Risk-Adjusted Performance of ESG and Non-ESG ETFs Across Market Regimes

Abstract

The rapid growth of environmental, social, and governance (ESG) investing has intensified the debate regarding whether ESG-oriented investment strategies exhibit performance patterns that differ from those of conventional investments, particularly during periods of market disruption. This study examines the risk-adjusted performance of ESG-oriented and non-ESG exchange-traded funds (ETFs) across market regimes surrounding the COVID-19 shock. The analysis classifies 28 passively managed ETFs into four sustainability-based categories and evaluates their performance using factor-based asset pricing models derived from the Fama–French framework. Additional analyses assess benchmark-relative performance using the S&P 500 and MSCI World indices and consider alternative ETF classifications based on investment mandates. The study estimates regime-specific regressions for the pre-COVID, COVID, and post-COVID periods. The results indicate that performance patterns vary across market regimes and ETF categories. Non-ESG ETFs tend to underperform on a risk-adjusted basis during the pre-COVID period, although this effect disappears thereafter. ESG-oriented ETFs generally exhibit limited evidence of abnormal performance, while factor exposures vary across regimes, reflecting changes in sector composition and macro-financial conditions. The findings suggest that, in addition to ESG orientation, market regimes and sectoral exposures play an important role in explaining differences in ETF performance.

1. Introduction

The rapid expansion of environmental, social, and governance (ESG) investing has generated growing interest among practitioners and academics in socially responsible investing (SRI). While SRI has existed in various forms for several decades, the United Nations Global Compact report provides formal guidance on integrating ESG considerations into asset management and investment decision-making (United Nations 2004). Modern Portfolio Theory (Markowitz 1952) and Post-Modern Portfolio Theory (Rom and Ferguson 1993; Sortino and van der Meer 1991) have long guided portfolio construction and investment decisions. However, these frameworks focus primarily on financial metrics such as expected returns and risk and do not explicitly incorporate non-financial considerations such as ESG criteria.

While ESG investing aims to offer benefits from a societal perspective, the broad integration of ESG criteria into investment decision-making remains puzzling, as it is unclear whether ESG strategies deliver favorable outcomes from an investor’s perspective. Specifically, it remains an open question whether investors who incorporate social responsibility considerations and restrict their investment choices must accept lower financial performance relative to those who do not. Some studies show that the returns of ESG-related exchange-traded funds (ETFs) outperform their non-socially responsible counterparts (ElBannan 2024; Meziani 2020; Rompotis 2023; Supatgiat et al. 2025; Vasiliauskaitė et al. 2025), while others arrive at opposing conclusions (Baklaci et al. 2024; Milonas et al. 2022; Rompotis 2022).

Meehan and Corbet (2025) argue that crises such as the COVID-19 pandemic have raised important questions regarding the performance and resilience of investment vehicles, particularly ETFs. In this context, the studies by ElBannan (2024), Gheorghe et al. (2025), and Supatgiat et al. (2025) report the favorable performance of ESG ETFs during the COVID-19 crisis. Nevertheless, this evidence is not supported by the findings of Folger-Laronde et al. (2022), Pavlova and de Boyrie (2022), and Vasiliauskaitė et al. (2025).

Framing these conflicting results, in the U.S., the influence of a broader social and political context needs consideration. In recent years, there has been a surge of anti-ESG pressures associated with political initiatives that may influence the selection of ESG-related investments. For example, Danyu-Zhang (2023) suggests that since fiduciary responsibilities are determined by politically motivated judges, retirement plan managers have good reasons to retreat from potential ESG asset choices to minimize legal liability risk. Tang et al. (2024) study the market reaction to anti-ESG regulations introduced in eighteen states between 2021 and 2023 and report that energy-related firms experienced cumulative abnormal returns between 0.8% and 3.5% around the regulation approval dates. The market’s response suggests that, in the absence of restrictive regulations, ESG investments are preferred substitutes for energy related securities. (Curtis 2025) and Harmes (2025) also report that a backlash has emerged against the use of ESG criteria by asset managers, in response to several states adopting measures to limit the inclusion of ESG investments in public pension funds. Moreover, the U.S. Department of Labor (DOL)’s guidance on fiduciary duties issued under the Employee Retirement Income Security Act (ERISA) in 2020 emphasized that fiduciaries must base investment decisions on economic risk and return, and prohibited them from sacrificing returns or taking on additional risk to promote non-financial or environmental, social, and governance (ESG) goals, making clear that ESG factors may be considered only to the extent they are expected to have a material effect on risk or return. While the initial guidance was relaxed later, it had an impact on portfolio managers’ asset selection (Department of Labor 2020).

Collectively, the existing evidence provides no clear consensus on the financial performance of ESG-oriented investment strategies. Prior studies report conflicting results across markets, asset classes, and periods of market stress, suggesting that the relationship between ESG orientation and investment performance may depend on broader economic and financial conditions. As noted by Joshi and Dash (2024), the performance implications of ESG investing remain relatively underexplored, particularly across different market regimes. The present study contributes to the ongoing discussion on ESG-related exchange-traded fund performance (ESG-ETFs). It provides a comprehensive evaluation of raw and risk-adjusted excess return differentials relative to the S&P 500 for a sample of 28 passively managed ETFs across distinct phases of the COVID-19 pandemic. The empirical framework also employs one- to six-factor asset pricing models (Carhart 1997; Fama and French 1993, 2015; Sharpe 1964) to distinguish abnormal performance from compensation for underlying risk exposures. The analysis spans the (1) pre-pandemic period from 3 January 2017, to 23 March 2020, (2) the pandemic period from 24 March 2020, to 12 July 2021, and (3) the post-COVID macro-financial adjustment period from 13 July 2021, to 31 December 2024. This subperiod approach allows a regime-based evaluation of the ETFs’ performance. ETFs are classified into four groups based on Morningstar sustainability ratings, namely ESG-aligned, ESG-neutral, ESG-unaligned, and non-ESG, enabling a systematic comparison across ESG orientations. In addition, four equally weighted portfolios are constructed for each ESG classification to facilitate a consistent comparison of performance dynamics across market regimes. This framework enables an examination of whether differences in ESG orientation are associated with systematically different risk-adjusted performance outcomes across periods of market stability, stress, and subsequent normalization.

