Faculty of law blogs / UNIVERSITY OF OXFORD

True Expectations, the Expected Mediocre Long-Term Returns of ESG Index Funds


Bernard S. Sharfman
Senior Corporate Governance Fellow, RealClearFoundation


Time to read

6 Minutes

OBLB types


OBLB keywords

Index funds

Investing in Environmental, Social, and Governance (‘ESG’) index funds is currently in vogue, now totaling around $250 billion. Yet, the US Department of Labor (‘DOL’) has taken a very negative approach to their use by employee pension benefit plans in a recently proposed rule,  Financial Factors in Selecting Plan Investments Proposed Regulation. What does the DOL know that so many investors around the globe just don’t get? 

The answer is that ESG index funds, even when the higher fees required to own these funds are excluded from the financial calculations, should only be expected to yield mediocre long-term risk-adjusted returns.  This is due to the indices that make up these funds not taking into consideration the positive skewness of market returns, lowering expected portfolio returns, and creating portfolios that have not fully diversified away unsystematic risk.  

To create ESG indices, portfolio selection will be based on some sort of screening mechanism.  As defined in my recent comment letter to the DOL, ESG portfolio screening is ‘a process by which a plan manager [investor] reduces its universe of eligible investments based on non-pecuniary [non-financial] factors [objectives].’ Screening criteria can be based on many factors including environmental factors, use of dual class shares, ESG ratings, alcohol- or tobacco-related, diversity, executive compensation, workforce compensation or working conditions, unionization, etc.

For example, consider the selection criteria utilized in the MSCI KLD 400 Social Index, the index used by BlackRock’s iShares MSCI KLD 400 Social ETF, an ESG ETF with approximately $2 billion in assets as of July 7, 2020:

The MSCI KLD 400 Social Index is maintained in two stages. First, securities of companies involved in Nuclear Power, Tobacco, Alcohol, Gambling, Military Weapons, Civilian Firearms, GMOs and Adult Entertainment are excluded. Then additions are made from the list of eligible companies based on considerations of ESG performance, sector alignment and size representation. The MSCI KLD 400 Social Index is designed to maintain similar sector weights as the MSCI USA Index and targets a minimum of 200 large and mid-cap constituents. Companies that are not existing constituents of The MSCI KLD 400 Social Index must have an MSCI ESG Rating above “BB” and the MSCI ESG Controversies Score greater than 2 to be eligible. At each quarterly Index Review, constituents are deleted if they are deleted from the MSCI USA IMI Index, fail the exclusion screens, or if their ESG ratings or scores fall below minimum standards. Additions are made to restore the number of constituents to 400. All eligible securities of each issuer are included in the index, so the index may have more than 400 securities. The selection universe for the MSCI KLD 400 Social Index are large, mid and small cap companies in the MSCI USA IMI Index.

Investment funds that use this index will have a significantly reduced number of stocks held in portfolio. First, there is an up-front screen to exclude a large number of investments based on moral and ethical reasons (negative screen). Second, another round of exclusions is based on an investment not having a minimum ESG rating or score. However, additions are made from the list of eligible companies based on considerations of ESG performance, sector alignment, and size representation (positive screen). All qualified securities are included in the index. Even so, the result is a relatively small portfolio of roughly 400 stocks out of a universe of 2,344 stocks that make up the MSCI USA IMI Index.


Portfolio Screening and Positive Skewness

Screening techniques based on non-pecuniary factors lead to a reduced number of stocks in a portfolio and therefore an increased probability that the big winners in the stock market will be excluded from or underweighted in an investment portfolio. The result will be reduced expected returns versus a comparable benchmark. This is a very important point for plan managers when selecting mutual funds or ETFs to invest in or making them available to plan participants and beneficiaries in self-directed accounts.

In Hendrik Bessembinder’s recent path-breaking article ‘Do Stocks Outperform Treasury Bills?’, he observed that there is a significant amount of positive skewness (‘a distribution with a longer right tail with higher probability for extremely high gains’) in the returns of individual public companies (common stock) that made up the stock market from July 1926 to December 2016. He found that ‘in terms of lifetime dollar wealth creation’ (ie, ‘accumulated December 2016 value in excess of the outcome that would have been obtained if the invested capital had earned one‐month Treasury bill returns’), ‘the best-performing 4% of listed companies explain the net gain for the entire US stock market since 1926, as other stocks collectively matched Treasury bills.’ His results also showed that the sum of the individual contributions to lifetime dollar wealth creation provided by the top 50 companies represented almost 40 percent of total lifetime dollar wealth creation. Thus, the returns earned by a relatively small number of best-performing companies were critical to the stock market earning returns above short-term Treasuries.

