funds. This is expected because the SRI fund market operates as a niche market within the entire mutual fund market. 14 In our sample, the number of SRI fund is 182 as of 2011, which is comparable to the 184 SRI funds in Nofsinger and Varma (2014). Figure 2 graphically shows how equal-weighted ESG and CBI scores change over time. 15 In most years, the total ESG scores of the SRI funds are slightly higher than those of the conventional funds. However, the total ESG score does not differentiate between the three individual categories of ethicality. Examining the environmental, social, and 13 Note that quantile regression is not equivalent to simply separating the unconditional distribution of the dependent variable into quantiles and then estimating the effects of independent variables using OLS for each subset. This erroneous approach would lead to catastrophic results, particularly when the data has outliers. In contrast, quantile regressions utilize all of the data to fit quantiles. 14 According to the Investment Company Fact Book 2014, the total net asset value of the entire mutual fund market is $11,831.3 billion as of 2011 while the SRI mutual fund market is only $316.1 billion based on the US SIF report We use equal-weighted ESG and CBI scores for Figure 2 since we want to compare SRI and conventional funds with the US market. For all other tables, we use value-weighted ESG and CBI scores. 17

18 governance scores separately shows that SRI funds typically have higher environmental and social scores than conventional funds but exhibit very similar governance scores. This result could be because conventional fund managers recognize that improvements in firm governance leads to better future firm performance, thus, conventional fund portfolios could contain more firms with better governance scores in order to maximize fund performance. Furthermore, we observe that both SRI and conventional funds environmental and social scores are above market average for all years but their governance scores are below market average until Overall, it appears that the higher ESG scores of SRI funds relative to conventional funds stem from their better environmental and social scores. We also observe an increasing trend in ESG scores in Figure 2 which could be a result of better ESG reporting over the years. As mentioned previously, the number of strength and concern indicators reported by MSCI STATS varies each year and generally increases over time. Thus, one might argue that the observed upward trend in ESG scores is an artefact of the better reporting by MSCI STATS over the sample period. However, we have taken two steps to mitigate this potential problem. First, we use the adjusted strength and concern scores as in Deng et al. (2013) so that scores across years can be compared. Second, if the upward trend observed in the ESG scores of both SRI and conventional funds is due to the better reporting over time, then there should also be an upward trend in the ESG score of the market portfolio. However, as shown in Figure 2, both SRI and conventional funds ESG scores deviate significantly from the market average, which indicates that the trend is not due to better reporting. We also observe a substantial difference in the CBI scores between SRI and conventional funds which suggests that the use of exclusionary screens could play pivotal role in reducing the proportion of sin stocks in SRI portfolios. [Insert Table 1 here] [Insert Figure 2 here] One interesting pattern in Figure 2 is that there are sudden jumps in ESG scores around the financial crisis periods. Both the environmental and total ESG scores exhibit 16 We calculate the market ESG and CBI scores in the same way that is used for SRI and conventional funds. The coverage of the market portfolio is around 3000 stocks available in the MSCI STATS database which includes the 3000 largest US companies and companies in the MSCI KLD 400 social index. 18

19 an abnormal increase in 2009 while the social and governance scores increase steadily in For the market portfolio, the direction of the pattern is opposite, except for the environmental score. The market portfolio s social, governance and total ESG scores declines during the crisis periods and then increases afterward. This observation could be due to the majority of firms reducing operational costs and shrinking social benefits to employees during the financial crisis period. The decoupling behavior around the crisis period is in line with the shielding effect of socially responsible stocks against crisis (Kim, Li, & Li, 2014) and the lower crash risk of socially responsible stocks during crisis periods (Nofsinger & Varma, 2014). [Insert Table 2 here] [Insert Table 3 here] To examine the impact of funds level of ethicality on their performance, we sort the funds according to their ESG scores (Table 2) and CBI scores (Table 3) into three groups and compute the corresponding fund returns. Rank 1 in Table 2 represents the most ethical fund group while rank 3 represents the least ethical fund group based on ESG scores. Across all ranks, the ESG scores for SRI funds are greater than those of conventional funds. This finding is consistent with those in Table 1. The difference is also formally tested by a mean difference t-test and a Wilcoxon/Mann-Whitney test for the median (Panel C of Table 2). The results show that the ESG score differences are statistically significant for both the mean and median at the 1% level. Our result is consistent with Kempf and Osthoff (2008), but not with Utz et al. (2014). Utz et al. (2014) find that SRI funds are similar to conventional funds in mean and maximum ESG scores, but have higher minimum scores over time. The different results stem from three sources. First, Utz et al. (2014) cover 27 SRI funds and use a sample period from 2003 to In contrast, we have a much larger sample of SRI funds over a similar time period. Second, the ESG scores of Utz et al. (2014) are computed using an inverse portfolio optimization technique whereas we follow the approach of Deng et al. (2013). Lastly, our study uses MSCI STATS database to compute ESG scores while Utz et al. (2014) uses Thomson Reuters ASSET4. Panels A and B in Table 2 show that portfolio groups with high ESG scores do not necessarily exhibit low CBI scores. This result implies that the use of ESG-related screens 19