By jointly examining ESG-aligned, ESG-neutral, ESG-unaligned, and anti-ESG ETFs within a unified empirical framework across pandemic-related market regimes, this study provides new evidence on how ESG orientation interacts with macro-financial conditions to influence ETF risk-adjusted performance.

The remainder of the article is organized as follows. Section 2 presents the theoretical basis of the study and a brief review of the empirical literature. Section 3 describes the data and outlines the research methodology. Section 4 presents the empirical findings, and Section 5 discusses the findings and concludes the paper.

2. Literature Review

2.1. Theoretical Basis

The relationship between asset returns and ESG performance can be interpreted through several complementary theoretical frameworks that account for heterogeneous market responses to ESG information. Del Gesso and Lodhi (2025) identify five dominant approaches, namely stakeholder, legitimacy, institutional, agency, and signaling theories, which are often combined within an integrated theoretical framework.

Stakeholder theory emphasizes that ESG initiatives may enhance firm value by strengthening relationships with key stakeholders and mitigating non-financial risks, supporting sustainable long-term performance rather than short-term gains (Freeman 1984). In the context of the present study, stakeholder theory would predict ESG-ETFs’ resilience to the COVID pandemic “stress” phase, while it would also explain a dampened response during the recovery phase, when cyclical, non-ESG funds typically thrive. By contrast, investment strategies that emphasize ESG-excluded sectors may offer higher expected returns as compensation for regulatory, reputational, and litigation risks, particularly during periods of economic expansion or sectoral rotation.

Legitimacy theory postulates that ESG practices are influenced by prevailing social norms and institutional contexts (Suchman 1995), while institutional theory explains how organizations perceive and respond to social change and institutional pressures (DiMaggio and Powell 1983; Meyer and Rowan 1977). Both theories suggest that the financial effects of ESG performance on ETF returns are likely to differ over time and across regions, something that may be empirically validated by observing the different ETF groups considered in this study through the pre-, post-, and COVID phases.

Agency theory further suggests that information disclosure reduces information asymmetry and agency costs (Jensen and Meckling 1976), so that greater ESG-related disclosure is expected to strengthen transparency and governance, as well as enhance monitoring mechanisms, potentially supporting improved financial outcomes over time (Carnini et al. 2022). In the context of the present study and consistent with stakeholder theory, agency theory suggests that improved ESG performance should be associated with better ETF performance during the more uncertain phases of the pandemic.

Finally, according to signaling theory, ESG disclosures, ratings, and changes in ESG performance may convey information about firm quality, risk management, and long-term orientation in settings characterized by information asymmetry, leading investors to reassess valuations and portfolio allocations (Spence 1973). The signaling theory related anticipation would be supportive of better ESG performance being reflected in better ETFs’ financial market performance.

While stakeholder and signaling theories imply that ESG-oriented firms and assets may benefit from greater resilience and investor support during periods of market stress, agency considerations also point to potential opportunity costs associated with ESG screening. Moreover, managers of firms and funds are expected to adjust to evolving social and institutional pressures over time. These theoretical perspectives imply that ESG performance may mitigate downside risk, although the costs associated with ESG may weigh on ETF performance during the recovery phase. Conversely, investment strategies that emphasize ESG-excluded sectors may offer higher expected returns as compensation for regulatory, reputational, and litigation risks, particularly during periods of economic expansion or sectoral rotation.

The empirical analysis of the different ETF groups’ performance during the three subperiods of the COVID pandemic offers an opportunity to corroborate the adequacy of the theoretical predictions proposed by these theories.

2.2. Review of Empirical Studies

A review of recent studies that report empirical evidence on ESG-ETFs’ market performance identified seven general topics, which may be used as a guide to organize this analysis. According to their findings, the studies may be grouped as follows: (1) ESG ETFs outperform conventional ETFs or benchmarks, (2) higher-ESG-graded ETFs outperform lower-ESG-rated ETFs, (3) ESG-ETFs exhibit distinctive stability during periods of uncertainty, (4) ESG ETFs do not generate a significant alpha or outperform conventional ETFs, (5) ESG ETFs underperform conventional ETFs, (6) anti-ESG funds outperform market benchmarks, and (7) anti-ESG funds underperform market benchmarks.

Studies reporting that ESG ETFs outperform conventional ETFs or other benchmarks include ElBannan (2024), which examines the role of ESG performance in explaining ETF returns and volatility during the market downturn associated with the COVID-19 pandemic; using a diversified sample of 160 iShares MSCI ESG/SRI, this study finds that they were resilient during the COVID pandemic-induced market crash and that they outperformed their conventional counterparts. Similarly, Rompotis (2023) reports that, on average, U.S. ESG ETFs outperform the S&P 500 in terms of raw returns, although this finding is not uniform across funds, as only a small subset of the ETFs examined displays positive and statistically significant alphas.

In their analysis of the role of ESG ratings in shaping market behavior, Supatgiat et al. (2025) report that higher-rated ESG ETFs outperform lower-rated counterparts, exhibiting higher or comparable returns, lower volatility, and smaller price jumps, particularly during periods of market stress such as the COVID-19 pandemic and the Russia–Ukraine conflict.

Gheorghe et al. (2025) analyze the regime-dependent and non-linear behavior of ESG and AI-themed ETFs and conclude that ESG ETFs display relatively stronger stability during periods of heightened uncertainty, particularly in terms of downside risk.

Studies reporting no significant alpha or outperformance of ESG ETFs relative to conventional benchmarks include Rompotis (2022), which examines 49 ESG ETFs in the United Kingdom and finds that, although these funds outperform the FTSE 100, they do not generate statistically significant alpha or differences in Sharpe and Treynor ratios relative to the benchmark. Similarly, Pavlova and de Boyrie (2022) evaluate the risk-adjusted performance of ESG ETFs before and during the COVID-19 market crash and report no significant differences in risk-adjusted performance across ESG rating groups during the pandemic-related downturn.

