Do ESG funds make stakeholder-friendly investments?

In March 2021, the Securities and Exchange Commission (SEC) created the Climate and ESG Task Force to proactively identify misconduct related to environmental, social, and governance (ESG) issues.¹ The taskforce was created in response to a recent explosion of interest in incorporating ESG factors into investment decisions. The asset management industry has responded to the demand for ESG investing by launching numerous “socially responsible” funds that nominally account for factors considered important to a firm’s overall sustainability: the environment (e.g., carbon emissions), social issues (e.g., employee treatment), and governance (e.g., executive compensation). According to the U.S. Forum for Sustainable and Responsible Investing, more than $12 trillion of assets under management (AUM) is explicitly linked to ESG issues. The availability of ESG funds is growing rapidly; Morningstar documents a nearly 50% increase in the number of ESG funds available in the United States from 2019 to 2020 alone.

However, the creation of the SEC’s new taskforce represents regulatory concern that ESG funds are not providing asset owners with products that actually reflect concern for stakeholder welfare. In April 2021, the SEC stated that it had identified several asset managers that were misleading investors by marketing funds as ESG-friendly but not making investment decisions consistent with such marketing or,² in some cases, even having a mechanism to “reasonably track” or screen portfolio firms’ ESG performance.³ This concern is shared by many members of the asset management industry. For example, in a recent op-ed, BlackRock’s former Chief Investment Officer for Sustainable Investing, Tariq Fancy, states,⁴ “Our messaging helped mainstream the concept that pursuing social good was also good for the bottom line. Sadly, that’s all it is, a hopeful idea. In truth, sustainable investing boils down to little more than marketing hype, PR spin and disingenuous promises from the investment community.”

To ensure that our results are not driven by heterogeneity in asset managers, we limit our main analyses to funds issued by financial institutions that also issued at least one non-ESG fund in the same year; i.e., we compare ESG funds to non-ESG funds managed by the same financial institutions in the same year.⁵ We begin with a comprehensive list, published by Morningstar, of mutual funds based in the United States that self-identify as ESG-oriented. We emphasize that although the list is compiled by Morningstar, it does not reflect Morningstar’s determination of which funds qualify as ESG-oriented. Rather, Morningstar is simply providing an aggregated list of self-identified ESG funds, primarily based on those funds’ prospectuses, fund reports, and websites. While there is substantial heterogeneity in how individual ESG funds claim to select stocks, our reading of several ESG funds’ prospectuses, summarized in Appendix A, suggests that the funds’ stock selection process can roughly be characterized as drawing on both ESG factors and valuation attributes (e.g., value or momentum). However, the funds largely characterize these two sets of criteria as distinct rather than interrelated, in many cases even cautioning that financial performance may suffer as a result of incorporating ESG factors into the investment process. After applying screens for data availability, we identify 147 distinct mutual funds, issued by 74 distinct asset managers, over the period 2010–2018 that claim to be ESG-oriented. We track the stakeholder-related behavior of stocks included in and added to these funds, relative to stocks held by 2428 non-ESG funds run by the same financial institution-years.

We find no evidence that ESG funds’ portfolio firms outperform non-ESG funds’ portfolio firms with respect to most of the measures of stakeholder-centric behavior that we consider in this paper. In fact, we find that ESG funds’ portfolio firms have significantly more violations of labor and environmental laws and pay more in fines for these violations, relative to non-ESG funds issued by the same financial institutions in the same year. Moreover, we find that ESG funds’ portfolio firms, on average, exhibit worse performance with respect to carbon emissions, in terms of both raw emissions output and emissions intensity (i.e., CO₂ emissions per unit of revenue). These results undermine such funds’ claims that they are picking socially responsible stocks for inclusion and suggest substantial greenwashing on the part of ESG funds.

Because carbon emissions disclosure is not mandatory, we also investigate voluntary emissions disclosure.⁶ We find that ESG funds are more likely to pick stocks that voluntarily disclose emissions. Given our other results, these tests suggest that ESG funds may be concerned about the existence of firms’ disclosures rather than the content of the information being disclosed. This finding is consistent with recent work (Drempetic et al. 2017; Lopez-de-Silanes et al. 2019) suggesting that firms’ ESG scores are influenced more by the existence of voluntary disclosures than by the information content of these disclosures.

