Do Social Responsibility Screens Matter When Assessing Mutual Fund Performance?

Abstract

Regarding the contribution of socially responsible (SR) screening to mutual fund performance, we propose a new decomposition of the variability of SR mutual fund returns that isolates the contribution of SR screening, allowing it to be compared with other, traditional sources of performance. Our results, based on a sample of SR equity mutual funds, show that SR screening does contribute to the variability of mutual fund performance, together with asset allocation decisions and active management. This contribution is, on average, between 4% and 10%, roughly two times lower than the contribution made by active portfolio choices.

Disclosure:

The authors report no conflicts of interest.

Editor’s Note

This article was externally reviewed using our double-blind peer-review process. When the article was accepted for publication, the authors thanked the reviewers in their acknowledgments. Denys Glushkov and Jenke ter Horst were the reviewers for this article.

Submitted 10 April 2015

Accepted 28 December 2016 by Stephen J. Brown

We thank the editorial team of the Financial Analysts Journal and Denys Glushkov and Jenke ter Horst for helpful comments and suggestions.

Notes

¹ A few studies have pointed to a significant underperformance of SR equity investments (Jones, van der Laan, Frost, and Loftus 2008; Renneboog, Ter Horst, and Zhang 2008). In contrast, a few others have documented the outperformance of portfolios constructed on the basis of ethical screening (Derwall, Guenster, Bauer, and Koedijk 2005; Kempf and Osthoff 2007; Derwall, Koedijk, and Ter Horst 2011).

² Although sectorial screens (e.g., those that exclude so-called sin stocks) have a negative effect on risk-adjusted returns, that is not the case with transversal screens (e.g., a commitment to the United Nations Global Compact principles). Funds that use the best-in-class approach are often indistinguishable from their conventional counterparts.

³ We considered other sources of classification, such as Morningstar. In line with Climent and Soriano (2011), we finally opted for Bloomberg (2013), because its definitions of funds in terms of both asset class focus and SR investments seemed more restrictive. Morningstar defines domestic stock funds as “funds with at least 70% of assets in domestic stocks” and international stock funds as having “40% or more of their equity holdings in foreign stocks” (see the Morningstar Investing Glossary at http://www.morningstar.com/InvGlossary/). Bloomberg’s definition is narrower, requiring equity funds’ assets to be at least 80% invested in equities. Moreover, Morningstar defines socially responsible funds as a group that includes “any fund that invests according to noneconomic guidelines. Funds may make investments based on such issues as environmental responsibility, human rights, or religious views.” Thus, thematic funds in water or green energies or even Islamic funds may be included in this category. From Bloomberg’s categorization, we chose socially responsible funds, defined as “investing in securities of companies meeting socially responsible standards” and as “environmental, social, and corporate governance (ESG) funds investing in companies compliant with ESG criteria,” omitting thematic and religious funds.

⁴ As stated by Fung and Hsieh (1997, p. 276), in Sharpe’s (1992) model, “the focus is on the location component of return, which tells us the asset categories the manager invests in.” A limited number of asset classes are required to replicate the performance of an extensive universe of mutual funds. In our study, we regressed fund returns on a number of chosen factors, with specific constraints (i.e., the residual of the regression is uncorrelated with the factors, each coefficient is bounded in the [0,1] interval, and the sum of the coefficients is equal to 1).

⁵ These criteria are based on the ratings created by FTSE International Ltd. and Ethical Investment Research Services (EIRIS). For a more detailed listing of various SR indexes, see Appendix A.

⁶ We tested several alternative specifications, all of which proved less powerful in explaining SR fund returns. We also performed all our estimations using the DJSI US and DJSI Europe indexes as proxies for the US and European SR equity indexes, respectively. In addition, we replaced the SR indexes listed in Appendix A with their conventional peers without any sectorial exclusions. These results are available upon request.

⁷ We thank one of the reviewers for this suggestion.

⁸ The MSCI index recorded an annualized average return of 7.30% and an annualized standard deviation of 14.41% over the study period (October 2004–August 2015).

⁹ MSCI uses a market-capitalization approach in conducting the size segmentation of the market. The large-cap index consists of the 300 largest companies by full market capitalization in the investable market segment, the mid-cap index comprises the next 450 companies, and the small-cap index consists of the remaining 1,750 companies. Value and growth indexes are constructed on the basis of a segmentation of the stock universe according to various stock characteristics. Stocks are classified as value stocks if they have a high book-value-to-price ratio, a high 12‐month forward-earnings-to-price ratio, and a high dividend yield. Growth stocks have a high growth rate for both long‐term and short-term forward earnings per share, a high current internal growth rate, a long‐term historical earnings per share growth trend, and a long‐term historical sales per share growth trend. Z-scores are calculated and used to determine the overall style characteristics of each security in the MSCI value and growth two-dimensional style space. For more details, see MSCI US Equity Indices Methodology (2011).

¹⁰ For a detailed presentation of the SR rating process, see MSCI ESG Research (2014, 2015).

¹¹ In line with Statman and Glushkov (2009), we excluded nonrated companies from each index.

¹² A positive (negative) interaction effect comes from the positive (negative) correlation between the total return and the residual term in the regression.

¹³ The funds in our sample exhibit significant loadings on the factors representing the difference between SR and conventional benchmarks after controlling for the market exposure. For 53% of the US funds and 56% of the global funds, these loadings are statistically significant at the 5% level.

¹⁴ In practice, asset management companies use an internal nonfinancial rating system that can depart significantly from the public systems used by index providers. Thus, SR portfolio managers may depart from benchmarks when applying their SR screens. Active portfolio management may appear not only as a tactical allocation practice aimed at reaching the highest risk–return profile but also as a way to introduce internal rating recommendations into funds.

¹⁵ There are significant opportunities for diversification in the fixed-income asset class (Brière and Szafarz 2008).