Additional evidence of ESG ETF underperformance appears in Meziani (2020), which finds that, despite some improvements in performance metrics, ESG ETFs continue to underperform conventional funds. Relatedly, Vasiliauskaitė et al. (2025) compare eight ETFs from the United States and Europe, including both ESG and conventional funds, and report that traditional U.S. ETFs achieve the highest returns and risk-adjusted performance, while European ESG ETFs exhibit the lowest downside risk. Their findings also indicate that ESG ETFs in the U.S. perform comparably to conventional funds, suggesting that ESG integration does not necessarily lead to inferior financial outcomes.

A segment of the capital markets industry provides investment vehicles and advisory strategies tailored to investors who question the need for firms to engage in ESG-related initiatives or oppose such expenditures on financial grounds. These approaches are commonly referred to as anti-ESG, and the literature has examined their performance to a more limited extent. Within this category, some studies report outperformance relative to conventional benchmarks. Hong and Kacperczyk (2009) show that so-called sin stocks earn higher returns than comparable firms even after controlling for standard risk factors, despite being avoided by norm-constrained investors and facing greater legal risk. Similarly, Chang and Krueger (2013) analyze the VICEX fund, which invests in anti-ESG firms, and find that it outperforms the S&P 500 over the 2003–2012 period.

Finally, some studies report that anti-ESG investments underperform their benchmarks. Meehan and Corbet (2025) examine the performance dynamics of socially responsible investment (SRI) and anti-ESG ETFs in the United States during the COVID-19 episode and present evidence that challenges earlier claims of an anti-ESG performance advantage, instead documenting greater resilience, particularly among SRI ETFs. Similarly, Rompotis (2024) compares 18 U.S. anti-ESG ETFs with the iShares ESG Aware MSCI USA ETF and the S&P 500 and finds that most anti-ESG ETFs underperform both the broad market and the selected ESG ETF in terms of absolute and risk-adjusted returns.

In summary, the empirical literature on the subject provides mixed and sometimes conflicting evidence on the performance of ESG-oriented and anti-ESG investment strategies. While some studies document enhanced resilience and favorable risk-adjusted performance of ESG ETFs during periods of heightened market stress, others find only limited or no differences at all in their risk-return performance, relative to conventional ETFs or benchmark portfolios. Furthermore, evidence on anti-ESG ETFs remains relatively limited and largely restricted to specific samples or time periods.

Few studies jointly analyze ESG-aligned, ESG-neutral, ESG-unaligned, and anti-ESG ETFs within a unified empirical framework that explicitly accounts for regime-dependent dynamics and risk-adjusted performance. This study contributes to the literature from a methodological perspective by employing widely used factor models to estimate risk-adjusted abnormal performance for both individual ETFs and equally weighted portfolios. Furthermore, the analysis evaluates raw and risk-adjusted excess return differentials and considers the distinct market regimes associated with different phases of the COVID-19 pandemic to examine the evolution of risk and return across the sample. The study further classifies ETFs according to both sustainability ratings and stated investment mandates, allowing a comparison of performance patterns across alternative classification schemes. These features provide a systematic assessment of ESG-oriented and non-ESG investment strategies across different market environments. Considering these gaps, the present study evaluates two related research hypotheses concerning the performance of ESG-oriented and non-ESG ETFs:

H₁: 

ESG-oriented ETFs exhibit risk-adjusted performance that differs from that of non-ESG ETFs and the market benchmark.

H₂: 

The risk-adjusted performance of ESG and non-ESG ETFs differs across market regimes, specifically across the pre-COVID, COVID, and post-COVID periods.

The hypotheses are evaluated using subperiod factor regressions to estimate abnormal returns (alphas) for each ETF category, followed by Wald tests that assess whether performance differs across regimes.

3. Data and Methodological Issues

3.1. Data

The selection of ETFs is based on the following criteria: (1) given the ongoing debate on whether managers add value in this asset class (e.g., Carneiro et al. 2022; Crane and Crotty 2018; Fox and Hammond 2022), the analysis focuses exclusively on passively managed funds; (2) price data are available from December 2016 onward; (3) the funds are traded on a U.S. exchange; (4) except for non-ESG ETFs, as defined in the following paragraph, the remaining funds are included in Liu (2020) classification; and (5) all ETFs are listed on ETFdb.com. The latter two criteria generally align the sample construction with the approach adopted by Pavlova and de Boyrie (2022). Application of these criteria resulted in a final sample of 28 ETFs spanning a broad range of sustainability-oriented and conventional investment strategies.

Table 1. Characteristics of the ETFs in the sample.