Our results suggest that ESG funds may rely on ESG scores rather than performing their own due diligence about firms’ environmental and social practices. This conjectured practice, in turn, may lead to investments in firms with poorer levels of stakeholder treatment relative to firms that do not actively incorporate ESG into investment decisions. We provide support for this argument by assessing ESG scores directly. While we do find that ESG funds’ portfolios contain firms with higher average ESG scores, we also find, consistent with Drempetic et al. (2017) and Lopez-de-Silanes et al. (2019), that these scores are only correlated with metrics that capture news coverage and the existence and quantity of voluntary disclosure (both with respect to emissions and more generally), but not with the actual content of such disclosures.

Given the findings documented above, we turn to fund managers’ potential motivations for offering ESG products. We find that ESG funds (i) obtain lower stock returns but (ii) charge higher management fees. Moreover, while it is plausible that ESG funds invest in companies with the goal of improving how these companies treat their stakeholders, we test and find no empirical support for this conjecture. Our results raise questions about what purchasers of shares in self-labelled ESG or “socially responsible” mutual funds get in exchange for this higher management fee. We view this as a particularly salient concern, given that in recent years it has become easier for investors to identify and purchase what they believe to be ESG-oriented investment products, which, in turn, has led to a significant increase in inflows to such funds (Hartzmark and Sussman 2019).

Another potential explanation for our findings thus far is that ESG funds may primarily focus on the “G” in ESG (i.e., corporate governance). We find mixed results in this respect. Relative to non-ESG funds, ESG funds’ portfolio firms have lower excess CEO compensation but also lower levels of board independence. The governance characteristics described are typical of high-technology firms, which often have powerful founder-CEOs and boards of directors with limited oversight and influence over top executives. This characterization is consistent with our finding that ESG-oriented funds contain 27% more technology stocks than non-ESG funds issued by the same financial institutions in the same year.

A key takeaway of our study is that asset managers do not necessarily “walk the talk” in following through on claims of picking stocks that engage in stakeholder-friendly behavior. In light of the recent explosion of interest in ESG investing, our study is timely because prior empirical research that examines whether ESG funds actually deliver on their promises to focus on stakeholder-oriented stocks is sparse. Our study complements contemporaneous working papers in another setting by Gibson et al. (2021), Liang, Sun, and Teo (2022), and Kim and Yoon (2021). These studies consider the ESG performance of asset managers that sign the United Nations Principles for Responsible Investment (PRI), pre and post signing. Gibson et al. (2021) find that international US-domiciled institutions that publicly commit to responsible investing by signing the PRI are less likely to pick firms with better environmental, social, and governance (ESG) scores. Liang, Sun and Teo (2022) find that hedge funds that sign the PRI but have lower weighted-average ESG scores also underperform financially. Finally, Kim and Yoon (2021) find that PRI signatories do not improve fund-level weighted average ESG scores, and exhibit a decrease in returns earned, after signing the PRI. Our results are consistent with these studies.

However, our study exhibits two key differences from theirs. First, we consider individual mutual funds, holding constant any differences across asset managers. In contrast, both Gibson et al. (2021) and Kim and Yoon (2021) consider behavior at the asset manager level. Second, and most important, we focus on fundamental measures of stakeholder treatment – based on firms’ misconduct against consumers, employees, and the environment as well as carbon emissions and key aspects of firms’ governance – rather than on commercial ESG scores (as the studies above do). This distinction is important because, unlike Gibson et al. (2021) and Kim and Yoon (2021), we do find a positive correlation between the ESG fund label and ESG scores. We provide evidence that these constructs – commercial ESG scores and fundamental track records toward stakeholders – do not correlate well. We show that this is because commercial ESG scores appear to be driven by news coverage of ESG (consistent with Yang 2021) as well as the existence of firms’ voluntary disclosures about ESG information. ESG scores, however, do not correlate with the actual contents of these voluntary disclosures, consistent with Drempetic et al. (2017) and Lopez-de-Silanes et al. (2019). In considering direct measures of stakeholder treatment rather than commercial ESG scores, our paper is also timely and relevant for the SEC’s new Climate and ESG Task Force. While ESG-oriented investors have long relied on commercial scores to reduce information acquisition costs, the SEC, in its April 2021 ESG Risk Alert bulletin, specifically highlighted overreliance on composite ESG scores as a sign of inadequate due diligence and of poor fund-level compliance more generally. The SEC’s bulletin emphasized the need to understand whether ESG funds that claim to incorporate specific environmental or social factors into portfolio allocation decisions actually pick stocks that obtain superior performance with respect to these stated factors.⁷