20 does not necessarily exclude sin stocks from portfolios. For example, if a tobacco company (which is considered a sin stock) performs well in ESG-related areas, it could be included in an SRI fund s portfolio that does not use an exclusionary screen for the tobacco industry. For the relation between ESG and fund returns, it is not apparent that higher ESG scores lead to higher current or subsequent period returns, since neither R t nor R t+1 increases with ESG scores for either fund type. The book-to-market ratio and market capitalization do not show any obvious patterns in relation to the ESG score. We now examine the relation between CBI scores and fund performance by sorting both types of funds into high-, medium-, and low-cbi portfolios. Table 3 shows the relation between CBI scores and the ESG scores, fund returns, the book-to-market ratio and market capitalization of funds. Rank 1 (3) contains the portfolio with the highest (lowest) CBI scores. Analogous to the ESG score, the mean and median differences of the CBI scores between SRI and conventional funds are statistically significant at the 1% level, which suggests that SRI funds are more likely to exclude sin stocks from their portfolios (Panel C of Table 3). In line with the findings of Table 2, CBI scores are not strongly correlated with ESG scores reinforcing the idea that these two scores capture different dimensions of fund ethicality. We observe that for both types of funds, CBI scores are positively associated with both current and subsequent returns. This result indicates that the proportion of sin stocks in portfolios has a considerable impact on fund returns and is in line with the findings of Hong and Kacperczyk (2009) who show that sin stocks have higher expected returns. We also see that funds with more sin stocks (i.e., greater CBI scores) tend to have stocks of greater value and market capitalization The impact of screens on the ethicality of portfolios Next, we examine the relation between ESG scores and screens employed by SRI funds. Although the screening process is the main tool that SRI funds use to differentiate themselves from their conventional counterparts, there has been little attempt to investigate whether these screens actually increase the ethicality of SRI fund portfolios. In principle, when more intense exclusionary screens are employed (measured by the number of screens), portfolio exposure to controversial businesses is expected to be reduced (i.e., a reduction in the CBI score). If ESG-related screens are implemented, then the ESG score of fund portfolios is expected to increase. 20

21 [Insert Table 4 here] The first column of Table 4 demonstrates that the intensity of exclusionary screens has a negative impact on SRI funds CBI scores significant at the 5% level. For the individual ESG scores, all ESG-related screening processes have significant positive coefficients. For example, in the second column of Table 4, environmental screens help improve the environmental score of the SRI fund by This result indicates that environmental, social, and governance screens help improve the scores of the individual categories of ethicality. When considering all ESG-related screens together in the regression specification in column five, the significance of the screens disappears. This result could imply that individual screens do not have any explanatory power on the total ESG score. 17 Finally, in the last column of Table 4, we see that when both exclusionary and ESG-related screens are applied, the coefficient on exclusionary screens becomes insignificant, highlighting the fact that the exclusionary screening process affects CBI scores and not ESG scores Impact of portfolio ethicality on fund performance In this section, we test whether a fund s level of controversial business involvement and ethicality are related to its subsequent performance (hypotheses 2a and 2b). Table 5 shows the impact of ESG and CBI scores on subsequent fund returns. Columns (2) and (5) of Table 5 show that funds with higher CBI scores have higher subsequent fund returns for both SRI and conventional funds, which is consistent with hypothesis 2a. A test of significance between the coefficients on the CBI variable for SRI and conventional funds indicates that the magnitude of the CBI variable is significantly larger for SRI funds. This result shows that since conventional funds do not apply exclusionary screens, their fund portfolios would contain more sin stocks than SRI fund portfolios. Consequently, an increase in the CBI score of conventional fund portfolios would lead to a smaller increase in expected fund returns due to diminishing marginal returns. However, since SRI funds apply exclusionary screens, which excludes sin stocks from their portfolios, the inclusion of sin stocks in SRI funds portfolios would induce much larger financial gains. [Insert Table 5 here] 17 The lack of statistical significance of individual screens in explaining the total ESG score could also be because they are highly correlated. Specifically, their correlation varies approximately from 0.55 to 0.65 with the total ESG score. 21

22 The ESG score also has an expected positive sign for both fund types as shown in columns (1) and (4) of Table 5. For SRI funds, a 10% increase in ESG score results in a 1.72% growth in subsequent fund return, all other things constant. In columns (3) and (6) of Table 5, we see that both CBI and ESG scores have a positive relation with subsequent fund returns with the impact of CBI scores being stronger than that of ESG scores. This result implies that for SRI funds there is a trade-off in financial benefits depending on the screens applied. Specifically, since exclusionary screens are shown to decrease CBI scores, the underperformance of SRI funds documented in the literature could be due to the extensive use of exclusionary screens. On the other hand, since conventional funds are not subject to any screening processes, they are able to maximize the expected returns of their portfolio by choosing stocks with the highest CBI and ESG scores. [Insert Table 6 here] To ensure robustness of the results in Table 5, we also use risk-adjusted returns as the dependent variable including excess fund returns over the risk-free rate and over the market return, and the Sharpe ratio. The estimation results are presented in Table The main findings in Table 6 are similar to those in Table 5, that is, both ESG and CBI scores have a positive impact on subsequent risk-adjusted returns for SRI and conventional funds. The significantly positive coefficient on the ESG score across all columns indicates that an increase in ESG score sufficiently compensates the decrease in the potential diversification risk as argued by Rudd (1979). [Insert Table 7 here] Next, we decompose the ESG scores into its subcategories and investigate the impact of each subcategory on the fund returns of the subsequent period. Table 7 shows the relation between individual ESG scores and subsequent fund returns. For SRI funds, the environmental and social scores have a significantly positive impact on subsequent fund returns, while the governance score has a insignificant coefficient. This result could be due to the fact that the governance score is not directly related to the governance screens 18 We do not use Jensen s alpha or other regression based risk-adjusted performance measures since the dependent variable is the one-step ahead return which restricts us to using only one-year return series for the estimation. 22