Ticker	Name	Core Strategy	Inception Date	Morningstar Globes	ESG Classification	Mandate Category
DSI	iShares MSCI KLD 400 Social ETF	U.S. social ESG index	14 November 2006	5	ESG-aligned	ESG-S
FIW	First Trust Water ETF	U.S. water management thematic	8 May 2007	5	ESG-aligned	THEM
GRID	First Trust NASDAQ Clean Edge Smart Grid ETF	Smart-grid infrastructure theme	16 November 2009	5	ESG-aligned	THEM
SUSA	iShares MSCI USA ESG Select ETF	ESG-screened U.S. large/mid caps	24 January 2005	5	ESG-aligned	ESG-S
ESGU	iShares ESG Aware MSCI USA ETF	Core U.S. equity exposure with integrated ESG optimization	1 December 2016	4	ESG-aligned	ESG-S
FAN	First Trust Global Wind Energy ETF	Global wind energy thematic	16 June 2008	4	ESG-aligned	THEM
ICLN	iShares Global Clean Energy ETF	Global renewable energy exposure	24 June 2008	4	ESG-aligned	THEM
SHE	SPDR MSCI USA Gender Diversity ETF	U.S. gender-diverse companies	7 March 2016	4	ESG-aligned	ESG-S
SPXE	ProShares S&P 500 ex-Energy	U.S. large- and mid-cap stocks, excluding the energy sector	22 September 2015	4	ESG-aligned	ESG-S
SPYX	SPDR S&P 500 Fossil Fuel Reserves Free ETF	S&P 500 excluding fossil fuel reserve firms	30 November 2015	4	ESG-aligned	ESG-S
CRBN	iShares MSCI ACWI Low Carbon Target ETF	Global low-carbon index; screens coal/oil	8 December 2014	3	ESG-neutral	ESG-S
PBD	Invesco Global Clean Energy ETF	Global clean energy equities	13 June 2007	3	ESG-neutral	THEM
PHO	Invesco Water Resources ETF	U.S. water infrastructure & purification companies	6 December 2005	3	ESG-neutral	THEM
PIO	Invesco Global Water ETF	Global water infrastructure equities	13 June 2007	3	ESG-neutral	THEM
SMOG	VanEck Low Carbon Energy ETF	Low-carbon energy focus	3 May 2007	2	ESG-unaligned	THEM
TAN	Invesco Solar ETF	Global solar energy thematic	15 April 2008	2	ESG-unaligned	THEM
PBW	Invesco WilderHill Clean Energy ETF	Global clean energy thematic (solar, wind, etc.)	3 March 2005	1	ESG-unaligned	THEM
AMLP	Alerian MLP ETF	Midstream energy MLPs	24 August 2010	2	Non-ESG	ENER
IEO	iShares U.S. Oil & Gas E&P ETF	U.S. E&P sector	1 May 2006	2	Non-ESG	ENER
IYE	iShares U.S. Energy ETF	Large-cap U.S. energy	12 June 2000	2	Non-ESG	ENER
MLPA	Global X MLP ETF	Midstream energy master limited partnerships (MLPs)	18 April 2012	2	Non-ESG	ENER
PEJ	Invesco Dynamic Leisure & Entertainment ETF	U.S. leisure & gambling sectors	23 June 2005	2	Non-ESG	DEF
PPA	Invesco Aerospace & Defense ETF	Aerospace and defense sector	26 October 2005	2	Non-ESG	DEF
VDE	Vanguard Energy ETF	Broad U.S. energy sector	23 September 2004	2	Non-ESG	ENER
XLE	Energy Select Sector SPDR Fund	U.S. energy sector (oil, gas, consumable fuels)	16 December 1998	2	Non-ESG	ENER
ITA	iShares U.S. Aerospace & Defense ETF	Aerospace & defense equities	1 May 2006	1	Non-ESG	DEF
XAR	SPDR S&P Aerospace & Defense ETF	U.S. aerospace & defense sector	28 September 2011	1	Non-ESG	DEF
XOP	SPDR S&P Oil & Gas E&P ETF	Oil & gas exploration & production	19 June 2006	1	Non-ESG	ENER

Source: Authors’ own.

presents the selected funds, which represent the complete set of ETFs meeting the study’s selection criteria at the time of sample construction.

We classify the ETFs in our sample based on Morningstar Sustainability Globe ratings. This rating follows a five-globe scale and reflects a fund’s ESG risk relative to other funds within the same Morningstar Global Category. ETFs receiving four or five Globes are classified as ESG-aligned, indicating lower ESG risk exposure relative to category peers. ETFs with three Globes are classified as ESG-neutral, reflecting sustainability characteristics broadly in line with peer-group averages. Funds receiving one or two Globes are classified as ESG-unaligned, indicating relatively higher ESG risk exposure. ETFs without explicit ESG screening or sustainability mandates are classified as non-ESG, including so-called sin funds.

Table 1 reports selected characteristics of the ETFs in the sample. Historical Sustainability Globe ratings for the full sample of the study were not available in publicly accessible databases. The classification is therefore treated as structural and cross-sectional rather than time varying. Because the ETFs in the ESG categories were launched with sustainability-oriented mandates, their core investment objectives remained stable throughout the sample period. However, the use of end-of-sample ratings introduces a potential look-ahead bias, which should be considered when interpreting the results.

To assess whether the use of end-of-sample Morningstar Sustainability Globe ratings could lead to misclassification, we reviewed historical prospectuses and registration filings for all ETFs in the sample using the EDGAR database of the U.S. Securities and Exchange Commission, complemented by ETF provider websites to verify subsequent mandates. The review indicates that the primary investment strategy of the ETFs remained broadly stable throughout the sample period. Most funds retained the same benchmark index, while a small number (e.g., SHE and SMOG) experienced benchmark changes without altering their core investment themes.

Therefore, from a mandate-based perspective and to further assess robustness, we conducted an additional analysis using an alternative ETF classification based on stated investment mandates and sectoral exposures rather than Morningstar ratings. Specifically, as shown in Table 1, we group ETFs into four categories: ESG-Screened or Exclusionary (ESG-S), Environmental or Thematic (THEM), Energy (ENER), and Defense (DEF). This classification relies on ex-ante fund objectives and provides a clearer separation between ESG orientation and sector-specific exposures.

The full sample spans eight years, from 3 January 2017, to 31 December 2024. Given the objective of examining ETF performance across distinct phases of the COVID-19 episode, the sample is divided into pre-COVID, COVID, and post-COVID market regimes. This classification is motivated by the COVID-19 finance literature and major macro-financial developments during the sample period. The first breakpoint corresponds to the March 2020 market trough and the abrupt increase in market volatility associated with the COVID-19 shock (Baker et al. 2020; Zaremba et al. 2021; Zhang et al. 2020). The second breakpoint captures the transition from the vaccine-driven recovery phase toward a post-COVID macro-financial environment characterized by economic reopening, sector rotation, rising inflation pressures, and expectations of monetary-policy normalization (BIS 2022; Board of Governors of the Federal Reserve System 2022; Gräb et al. 2021). Based on these considerations, the sample is divided into three regimes: pre-COVID (3 January 2017, to 23 March 2020), COVID (24 March 2020, to 12 July 2021), and post-COVID (13 July 2021, to 31 December 2024).

As a diagnostic check, we applied Bai and Perron’s (2003) multiple structural break methodology to the logarithmic daily returns of the S&P 500 Index. Three versions of the test are considered: global information criteria, sequential F-tests of L + 1 versus L breaks, and global F-tests allowing for up to M breaks. Although the tests do not identify statistically significant structural breaks at the 5% level, the estimated breakpoints cluster around the economically motivated dates described above. Thus, while the Bai–Perron results do not provide formal statistical evidence supporting the selected breakpoints, they are broadly consistent with the timing of major market transitions associated with the COVID-19 period.