A skeptic may argue that the goal of many ESG funds is simply to pick stocks that manage ESG risks in a way that maximizes firm value (i.e., financial materiality and not environmental and social materiality).⁸ That is, do fund managers claim to focus on socially responsible stocks, as we implicitly assume, or do they instead claim to focus on stocks that effectively manage the financial impacts of ESG risks? For instance, if BlackRock offers an ESG fund based on MSCI’s ESG indices and claims that the fund selects stocks that are resilient to ESG risks, is this equivalent to claiming that the fund is picking socially responsible stocks?

We acknowledge this alternative explanation with three counterpoints. First, in section 3.1, we provide a detailed discussion of how Morningstar compiles its list of sustainable funds and hence our sample. Notably, Morningstar states that it includes funds whose prospectuses explicitly allude to ESG considerations as a direct reason for investing in specific companies but excludes funds that simply incorporate ESG factors in a “more limited way.” Second, we selected the top 25 funds, sorted by assets under management, in our sample – reflecting 97% of total AUM for the last year of our sample – and read the prospectuses associated with those funds. In 22 out of 25 cases tabulated in Appendix A, we found language suggesting that the funds explicitly consider ESG-related issues as reasons to invest in, avoid, or divest from individual stocks, without citing the potential financial consequences of those firms’ ESG issues as a reason for picking or avoiding them. Collectively, this suggests to us that many – if not all – ESG fund managers do explicitly claim to focus on social responsibility for reasons other than financial materiality. Third, because of disclosure and data limitations, linking ESG factors to explicit financial consequences is non-trivially difficult in practice (Rajgopal 2022). It is therefore not obvious what precise analyses ESG funds conduct to identify value-enhancing strategies vis-à-vis ESG risks. Nonetheless, we agree that this may not be how all fund managers behave. We thus acknowledge the alternative hypothesis that some ESG funds in our sample do not explicitly attempt to pick socially responsible funds and that this may partially explain the results we document.

In this section, we summarize prior work related to our paper and outline how we measure whether ESG funds follow through on their promises of social responsibility.

Our objective in this paper is to focus on funds that claim to select only socially responsible portfolio firms, as these reflect a convenient way for investors to “put their money where their mouth is” with respect to ESG issues. We test whether ESG funds appear to pick stocks that have superior performance related to a number of fundamental measures of stakeholder treatment. In our tests described in this section and tabulated in Sections 5.1 through 5.5, the unit of analysis is the fund level; i.e., we consider the characteristics of ESG funds’ full portfolios and compare them to the full portfolios of non-ESG funds. Our aim in Sections 4 and 5 is to compare stocks held by ESG and non-ESG funds.

We concentrate on compliance violations as a measure of fundamental underlying ESG performance. We expect high-ESG firms to treat their employees and the environment better, which should be reflected in a lower rate of compliance violations. To test whether ESG-oriented mutual funds choose stock portfolios comprised of funds with superior labor and environmental records, we therefore compare the violation records of ESG funds’ portfolio firms to the violation records of non-ESG funds’ portfolio firms. More specifically, for each mutual fund-year, we construct a value-weighted outcome measure as follows: where v_int denotes the total value of fund i’s shareholdings in portfolio stock n in year t, and FirmBehavior_nt denotes one of our dependent measures corresponding to firm n in year t (e.g., an indicator variable for the presence of violations or the dollar amount of said violations, proxies for corporate governance). For example, if FirmBehavior_nt is an indicator for whether firm n committed a labor violation in year t, and mutual fund i held $10 worth of shares in companies that committed labor violations in year t and $30 worth of shares in companies that did not commit labor violations in year t, BEHAVIOR_it would equal \( \frac{10}{10+30}=0.25 \). We compute v_int based on calendar-year average holdings for each fund i in each stock n. When considering violations, we exclude from both the numerator and denominator of BEHAVIOR_it any firms that never show up in Good Jobs First’s Violation Tracker or Subsidy Tracker databases (i.e., firms for which we never see a violation or a subsidy). For example, if a fund holds $95 in portfolio firms covered in the Good Jobs First data universe and $5 in portfolio firms not covered in the Good Jobs First data universe, the denominator of BEHAVIOR_it – that is, \( {\sum}_n^N{v}_{int} \) – would equal $95. We take this step to avoid erroneously labeling a firm as a non-violator when it may simply be the case that Good Jobs First has not provided parent-subsidiary linkages for that firm. Because Violation Tracker and Subsidy Tracker cover nearly all of the Russell 3000 – and more than 90% of US public equities by market capitalization – we do not view this as a significant constraint. For other measures of FirmBehavior_nt, we exclude from the numerator and denominator of BEHAVIOR_it any firms not covered in the relevant databases (e.g., we exclude firms not covered by Trucost from the calculation of \( {\sum}_n^N{v}_{int}{FirmBehavior}_{nt} \) and \( {\sum}_n^N{v}_{int} \) when BEHAVIOR_it reflects carbon emissions).