23 employed by SRI funds. In fact, Servaes and Tamayo (2013) point out that the governance score does not fully reflect firms CSR practices, given that corporate governance is about the mechanisms that allow principals (shareholders) to reward and exert control on agents (managers). Inspecting the individual items which form the governance score shows that there are two strength indicators (reporting quality and public policy) and four concern indicators (reporting quality, public policy, governance structure controversies, and other controversies). Most of these variables do not have sound definitions and are less clear in practice. For example, the governance structure controversy indicator is defined as measuring the severity of controversies related to a firm s executive compensation and governance practices. This may lead to more subjective decisions made on part of the SRI fund when implementing governance screens. When comparing the magnitudes of the coefficients between the environmental and social scores, the social measure is almost twice as large as the environmental measure for both SRI and conventional funds. Since an improvement in environmental aspects usually requires some type of investment as well as constant monitoring effort, activities that improve the environmental score would be more costly than those activities that help improve the social score. Consequently, the impact of environmental factors on subsequent returns is lower than that of social factors. For conventional funds, all three subcategories appear to be significantly positive. The positive impact of CBI scores on subsequent returns remains largely the same across all columns. [Insert Table 8 here] Table 8 presents the estimation results of the regression of subsequent fund returns on the strength and concern of ESG scores for both SRI and conventional funds. Columns (1) to (4) and (5) to (8) show the results for SRI and conventional funds, respectively. As shown in columns (4) and (8), the total ESG strength score is significantly positive while the total ESG concern score is insignificant. This result indicates that the good news contained in the ESG strength score has value-relevant information which contributes towards subsequent fund performance. For SRI funds, the environmental and social strength scores are all positive and significant but their respective concern scores are insignificant. Both the strength and concern scores for governance are insignificant. These results imply that for SRI funds, the value-relevant information contained in the total ESG 23

24 strength score originates from the environmental and social categories. For conventional funds, all of the individual ESG strength scores are positive and significant indicating that conventional funds tend to capitalize on good CSR practices across all three categories of ethicality to improve fund performance Asymmetric impact of portfolio ethicality In this section, we examine whether portfolio ethicality has an asymmetric effect on subsequent fund returns, that is, whether fund ethicality has a different impact on the subsequent fund returns of high-performance portfolios compared to low-performance portfolios. We run a quantile regression from the 0.05 quantile to the 0.95 quantile of the subsequent period fund returns, with quantile increments of Table 9 reports the quantile regression estimation results at the 0.1, 0.3, 0.5 (median), 0.7, and 0.9 quantiles. The median regression shows that the median coefficient estimates of both the CBI and ESG scores are similar to the OLS estimates from Table 5. Figure 3 plots the point and interval coefficient estimates of the ESG and CBI scores at all quantiles estimated. We see that while CBI scores appear to have an asymmetric impact on subsequent period fund returns for SRI and conventional funds, ESG scores do not. Specifically, for both types of funds, the positive relation between CBI scores and subsequent fund returns is much stronger (weaker) for high-performance (low-performance) funds which implies that the financial performance of SRI and conventional funds largely depends on the inclusion of sin stocks in their portfolios. Although the ESG score appears to have an inverse U-shape relation with subsequent fund returns across quantiles for SRI funds, the magnitude of the variation is minimal (coefficients range from 0.04 to 0.18) which means there is unlikely to be any asymmetric effects. For conventional funds, the effect of the ESG score is fairly flat, which implies that conventional funds are unlikely to utilize ESG criteria in their investment decision making. [Insert Table 9 here] [Insert Figure 3 here] 24

25 7. Conclusions Socially responsible funds have become increasingly popular over the last two decades and their assets under management are expected to grow further in the future. While several studies have attempted to determine the financial performance differences between SRI and conventional funds, differences in portfolio ethicality have received little attention, despite the fact that SRI funds tend to charge higher fees than conventional funds primarily due to the additional ethical research. We find that the exclusionary screens employed by SRI funds help decrease investments in irresponsible companies and that the ESG-related screens increase the corresponding individual ESG scores. We empirically examine the ethicality of SRI and conventional funds and find that SRI funds have higher ESG scores. Moreover, it appears that both types of funds have tilted their portfolios toward more ethical stocks, especially around the recent financial crisis periods. As Nofsinger and Varma (2014) argue, if responsible portfolios are less risky in crisis periods, it is sensible to move from irresponsible to responsible portfolios during periods of turmoil, which leads to an increase in fund ESG scores. We also empirically study the financial performance of SRI and conventional funds by examining the impacts of exclusionary and ESG-related screens simultaneously. Using a sample of SRI and conventional funds from 2003 to 2012, we find that an increase in ESG scores leads to higher subsequent returns, while a decrease in CBI scores leads to inferior financial performance in the following period for both SRI and conventional funds. These findings are consistent with the errors-in-expectation and taste-based hypotheses. However, it appears that the impact of CBI scores on fund performance dominates that of the ESG scores when the relative effects are compared. This result suggests that SRI funds that only impose exclusionary screens are more likely to underperform conventional funds, but funds with only ESG screens might be able to outperform conventional funds. 25