To further assess whether the proposed regime classification captures meaningful differences in return dynamics, we conducted the Chow (1960) breakpoint test using 23 March 2020, and 12 July 2021, as break dates. The joint Chow test rejects the null hypothesis of constant coefficients at the 5% level, indicating that regression parameters differ across the three subperiods. Thus, the economic motivation, the clustering of Bai–Perron estimated breakpoints around the selected dates, and the statistically significant Chow test provide support for the use of the proposed regime classification in the empirical analysis. All the analyses in the study were conducted using R (version 4.5.1) and EViews 9.

3.2. Factor Models

Following prior studies (e.g., ElBannan 2024; Pavlova and de Boyrie 2022; Rompotis 2024; Sabbaghi 2025) we employ several factor models to assess the risk-adjusted abnormal performance of the ETFs and equally weighted portfolios: (1) the CAPM (Sharpe 1964), (2) the Fama–French three-factor model (Fama and French 1993), (3) the Carhart (1997) four-factor model, (4) the (Fama and French 2015) five-factor model, and (5) the Fama–French five-factor and momentum factor model.

$$R_{i,t}-R_{f,t}={\alpha }_B{D}_{B,t}+{\alpha }_D{D}_{D,t}+{\alpha }_A{D}_{A,t}+{\beta }_1\left(R_{m,t}-R_{f,t}\right)+{\epsilon }_t$$

(1)

$$R_{i,t}-R_{f,t}={\alpha }_B{D}_{B,t}+{\alpha }_D{D}_{D,t}+{\alpha }_A{D}_{A,t}+{\beta }_1\left(R_{m,t}-R_{f,t}\right)+{\beta }_2{SMB}_t+{\beta }_3{HML}_t+{\epsilon }_t$$

(2)

$$R_{i,t}-R_{f,t}={\alpha }_B{D}_{B,t}+{\alpha }_D{D}_{D,t}+{\alpha }_A{D}_{A,t}+{\beta }_1\left(R_{m,t}-R_{f,t}\right)+{\beta }_2{SMB}_t+{\beta }_3{HML}_t+{\beta }_6{WML}_t+{\epsilon }_t$$

(3)

$$R_{i,t}-R_{f,t}={\alpha }_B{D}_{B,t}+{\alpha }_D{D}_{D,t}+{\alpha }_A{D}_{A,t}+{\beta }_1\left(R_{m,t}-R_{f,t}\right)+{\beta }_2{SMB}_t+{\beta }_3{HML}_t+{\beta }_4{RMW}_t+{\beta }_5{CMA}_t+{\epsilon }_t$$

(4)

$$R_{i,t}-R_{f,t}={\alpha }_B{D}_{B,t}+{\alpha }_D{D}_{D,t}+{\alpha }_A{D}_{A,t}+{\beta }_1\left(R_{m,t}-R_{f,t}\right)+{\beta }_2{SMB}_t+{\beta }_3{HML}_t+{\beta }_4{RMW}_t+{\beta }_5{CMA}_t+{\beta }_6{WML}_t+{\epsilon }_t$$

(5)

where R_i,t is the return of the asset on day t, R_f,t is the risk-free rate, R_i,t − R_f,t is the asset excess return, and R_m,t − R_f,t is the market excess return. SMB_t and HML_t correspond to the size and value factors, respectively. WML_t represents the momentum factor, while RMW_t and CMA_t capture the profitability and investment factors, defined as the return differentials between portfolios of firms with robust versus weak profitability and low versus high investment. D_B,_t, D_D,_t, and D_A,_t are dummy variables that take the value of one in the pre-COVID, COVID, and post-COVID periods, respectively, and zero otherwise1. The analysis estimates standard errors using the Newey–West HAC procedure. We use t-tests on the dummy coefficients to assess whether the average abnormal return (alpha) in each subperiod differs statistically from zero.

Risk-adjusted performance is evaluated relative to the S&P 500 and standard U.S. equity risk factors. Although several ETFs in the sample track global or thematic indices, the S&P 500 provides a consistent and widely used benchmark from the perspective of U.S.-based investors. The use of Fama–French–Carhart factors also controls for common systematic risk exposures affecting a broad set of equity investments. Since the study focuses on comparing ESG-oriented and non-ESG ETFs within a unified framework, employing a single benchmark and factor structure ensures methodological consistency across categories. However, some ETFs have mandates extending beyond the U.S. market, which may introduce additional variation in factor exposures.

We compute Variance Inflation Factors (VIF) to assess multicollinearity. VIF values of 5 or higher indicate elevated multicollinearity and are considered problematic, whereas values below 5 suggest moderate multicollinearity and are regarded as acceptable within the model (Chawarura et al. 2025). In addition, following Haggard and Witte (2012), we evaluate regime dependence using Wald tests of equality across subperiod dummy parameters. Specifically, we first test the joint hypothesis that α_B = α_D = α_A; if this null hypothesis is rejected, we then examine each pairwise equality restriction. As a robustness check, we also assess all pairwise equality restrictions for the ETF equally weighted portfolios, irrespective of the outcome of the joint test.

After the estimation of models (1) through (5), we compute two alternative risk-adjusted return measures. The first one is the Sharpe ratio (SR_i) shown in Equation (6):

$${SR}_i{\displaystyle \frac{R_{i,t}-R_{f,t}}{{\sigma }_i}}$$

(6)

The second risk-adjusted return measure is the Treynor ratio (TR_i), as defined in Equation (7):

$${TR}_i{\displaystyle \frac{R_{i,t}-R_{f,t}}{{\beta }_i}}$$

(7)

In Equation (6) σ_i is the standard deviation of the asset’s excess return, while β_i in Equation (7) represents the asset’s market risk. The definitions of R_i,t, R_f,t, and R_i,t − R_f,t in both equations are the same as above. Consistent with prior research (Daniel et al. 2020; Ong and Herremans 2023), we use a 126-day rolling window (approximately six months) to compute daily Sharpe ratio values based on annualized excess return differentials, following Omura et al. (2021). Similarly, to compute the Treynor ratio, we estimate 126-day rolling beta coefficients by regressing each asset’s excess return differentials on market excess return differentials, as suggested by Daniel et al. (2020) and Harris et al. (2019).