The understanding that positive skewness exists in stock-market returns means that investors are best served if those select few firms that are expected to be the best performers are given the maximum opportunity to show up in an investment fund’s portfolio. If investment funds want to maximize risk-adjusted returns, weeding out investments based on non-pecuniary factors is not the way to accomplish this objective. It is simply an additional constraint on the ability to maximize. 

Portfolio Screening and Overweighting

The use of portfolio screening based on non-pecuniary factors may also result in the overweighting of certain industries. This lack of portfolio diversification adds extra unsystematic risk to the ex-ante risk-adjusted return calculation. This extra risk cannot be ignored when an ESG fund is being evaluated for its risk-adjusted return.

Overweighting in certain sectors can also give the appearance that a portfolio’s stocks, such as many ESG funds, are performing much better than they appear, relative to their peers. As Vincent Deluard observed, ESG funds are currently overweighted in the health-care and technology industries, the two best-performing sectors in the first part of 2020.  As pointed out by Mitch Goldberg, ‘there are two likely reasons why a fund could outperform its benchmark. Either by overweighting the outperforming sector, or by lowering the expense ratio. In the case of the recent strong run for some ESG funds, it looks like the answer is an overweight to the technology sector.’

The result of this recent overweighting in the health-care and technology industries in ESG funds has led some to claim that ESG is an ‘equity vaccine’ in times of declining share prices. However, this is not correct. As stated by James Mackintosh:

Even where an ESG index did beat the market, it had little to do with environmental, social or governance issues. Instead, it came down to luck; did they happen to pick the stocks that best rode out coronavirus lockdowns? It is better to be lucky than right; but having, as some did, less exposure to cruise liners or long-haul airlines because of their carbon footprint was luck, not a well-thought-out way to avoid the stocks hurt most by Covid-19. There are several reasons why Microsoft tends to score well on ESG, but its cloud services being in demand because everyone is working from home isn’t among them.

Moreover, a recent empirical study ‘ESG Didn’t Immunize Stocks Against the COVID-19 Market Crash’ (Demers, Joos, Hendrikse, Lev, 2020) demonstrated that ESG did not serve as an equity vaccine during the COVID-19 market crash. They found that “ESG is insignificant in fully specified returns regressions for the first quarter of 2020 COVID crisis period, and it is negatively associated with returns during the market’s ‘recovery’ period in the second quarter of 2020.” Moreover, ESG scores provide very little (1% in the first quarter of 2020 and 3% in the second quarter of 2020) in the way of explanatory power. Instead, they found that ‘industry affiliation, market-based measures of risk, and accounting-based variables that capture the firm’s financial flexibility (liquidity and leverage) and their investments in internally-developed intangible assets together dominate the explanatory power of the COVID returns models.’ Importantly, the use of an extensive menu of control variables, including these and others, is what distinguishes their work from other research reports:

Contrary to the findings of contemporaneous studies that do not include such a full set of controls …, as well as to the widespread claims by fund managers [eg, claims made by BlackRock, Inc.], ESG data purveyors, and the financial press who seem to arrive at their conclusions on the basis of simple pairwise correlations, our results provide robust evidence that ESG is not significantly associated with stock market performance during the first quarter of 2020 once the full array of other expected determinants of returns have been controlled for.

In sum, portfolio overweighting that results from the use of non-pecuniary factors is a risk factor, not an enhancement to the expected financial performance of a fund, no matter how well the fund appears to perform in the short term.


Bernard S. Sharfman is a Senior Corporate Governance Fellow at the RealClearFoundation. 

Mr Sharfman would like to thank Attorney Albert Feuer for his helpful comments. Mr Feuer takes no position on this post’s conclusions. The opinions expressed here are the author’s alone and do not represent the official position of RealClearFoundation or any other organization with which he is currently affiliated.  


With the support of