26 References Amihud, Y. (2002). Illiquidity and stock returns: Cross-section and time series effects. Journal of Financial Markets, 5 (1), Barnett, M. L., & Salomon, R. M. (2006). Beyond dichotomy: The curvilinear relationship between social responsibility and financial performance. Strategic Management Journal, 27 (11), Bauer, R., Koedijk, K., & Otten, R. (2005). International evidence on ethical mutual fund performance and investment style. Journal of Banking and Finance, 29, Bello, Z. Y. (2005). Socially responsible investing and portfolio diversification. Journal of Financial Research, 28 (1), Cornell, B., & Shapiro, A. C. (1987). Corporate stakeholders and corporate finance. Financial Management, 16 (1), Deng, X., Kang, J. K., & Low, B. S. (2013). Corporate social responsibility and stakeholder value maximization: Evidence from mergers. Journal of Financial Economics, 110 (1), Derwall, J., Guenster, N., Bauer, R., & Koedijk, K. (2005). The eco-efficiency premium puzzle. Financial Analysts Journal, 61 (2), Derwall, J., Guenster, N., Bauer, R., & Koedijk, K. (2011). A tale of values-driven and profit-seeking social investors. Journal of Banking and Finance, 35 (8), Edmans, A. D. (2011). Does the stock market fully value intangibles? Employee satisfaction and equity prices. Journal of Financial Economics, 101 (3), Fama, E. F., & French, K. R. (1993). Common risk factors in the returns on stocks and bonds. Journal of Financial Economics, 33 (1), Fama, E. F., & French, K. R. (2007). Disagreement, tastes, and asset prices. Journal of Financial Economics, 83 (3), Fombrun, C. J., Gardberg, N. A., & Barnett, M. L. (2000). Opportunity platforms and safety nets: Corporate citizenship and reputational risk. Business and Society Review, 105 (1), Ghoul, S. E., Guedhami, O., Kwok, C. C. Y., & Mishra, D. R. (2011). Does corporate social responsibility affect the cost of capital? Journal of Banking and Finance, 35 (9), Girard, E., Rahman, H., & Stone, B. A. (2007). Socially responsible investments: Goody-two-shoes or bad to the bone? Journal of Investing, 16 (1), Goldreyer, E. F., & Diltz, J. D. (1999). The performance of socially responsible mutual funds: Incorporating sociopolitical information in portfolio selection. Managerial Finance, 25 (1), Hamilton, S., Jo, H., & Statman, M. (1993). Doing well while doing good? The investment performance of socially responsible mutual funds. Financial Analysts Journal, 49 (6), 26

27 Heinkel, R., Kraus, A., & Zechner, J. (2001). The effect of green investment on corporate behavior. Journal of Financial and Quantitative Analysis, 35 (4), Hong, H., & Kacperczyk, M. (2009). The price of sin: The effects of social norms on markets. Journal of Financial Economics, 93 (1), Humphrey, J. E., & Lee, D. D. (2011). Australian socially responsible funds: Performance, risk and screening intensity. Journal of Business Ethics, 102 (4), In, F., Kim, M., Park, R. J., Kim, S., & Kim, T. S. (2014). Competition of socially responsible and conventional mutual funds and its impact on fund performance. Journal of Banking and Finance, 44, Kempf, A., & Osthoff, P. (2007). The effect of socially responsible investing on financial performance. European Financial Management, 13 (5), Kempf, A., & Osthoff, P. (2008). Sri funds: Nomen est omen. Journal of Business Finance and Accounting, 35, Kim, Y., Li, H., & Li, S. (2014). Corporate social responsibility and stock price crash risk. Journal of Banking and Finance, 43, Koenker, R., & Bassett, G. (1978). Regression quantiles. Econometrica, 46 (1), Koenker, R., & Hallock, K. F. (2001). Quantile regression. Journal of Economic Perspective, 15 (4), Krüger, P. (2015). Corporate goodness and shareholder wealth. Journal of Financial Economics, 115 (2), Lee, D. D., Humphrey, J. E., Benson, K. L., & Ahn, J. Y. K. (2010). Socially responsible investment fund performance: The impact of screening intensity. Accounting and Finance, 50 (2), Markowitz, H. (1959). Portfolio selection: Efficient diversification of investments. New York: Wiley. Nofsinger, J., & Varma, A. (2014). Socially responsible funds and market crises. Journal of Banking and Finance, 48, Petersen, M. (2009). Estimating standard errors in finance panel data sets: Comparing approaches. Review of Financial Studies, 22 (1), Renneboog, L., Horst, J. T., & Zhang, C. (2008). The price of ethics and stakeholder governance: The performance of socially responsible mutual funds. Journal of Corporate Finance, 14 (3), Renneboog, L., Horst, J. T., & Zhang, C. (2011). Is ethical money financially smart? Nonfinancial attributes and money flows of socially responsible investment funds. Journal of Financial Intermediation, 20 (4), Rudd, A. (1979). Divestment of south african equities: How risky? Journal of Portfolio Management, 5 (3), Servaes, H., & Tamayo, A. (2013). The impact of corporate social responsibility on firm 27