4. Results

4.1. Descriptive Statistics

Table 2. ETFs Equally Weighted Portfolios Descriptive Statistics.

	Full Sample				Pre-COVID
	ESG-Aligned	ESG-Neutral	ESG-Unaligned	Non-ESG	ESG-Aligned	ESG-Neutral	ESG-Unaligned	Non-ESG
Mean	0.038	0.027	0.024	0.018	−0.001	−0.001	0.019	−0.107
Median	0.051	0.063	0.074	0.058	0.057	0.078	0.125	0.032
Maximum	9.393	8.790	12.863	10.847	7.513	7.680	5.651	8.651
Minimum	−11.483	−13.330	−13.759	−22.641	−11.483	−13.330	−13.759	−22.641
Std. Dev.	1.175	1.208	2.106	1.666	1.129	1.138	1.571	1.672
Skewness	−0.863	−0.982	−0.242	−1.812	−3.008	−3.557	−2.765	−4.923
Kurtosis	17.807	18.617	7.416	29.986	35.758	43.632	25.203	60.474
Jarque-Bera	18,629.21	20,770.40	1654.110	62,153.93	37,439.00	57,427.10	17,669.14	114,758.40
Jarque-Bera p-value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Sum	77.430	54.650	47.512	35.747	−0.595	−0.994	15.304	−86.702
Sum Sq. Dev.	2775.331	2935.175	8920.134	5582.517	1030.422	1048.564	1995.600	2260.975
Observations	2012	2012	2012	2012	810	810	810	810
	COVID				Post-COVID
	ESG-Aligned	ESG-Neutral	ESG-Unaligned	Non-ESG	ESG-Aligned	ESG-Neutral	ESG-Unaligned	Non-ESG
Mean	0.237	0.243	0.398	0.270	0.000	−0.027	−0.113	0.039
Median	0.279	0.297	0.529	0.072	−0.024	−0.027	−0.235	0.084
Maximum	9.393	8.790	12.863	10.847	5.775	6.237	8.596	3.986
Minimum	−6.246	−5.901	−7.520	−9.458	−4.234	−4.242	−8.090	−7.173
Std. Dev.	1.445	1.431	2.763	2.348	1.095	1.172	2.236	1.304
Skewness	0.540	0.466	0.096	0.441	−0.013	0.118	0.251	−0.377
Kurtosis	10.145	8.972	4.678	5.887	4.567	4.368	3.578	4.951
Jarque-Bera	713.6821	499.2368	38.98099	124.5858	89.42385	70.16553	21.33243	159.3096
Jarque-Bera p-value	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
Sum	77.632	79.576	130.692	88.400	0.393	−23.932	−98.484	34.049
Sum Sq. Dev.	682.760	669.838	2497.263	1802.334	1046.754	1198.294	4364.932	1485.411
Observations	328	328	328	328	874	874	874	874

Source: Authors’ own.

reports the descriptive statistics for the equally weighted portfolios constructed according to the four ESG-related ETF classifications. During the pre-COVID subsample, the non-ESG portfolio records negative average daily returns, whereas the remaining categories exhibit average returns close to zero. In contrast, during the post-pandemic subperiod, the pattern reverses for the non-ESG portfolio, while the ESG-neutral and ESG-unaligned portfolios show negative average returns. All portfolios generate positive returns during the COVID-19 subsample. Volatility, measured by the standard deviation, increases across all categories during the COVID-19 episode relative to the pre-pandemic period, with the ESG-unaligned portfolio exhibiting the largest increase. In the post-COVID subperiod, the standard deviation declines for all four portfolios compared to the pandemic subsample; however, the non-ESG portfolio records the largest reduction, while the ESG-neutral portfolio shows the smallest decrease. Finally, return distributions remain non-normal across subperiods, characterized by persistent excess kurtosis and statistically significant departures from normality, supporting the use of heteroskedasticity- and autocorrelation-robust inference in subsequent regressions.

The descriptive statistics for individual ETFs are broadly consistent with the portfolio-level results2. During the COVID-19 subsample, average returns are uniformly positive across all ETFs, indicating that this period coincides with a broad market rebound rather than category-specific gains.

4.2. Factor Model Results

Figure 1 presents the results of the Fama–French five-factor model augmented with the momentum factor for ESG-aligned, ESG-neutral, ESG-unaligned, and non-ESG ETFs across the pre-COVID (Panel A), COVID (Panel B), and post-COVID (Panel C) subperiods. The results indicate that the relationship between ESG orientation and risk-adjusted performance varies across market regimes.

For ESG-aligned ETFs, the factor model estimates provide limited evidence of abnormal performance during the pre-COVID and COVID subperiods. By contrast, in the post-COVID period, a considerable subset of ESG-oriented ETFs exhibits statistically significant negative coefficients, particularly funds with exposure to clean energy and renewable sectors. The magnitude and consistency of these estimates across alternative factor specifications indicate a deterioration in relative performance during the post-COVID macro-financial adjustment period.

A similar pattern emerges for ESG-neutral and ESG-unaligned ETFs, which also exhibit statistically significant negative coefficients in the post-COVID subperiod, while showing no robust evidence of abnormal performance in earlier regimes. These findings indicate that the post-pandemic underperformance is not confined to narrowly defined sustainability categories but extends across a broader range of ESG-oriented investment strategies. Although the CAPM, the Fama–French three-factor and five-factor models, and the Fama–French three-factor model expanded with momentum are not reported for reasons of brevity, their results remain qualitatively consistent with those presented earlier.

Conversely, non-ESG ETFs exhibit a distinctly different regime-dependent pattern. The estimates indicate statistically significant underperformance during the pre-COVID subperiod, particularly for funds with considerable exposure to traditional energy sectors. This pattern does not persist during the COVID and post-COVID periods, in which the estimated coefficients are generally small and statistically indistinguishable from zero across specifications.