Using the measures of BEHAVIOR_it above, we compare ESG funds to non-ESG funds by estimating the following series of regressions: where i indexes the CRSP portfolio, j indexes the asset manager (for example, BlackRock or JPMorgan), and t indexes the year. Our research design relies on asset manager-by-year fixed effects η_jt. We employ this fixed effect structure to ensure that our results are not driven by unobservable differences in portfolio selection across asset managers (e.g., the possibility that some asset managers are across-the-board better than others in accounting for how portfolio firms treat stakeholders) or across time (e.g., the possibility that ESG mutual funds, in general, have changed their ESG preferences over time). Moreover, our use of asset manager-by-year fixed effects rather than separate asset manager and year fixed effects ensures that we also account for any changes in investment strategy by asset managers over time.¹⁷ To enable maximal comparability of ESG funds and non-ESG funds, we also limit the sample to portfolios run by asset managers that also manage at least one ESG fund in the same year.¹⁸ Our approach of comparing ESG funds to other funds issued by the same asset manager-year combination also allows us to construct a natural benchmark for each of the ESG funds in our sample. More specifically, in labelling certain products as “ESG” funds, the asset manager is signaling that funds with the “ESG” label should place a greater emphasis, vis-à-vis its other fund offerings, on stakeholder-centric issues. We test whether such signaling is reflected in the asset manager’s ESG fund offerings.¹⁹

We employ several common fund-level control variables in Eq. (1) above. These include the extent to which the fund is available to retail investors; the natural logarithm of the fund’s age, in years; the natural logarithm of the fund’s total net assets (aggregated to the CRSP portfolio level, i.e., across variants of the same portfolio offered as different funds); and the fund’s financial performance and risk, measured as annual returns and annual return volatility, respectively. In addition, although a mutual fund does not have an “industry” in and of itself, it is possible that funds with different industry mixes in their portfolios will display mechanical differences in ESG performance. For example, a fund that focuses on technology stocks is less likely to have a high number of environmental violations by its portfolio firms, while a fund that focuses on oil and gas firms is likely to have more environmental violations. To ensure that our results are picking up more than just differential industry compositions across ESG funds’ and non-ESG funds’ portfolios, we also control for the percentage of the fund’s assets that are invested into (i) technology stocks; (ii) oil, gas, and coal stocks; and (iii) sin stocks. We measure technology stocks following the Bureau of Labor Statistics’ approach, as outlined in Hecker (2005); oil, gas, and coal stocks as being in two-digit SIC codes 12 and 13; and sin stocks following Hong and Kacperczyk (2009).

A combined read of the evidence presented in the paper suggests that the correlation between self-proclaimed high-ESG funds and their return and stakeholder-friendliness records is underwhelming. These results raise several questions about whether the declaration of high-minded ideals by ESG funds that charge higher management fees is cheap talk and whether commercially available ESG ratings really capture a firm’s ESG orientation. We caveat that, in light of recent work by Gibson et al. (2021) suggesting that greenwashing is stronger in the United States, the extent to which our results may generalize to other countries is an open question. Nonetheless, in light of a recent explosion in ESG investing by both retail and institutional investors, our work suggests a need for better verification of ESG funds’ claims of picking stakeholder-friendly stocks.

We end with the caveat that some ESG funds in our sample may not explicitly focus on potential portfolio firms’ social responsibility, and that this may partially explain the results we document. However, our review of the prospectuses of the top 25 funds in our sample suggests that fund managers pick stocks using a combination of valuation-based factors and ESG considerations. How much weight is attached in practice by ESG funds to valuation considerations as opposed to ESG factors is a question worthy of future research.