28 value: The role of customer awareness. Management Science, 59 (5), Social Investment Forum. (2014). Report on US sustainable, responsible and impact investing trends. US Social Investment Forum. Statman, M. (2000). Socially responsible mutual funds. Financial Analysts Journal, 56 (3), Statman, M., & Glushkov, D. (2009). The wages of social responsibility. Financial Analysts Journal, 65 (4), Thompson, S. B. (2011). Simple formulas for standard errors that cluster by both firm and time. Journal of Financial Economics, 99 (1), Utz, S., Wimmer, M., Hirschberger, M., & Steuer, R. E. (2014). Tri-criterion inverse portfolio optimization with application to socially responsible mutual funds. European Journal of Operational Research, 234 (2), Walker, K., & Wan, F. (2012). The harm of symbolic actions and green-washing: Corporate actions and communications on environmental performance and their financial implications. Journal of Business Ethics, 109 (2), White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for heteroskedasticity. Econometrica, 48 (4), Zhang, X. F. (2006). Information uncertainty and stock returns. Journal of Finance, 61 (1),

29 Tables Table 1 Time-series of ESG and CBI scores. Year # of funds E S G ESG CBI Panel A: SRI funds Panel B: Conventional funds This table reports the number of SRI and conventional funds used in our analysis and the time series of value-weighted ESG and CBI scores. Panels A and B report the ESG and CBI scores for SRI and conventional funds, respectively. The variables E, S, and G refer to the individual environmental, social, and governance scores, respectively. We exclude the years 2001 and 2002 since the holding information of SRI funds is largely incomplete. 29

30 Table 2 The relation between ESG scores and fund performance. Rank ESG CBI R t R t+1 BM Cap Panel A: SRI funds Total Panel B: Conventional funds Total Panel C: Mean and median tests ESG difference Mean Median SRI Total -Conventional Total SRI 1 -Conventional SRI 2 -Conventional SRI 3 -Conventional This table reports the CBI score, current and subsequent period fund returns, BM and Cap of SRI (Panel A) and conventional (Panel B) funds sorted according to the ESG score. Rank 1 (3) contains the portfolio with the highest (lowest) ESG scores. ESG and CBI are the ESG and CBI scores of the portfolio, respectively. The variables R t and R t+1 are the current and subsequent period fund returns. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) at the end of the year. Panel C reports the mean and median difference test for the ESG score between SRI and conventional funds. The t-statistic is provided under the column titled Mean and the Wilcoxon/Mann-Whitney test statistic is provided under the column titled Median. denotes significance at the 1% level. 30

31 Table 3 The relation between CBI scores and fund performance. Rank CBI ESG R t R t+1 BM Cap Panel A: SRI funds Total Panel B: Conventional funds Total Panel C: Mean and median tests CBI difference Mean Median SRI Total -Conventional Total SRI 1 -Conventional SRI 2 -Conventional SRI 3 -Conventional This table reports the ESG score, current and subsequent period fund returns, BM and Cap of SRI (Panel A) and conventional (Panel B) funds sorted according to the CBI score. Rank 1 (3) contains the portfolio with the highest (lowest) CBI scores. ESG and CBI are the ESG and CBI scores of the portfolio, respectively. The variables R t and R t+1 are the current and subsequent period fund returns. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) at the end of the year. Panel C reports the mean and median difference test for the CBI score between SRI and conventional funds. The t-statistic is provided under the column titled Mean and the Wilcoxon/Mann-Whitney test statistic is provided under the column titled Median. denotes significance at the 1% level. 31

32 Table 4 Screening intensity and fund ethicality. CBI E S G ESG ESG Intercept (-2.19) (-1.19) (-1.62) (-7.08) (-4.36) (-4.67) Exclusionary (-1.98) (-1.39) ES (2.70) (0.71) (0.09) SS (2.52) (0.39) (0.82) GS (1.93) (0.73) (1.12) T NA (-4.14) (-2.94) (-3.22) (-3.52) (-3.56) (-3.05) F amily T NA (7.52) (2.68) (2.52) (3.27) (3.31) (2.94) Age (7.51) (6.70) (6.88) (6.77) (7.26) (7.42) Expense (5.64) (14.77) (13.99) (14.00) (13.91) (13.98) F.T urnover (-0.12) (-4.56) (-4.50) (-4.40) (-4.51) (-4.60) Adj. R This table presents the estimation results of the regression of portfolio ethicality on the screens used by SRI funds. The first row denotes the dependent variable of each regression. ESG and CBI are the ESG and CBI scores of the fund portfolio, respectively. The variables E, S, and G refer to the individual environmental, social, and governance scores, respectively. Exclusionary is the number of exclusionary screens used and ES, SS and GS are dummy variables that have a value of one if an SRI fund employs environmental, social and governance screens, respectively, and zero otherwise. T N A is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratio of the fund. For all specifications, year fixed effects are included; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 32