Table 3. Factor Model t-Test and F-Test Results for Equally Weighted ETF Portfolios.

Panel A: Sustainability Globes Classification

CAPM

FF-3

Momentum

FF-5

FF-5 + Momentum

F-Stat.

F-Stat.

Portfolio

Period

α

t-Stat.

α

t-Stat.

α

t-Stat.

α

t-Stat.

α

t-Stat.

αB = αD = αA

αi = αj

ESG-Aligned

Pre-COVID

−0.0074

−0.764

−0.0037

−0.397

−0.0035

−0.372

−0.0036

−0.383

−0.0033

−0.347

0.362

COVID

0.0158

0.765

0.0115

0.575

0.0106

0.533

0.0112

0.561

0.0097

0.489

1.559

2.048

Post-COVID

−0.0258

−2.529

***

−0.0238

−2.358

**

−0.0232

−2.315

**

−0.0235

−2.329

**

−0.0226

−2.275

**

2.131

ESG-Neutral

Pre-COVID

−0.0078

−0.551

−0.0001

−0.005

0.0003

0.024

0.0004

0.030

0.0010

0.071

0.460

COVID

0.0261

0.859

0.0168

0.587

0.0150

0.523

0.0179

0.634

0.0149

0.526

3.283

**

2.291

**

Post-COVID

−0.0531

−3.047

***

−0.0484

−2.973

***

−0.0471

−2.938

***

−0.0472

−2.902

***

−0.0453

−2.857

***

3.525

*

ESG-Unaligned

Pre-COVID

0.0191

0.549

0.0275

0.797

0.0281

0.810

0.0328

0.992

0.0335

1.006

0.596

COVID

0.1005

0.881

0.0713

0.744

0.0688

0.715

0.1043

1.119

0.1008

1.077

3.334

**

5.228

**

Post-COVID

−0.1422

−2.787

***

−0.1103

−2.380

**

−0.1084

−2.348

**

−0.0964

−2.125

**

−0.0942

−2.090

**

3.749

**

Non-ESG

Pre-COVID

−0.1142

−2.775

***

−0.0542

−1.802

*

−0.0536

−1.775

*

−0.0509

−1.704

*

−0.0499

−1.651

*

0.745

COVID

0.0339

0.369

−0.0081

−0.139

−0.0107

−0.189

0.0067

0.117

0.0015

0.027

1.468

2.829

*

Post-COVID

0.0110

0.294

0.0041

0.151

0.0061

0.220

0.0125

0.480

0.0158

0.598

0.058

Panel B: ETF Mandate Classification

CAPM

FF-3

Momentum

FF-5

FF-5 + Momentum

F-Stat.

F-Stat.

Portfolio

Period

α

t-Stat.

α

t-Stat.

α

t-Stat.

α

t-Stat.

α

t-Stat.

αB = αD = αA

αi = αj

ESG-Screened

Pre-COVID

−0.0064

−1.246

−0.0077

−1.609

−0.0075

−1.536

−0.0080

−1.677

*

−0.0078

−1.587

0.018

COVID

−0.0057

−0.865

−0.0033

−0.621

−0.0043

−0.851

−0.0055

−1.054

−0.0068

−1.364

0.141

0.105

Post-COVID

−0.0077

−2.365

**

−0.0096

−3.240

***

−0.0089

−3.104

***

−0.0105

−3.516

***

−0.0096

−3.395

***

0.237

Thematic

Pre-COVID

−0.0022

−0.103

0.0081

0.381

0.0084

0.395

0.0102

0.491

0.0107

0.511

0.470

COVID

0.0603

0.961

0.0418

0.744

0.0402

0.712

0.0534

0.973

0.0507

0.919

3.338

**

5.318

**

Post-COVID

−0.0882

−2.973

***

−0.0734

−2.703

***

−0.0722

−2.676

***

−0.0679

−2.519

**

−0.0662

−2.485

**

3.651

*

Energy

Pre-COVID

−0.1594

−2.947

***

−0.0844

−2.100

**

−0.0837

−2.083

**

−0.0792

−2.002

**

−0.0778

−1.963

**

1.423

COVID

0.0596

0.493

0.0093

0.109

0.0060

0.071

0.0336

0.399

0.0266

0.331

2.191

3.963

**

Post-COVID

0.0233

0.450

0.0115

0.293

0.0141

0.353

0.0248

0.654

0.0293

0.758

0.001

Defense/Leisure

Pre-COVID

−0.0352

−1.228

−0.0014

−0.053

−0.0011

−0.042

−0.0015

−0.056

−0.0010

−0.040

0.616

COVID

−0.0111

−0.169

−0.0387

−0.827

−0.0399

−0.865

−0.0405

−0.855

−0.0426

−0.912

0.312

0.047

Post-COVID

−0.0105

−0.482

−0.0089

−0.501

−0.0079

−0.443

−0.0091

−0.522

−0.0078

−0.445

0.486

Source: Authors’ own. *** p < 0.01, ** p < 0.05, * p < 0.1. Where applicable, the last two columns present the F-statistic values and significance levels of α_B = α_D, α_B = α_A, and α_D = α_A, respectively.

reports the factor model estimates for the equally weighted portfolios constructed according to sustainability and mandate classifications and largely corroborates the ETF-level findings. In particular, the ESG-aligned, ESG-neutral, ESG-unaligned, ESG-screened, and thematic portfolios exhibit statistically insignificant coefficients during the pre-COVID and COVID subperiods but record significantly negative coefficients in the post-COVID period. By contrast, the non-ESG and energy portfolios underperform primarily in the pre-COVID subperiod, while the defense portfolio coefficients remain statistically insignificant throughout the three subperiods.