33 Table 5 The impact of CBI and ESG scores on subsequent fund returns. SRI funds Conventional funds (1) (2) (3) (4) (5) (6) Intercept (6.39) (6.76) (6.98) (0.61) (0.37) (0.84) ESG (3.23) (2.50) (5.17) (4.11) CBI (3.61) (3.04) (5.49) (3.85) F low (-2.46) (-2.56) (-2.46) (1.39) (1.51) (1.31) T NA (-0.68) (-0.72) (-0.93) (-0.09) (-0.11) (-0.04) F amily T NA (-0.90) (-1.61) (-1.47) (-0.62) (-1.09) (-0.85) Age (0.12) (-0.12) (0.04) (1.41) (1.37) (1.36) Expense (3.25) (3.78) (3.26) (0.77) (0.81) (0.81) F.T urnover (0.12) (-0.36) (-0.45) (2.27) (2.62) (2.42) BM (2.25) (2.22) (2.08) (-1.02) (-1.17) (-1.03) Cap (1.84) (1.72) (1.67) (3.66) (4.28) (3.26) Leverage (-0.47) (-0.98) (-0.54) (-3.47) (-3.31) (-3.21) DivY ield (-2.50) (-2.04) (-2.20) (-0.86) (-1.62) (-1.40) ROA (-2.11) (-2.12) (-2.28) (-4.97) (-5.44) (-4.90) CF V olt (-0.22) (-1.63) (-1.49) (-6.99) (-5.91) (-6.87) RetV olt (-1.34) (-2.00) (-1.30) (7.97) (6.21) (8.26) T urnover (-0.03) (0.35) (0.07) (-3.63) (-3.23) (-3.48) Illiquidity (1.37) (2.03) (1.33) (-7.96) (-6.14) (-8.24) Adj. R This table presents the estimation results of the regression of subsequent fund returns on the ESG and CBI scores of both SRI and conventional funds. ESG and CBI are the ESG and CBI scores of the fund portfolio, respectively. F low is the net fund flow. T NA is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratios of a fund. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) of the stock at the end of the year. Leverage refers to the debt-to-equity ratio. DivY ield refers to the annual percentage dividend yield calculated for each stock as the annual dividend divided by price. ROA is the net income divided by average total assets. CF V olt is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum of three years), scaled by the average total assets. RetV olt is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. T urnover is calculated as the trading volume divided by shares outstanding and Illiquidity is Amihud s (2002) illiquidity ratio. For all specifications, year fixed effects are included and standard errors are clustered by fund and time; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 33

34 Table 6 The impact of CBI and ESG scores on subsequent risk-adjusted fund returns. SRI funds Conventional funds (1) (2) (3) (4) (5) (6) Intercept (6.79) (2.48) (12.10) (0.70) (-2.44) (1.98) ESG (2.50) (2.47) (4.53) (4.12) (4.08) (1.88) CBI (3.04) (3.05) (4.75) (3.85) (3.86) (3.09) F low (-2.47) (-2.43) (-1.46) (1.31) (1.31) (1.81) T NA (-0.92) (-0.99) (-0.35) (-0.05) (-0.03) (0.56) F amily T NA (-1.47) (-1.47) (-3.77) (-0.85) (-0.82) (-0.76) Age (0.03) (0.12) (-0.89) (1.36) (1.36) (0.37) Expense (3.26) (3.23) (-1.84) (0.81) (0.81) (0.89) F.T urnover (-0.44) (-0.52) (-0.57) (2.42) (2.41) (2.04) BM (2.08) (2.12) (0.70) (-1.03) (-0.97) (-1.05) Cap (1.67) (1.68) (1.46) (3.26) (3.19) (2.94) Leverage (-0.54) (-0.53) (0.61) (-3.22) (-3.12) (-2.53) DivY ield (-2.20) (-2.21) (-1.28) (-1.40) (-1.44) (0.93) ROA (-2.28) (-2.27) (-1.12) (-4.90) (-4.91) (-3.62) CF V olt (-1.49) (-1.51) (-0.79) (-6.88) (-6.81) (-8.76) RetV olt (-1.29) (-1.37) (-1.94) (8.28) (8.04) (4.33) T urnover (0.07) (0.08) (-1.11) (-3.49) (-3.33) (-1.26) Illiquidity (1.32) (1.40) (1.98) (-8.25) (-8.02) (-4.30) Adj. R This table presents the estimation results of the regression of subsequent risk-adjusted fund returns on the ESG and CBI scores of both SRI and conventional funds. The dependent variable is: (i) the excess fund returns over the risk-free rate in columns (1) and (4); (ii) the excess fund returns over the market return in columns (2) and (5); and (iii) the Sharpe ratio in columns (3) and (6). ESG and CBI are the ESG and CBI scores of the fund portfolio, respectively. F low is the net fund flow. T NA is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratios of a fund. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) of the stock at the end of the year. Leverage refers to the debt-to-equity ratio. DivY ield refers to the annual percentage dividend yield calculated for each stock as the annual dividend divided by price. ROA is the net income divided by average total assets. CF V olt is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum of three years), scaled by the average total assets. RetV olt is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. T urnover is calculated as the trading volume divided by shares outstanding and Illiquidity is Amihud s (2002) illiquidity ratio. For all specifications, year fixed effects are included and standard errors are clustered by fund and time; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 34

35 Table 7 The impact of individual ESG scores on subsequent fund returns. SRI funds Conventional funds (1) (2) (3) (4) (5) (6) Intercept (6.99) (6.92) (6.64) (0.72) (0.48) (0.87) Environmental (2.75) (4.50) Social (4.28) (5.43) Governance (0.01) (2.32) CBI (3.14) (3.46) (3.46) (4.40) (5.60) (2.88) F low (-2.40) (-2.44) (-2.56) (1.24) (1.41) (1.40) T NA (-1.05) (-1.03) (-0.72) (-0.03) (-0.06) (-0.07) F amily T NA (-1.50) (-1.59) (-1.59) (-0.89) (-0.86) (-0.90) Age (0.14) (0.11) (-0.12) (1.34) (1.39) (1.36) Expense (2.86) (3.16) (3.82) (0.81) (0.80) (0.80) F.T urnover (-0.42) (-0.29) (-0.36) (2.44) (2.51) (2.44) BM (2.18) (2.24) (2.26) (-1.10) (-0.88) (-1.10) Cap (1.25) (0.97) (1.66) (2.85) (2.33) (4.56) Leverage (-0.53) (-0.62) (-0.99) (-3.28) (-3.27) (-3.17) DivY ield (-2.13) (-2.33) (-1.97) (-1.32) (-1.36) (-1.55) ROA (-2.16) (-2.31) (-2.09) (-5.08) (-4.81) (-5.04) CF V olt (-1.66) (-2.02) (-1.58) (-6.66) (-6.60) (-6.60) RetV olt (-1.21) (-1.07) (-1.96) (8.49) (7.18) (7.59) T urnover (0.05) (0.21) (0.35) (-3.69) (-3.63) (-3.24) Illiquidity (1.25) (1.10) (1.99) (-8.46) (-7.10) (-7.56) Adj. R This table presents the estimation results of the regression of subsequent fund returns on the individual ESG scores of both SRI and conventional funds. The variables Environmental, Social, and Governance refer to the individual environmental, social, and governance scores of the fund portfolio, respectively. CBI is the CBI score of the fund portfolio, respectively. F low is the net fund flow. T NA is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratios of a fund. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) of the stock at the end of the year. Leverage refers to the debt-to-equity ratio. DivY ield refers to the annual percentage dividend yield calculated for each stock as the annual dividend divided by price. ROA is the net income divided by average total assets. CF V olt is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum of three years), scaled by the average total assets. RetV olt is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. T urnover is calculated as the trading volume divided by shares outstanding and Illiquidity is Amihud s (2002) illiquidity ratio. For all specifications, year fixed effects are included and standard errors are clustered by fund and time; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 35

36 Table 8 The impact of ESG strength and concern scores on subsequent fund returns. SRI funds Conventional funds (1) (2) (3) (4) (5) (6) (7) (8) Intercept (7.16) (7.01) (6.63) (7.14) (1.71) (1.04) (1.18) (2.05) E(+) (2.70) (5.11) E( ) (0.63) (1.16) S(+) (4.08) (5.28) S( ) (-1.25) (-0.91) G(+) (-0.05) (6.69) G( ) (-0.09) (1.11) ESG(+) (2.82) (8.80) ESG( ) (1.01) (1.02) CBI (1.69) (2.88) (3.34) (2.19) (2.10) (5.89) (2.73) (3.22) F low (-2.41) (-2.35) (-2.49) (-2.27) (1.37) (1.45) (1.30) (1.40) T NA (-0.94) (-1.02) (-0.73) (-0.85) (0.00) (-0.02) (0.03) (0.04) F amily T NA (-1.57) (-1.65) (-1.51) (-1.72) (-0.90) (-0.87) (-0.84) (-0.86) Age (0.08) (0.12) (-0.12) (0.05) (1.30) (1.35) (1.29) (1.27) Expense (2.14) (3.15) (3.76) (2.57) (0.83) (0.82) (0.84) (0.85) F.T urnover (-0.48) (-0.29) (-0.37) (-0.43) (2.75) (2.62) (3.03) (2.91) BM (2.22) (2.24) (2.24) (2.11) (-1.47) (-0.91) (-1.45) (-1.33) Cap (0.89) (0.68) (1.39) (0.40) (1.46) (1.34) (1.27) (0.43) Leverage (-0.32) (-0.63) (-1.00) (-0.51) (-2.58) (-2.93) (-2.03) (-2.14) DivY ield (-2.20) (-2.31) (-1.97) (-2.25) (-1.81) (-2.09) (-1.74) (-2.41) ROA (-2.04) (-2.21) (-2.05) (-2.00) (-4.51) (-4.93) (-4.95) (-4.85) CF V olt (-2.52) (-2.22) (-1.20) (-2.72) (-7.89) (-4.08) (-6.47) (-5.95) RetV olt (-0.93) (-1.03) (-1.92) (-1.00) (8.82) (6.98) (7.60) (8.26) T urnover (0.14) (0.27) (0.35) (0.34) (-3.29) (-2.68) (-2.03) (-2.28) Illiquidity (0.96) (1.06) (1.95) (1.03) (-8.80) (-6.90) (-7.54) (-8.20) Adj. R This table presents the estimation results of the regression of subsequent fund returns on the strength and concern of ESG scores for both SRI and conventional funds. The variables E(+), S(+), G(+), and ESG(+) refer to the strength of the individual environmental, social, governance, and total ESG scores, respectively, while E( ), S( ), G( ), and ESG( ) refer to the concern of the individual environmental, social, governance, and total ESG scores, respectively. CBI is the CBI score of the fund portfolio, respectively. F low is the net fund flow. T NA is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratios of a fund. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) of the stock at the end of the year. Leverage refers to the debt-to-equity ratio. DivY ield refers to the annual percentage dividend yield calculated for each stock as the annual dividend divided by price. ROA is the net income divided by average total assets. CF V olt is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum of three years), scaled by the average total assets. RetV olt is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. T urnover is calculated as the trading volume divided by shares outstanding and Illiquidity is Amihud s (2002) illiquidity ratio. For all specifications, year fixed effects are included and standard errors are clustered by fund and time; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 36

37 Table 9 Quantile regression of subsequent fund returns on ESG and CBI scores. SRI funds Conventional funds Quantile Intercept (24.06) (16.09) (15.91) (19.57) (28.28) (0.75) (8.38) (16.20) (15.51) (17.29) ESG (12.14) (6.82) (6.95) (2.00) (2.34) (11.99) (20.09) (21.38) (19.72) (19.99) CBI (5.47) (8.78) (7.28) (11.73) (12.26) (1.76) (10.33) (12.51) (11.33) (8.43) F low (-18.49) (-3.06) (-3.10) (-1.00) (-4.26) (1.41) (0.42) (2.26) (1.09) (2.00) T NA (-15.22) (-6.66) (-1.95) (-0.32) (-1.24) (1.07) (-1.99) (-5.02) (-3.93) (-3.76) F amily T NA (15.66) (-5.18) (-6.22) (-6.69) (-8.72) (7.13) (7.83) (8.31) (4.42) (-1.97) Age (12.43) (0.43) (-0.28) (-1.18) (-5.48) (-1.50) (1.90) (4.65) (3.85) (4.05) Expense (-26.45) (0.62) (2.32) (3.97) (4.23) (-5.29) (-4.97) (-4.65) (-2.11) (2.10) F.T urnover (7.98) (-2.20) (-0.29) (-3.35) (-0.92) (1.92) (1.56) (2.27) (3.43) (-0.29) BM (42.33) (0.13) (3.59) (2.49) (0.71) (1.69) (0.10) (0.95) (0.01) (-3.66) Cap (66.96) (4.96) (0.13) (-2.34) (-2.94) (-1.45) (-4.62) (-7.76) (-6.70) (1.95) Leverage (10.56) (-2.01) (1.69) (-2.77) (6.02) (-5.44) (-5.84) (-2.95) (-3.73) (0.05) DivY ield (-36.08) (-1.20) (-5.08) (-2.01) (-4.51) (-4.04) (-7.44) (-12.24) (-10.62) (-11.00) ROA (49.12) (3.27) (-0.41) (-4.97) (-8.97) (0.58) (-5.05) (-6.28) (-6.84) (-14.96) CF V olt (14.84) (-1.99) (-2.12) (-6.03) (-4.32) (-3.12) (-9.21) (-10.87) (-10.42) (-11.20) RetV olt ( ) (-4.05) (-1.43) (-2.59) (2.98) (8.04) (9.42) (7.56) (8.22) (14.17) T urnover (-17.01) (-5.92) (-2.42) (4.71) (8.11) (-9.24) (-5.13) (-4.03) (-0.14) (2.20) Illiquidity (118.77) (4.14) (1.48) (2.62) (-2.94) (-7.96) (-9.38) (-7.54) (-8.23) (-14.21) This table presents the quantile estimation results of the subsequent fund returns on the ESG and CBI scores of both SRI and conventional funds. ESG and CBI are the ESG and CBI scores of the fund portfolio, respectively. F low is the net fund flow. T NA is the logarithm of the total net asset value of a fund and F amily T NA is the logarithm of the total net asset value of a fund s family. Age is the logarithm of the age of a fund since its inception. Expense and F.T urnover are the expense and turnover ratios of a fund. BM is the book value of equity divided by the market capitalization. Cap is calculated as the market capitalization (in billions) of the stock at the end of the year. Leverage refers to the debt-to-equity ratio. DivY ield refers to the annual percentage dividend yield calculated for each stock as the annual dividend divided by price. ROA is the net income divided by average total assets. CF V olt is the standard deviation of the net cash flow from operating activities over the previous five financial years (minimum of three years), scaled by the average total assets. RetV olt is the standard deviation of daily excess returns (excess over the CRSP value-weighted index) over the last calendar year. T urnover is calculated as the trading volume divided by shares outstanding and Illiquidity is Amihud s (2002) illiquidity ratio. For all specifications, year fixed effects are included and standard errors are clustered by fund and time; t-statistics are shown in the parenthesis.,, and indicate significance at the 10%, 5%, and 1% level, respectively. 37

38 Figures Fig. 1. The mechanism of screening intensity. This figure describes the mechanism of the screening process. Existing studies have focused on the impact of the screening process on fund performance. We argue that the screens used by SRI funds are designed to increase the ethicality of SRI funds. In addition, the screening process may not have a direct impact on fund performance since there is no guarantee that increases in screening intensity leads to portfolios with higher ethicality. Our study focuses on the relation between ESG and CBI scores, which contains value-relevant information, and fund performance. 38