Furthermore, the Wald test results based on the Fama–French five-factor and momentum factor model indicate that the ESG-neutral and ESG-unaligned portfolios, as well as the thematic portfolio, exhibit significant regime shifts, specifically between the pre- and post-COVID subperiods and between the COVID and post-COVID subperiods, but not between the pre-COVID and COVID subsamples. However, although the joint F-test does not reject equality across all subperiods for ETFs classified either as non-ESG or energy portfolios, the pairwise tests suggest limited evidence of a shift between the pre-COVID and post-COVID subperiods at the 10% significance level. At the individual ETF level, a similar pattern emerges for ICLN (ESG-aligned/thematic), PBD (ESG-neutral/thematic), and TAN and PBW (ESG-unaligned/thematic). In contrast, the non-ESG/energy midstream energy MLP ETFs AMLP and MLPA display evidence of regime change between the pre-COVID and COVID subperiods3.

In brief, the factor model evidence indicates that ESG-related ETFs, particularly thematic sustainability portfolios with substantial exposure to clean energy and other long-duration sectors, as well as ESG-screened portfolios, exhibit adverse risk-adjusted performance during the post-COVID macro-financial adjustment period. By contrast, energy-related non-ESG ETFs underperform primarily during the pre-pandemic period, with no robust evidence of abnormal performance thereafter, while the defense portfolio does not exhibit statistically significant abnormal returns across subperiods. These findings remain consistent across alternative factor specifications, supporting the robustness of the observed regime-dependent effects.

4.3. Raw and Risk-Adjusted Excess Return Differentials

Panels A through C of Figure 2A report the Sharpe ratio-based excess return differentials of the sample ETFs relative to the S&P 500 during the pre-COVID, COVID, and post-COVID subperiods, respectively4. Although raw return differentials and Treynor-based risk-adjusted measures are omitted for brevity, the corresponding results are qualitatively similar. Overall, the findings confirm the patterns identified in the factor model analysis and provide additional insight into the benchmark-relative performance of ESG-related and sector-specific investment strategies.

ESG-aligned ETFs exhibit mixed benchmark-relative performance across subperiods. While several ESG-oriented funds, particularly broad U.S. large- and mid-cap ESG-screened equity ETFs, outperform the S&P 500 during the pre-COVID and COVID subperiods, this pattern does not persist in the post-COVID macro-financial adjustment period. By contrast, ETFs within the ESG-aligned category with substantial exposure to clean energy and renewable sectors display significantly negative excess return differentials during the post-COVID period, indicating a deterioration in benchmark-relative performance.

ESG-neutral and ESG-unaligned ETFs, which primarily consist of thematic sustainability funds (i.e., clean energy and renewable sectors), exhibit excess return differentials that are statistically indistinguishable from zero during the pre-COVID and COVID subperiods but become significantly negative for several ETFs in the post-COVID period across the performance measures considered. The consistency of these results across both raw and risk-adjusted metrics indicates that the observed underperformance is not driven solely by differences in systematic risk exposure or volatility.

In contrast, non-ESG ETFs exhibit significant underperformance relative to the S&P 500 during the pre-COVID subperiod, whereas their performance during the COVID and post-COVID periods is, in general, statistically indistinguishable from that of the index. These results reveal a reduction in relative underperformance rather than the emergence of systematic outperformance. This attenuation of underperformance in the later subperiod coincides with a macro-financial environment characterized by higher energy prices, Federal Reserve monetary tightening, and geopolitical tensions following the Russia–Ukraine conflict. It also corresponds to a period of increased political scrutiny of ESG investing in the United States, which may have contributed to changes in investor sentiment toward ESG-oriented assets.

Panels A, B, and C of Figure 2B report the Sharpe ratio-based excess return differentials of the individual ETFs relative to the MSCI World Index during the pre-COVID, COVID, and post-COVID subperiods, respectively. Overall, these results are qualitatively consistent with those reported in Figure 2A. During the pre-COVID and COVID subperiods, the returns of ESG-aligned, ESG-neutral, and ESG-unaligned ETFs, which primarily comprise ESG-screened and thematic portfolios, were generally statistically indistinguishable from the index returns. Nevertheless, several ETFs within these categories, particularly those classified as thematic sustainability funds, exhibited significantly negative Sharpe ratio differentials during the post-COVID period. Similarly, with a few exceptions, the returns of non-ESG ETFs, comprising energy- and defense-related portfolios, were statistically indistinguishable from the index during the COVID and post-COVID subperiods but displayed significantly negative Sharpe ratio differentials during the pre-COVID phase.

As Panel A of

Table 4. Raw and Risk-Adjusted Return Differentials for Equally Weighted ETF Portfolios Classified by Sustainability Category Relative to Alternative Benchmark Indices.

A. Raw and Risk-Adjusted Return Differentials Relative to the S&P 500 for Equally Weighted ETF Portfolios Classified by Sustainability Category.
		Diff. Raw Return			Diff. Sharpe			Diff. Treynor
Portfolio	Period	Coeff.	t-Stat.		Coeff.	t-Stat.		Coeff.	t-Stat.
ESG-Aligned	Pre-COVID	1.4087	0.686		0.0085	0.698		2.0094	0.722
	COVID	7.9819	1.369		0.0102	0.563		8.1945	1.365
	Post-COVID	−5.0978	−1.892	*	−0.0315	−2.804	***	−6.1190	−2.187	**
ESG-Neutral	Pre-COVID	1.2845	0.412		0.0042	0.237		1.4172	0.278
	COVID	9.4758	1.136		0.0131	0.515		11.8561	1.610
	Post-COVID	−12.1116	−2.631	***	−0.0614	−3.443	***	−13.5979	−3.015	***
ESG-Unaligned	Pre-COVID	7.0520	0.750		0.0127	0.368		4.4083	0.477
	COVID	47.1596	1.523		0.0150	0.332		31.4592	1.484
	Post-COVID	−35.5664	−2.593	***	−0.0933	−3.340	***	−27.1811	−2.875	***
Non-ESG	Pre-COVID	−25.3804	−2.502	***	−0.0717	−2.537	***	−18.8886	−2.169	**
	COVID	16.2555	0.701		−0.0194	−0.511		5.1083	0.265
	Post-COVID	4.6063	0.488		−0.0005	−0.015		6.1818	0.522
*** p < 0.01, ** p < 0.05, * p < 0.1
B. Raw and Risk-Adjusted Return Differentials Relative to the MSCI World Index for Equally Weighted ETF Portfolios Classified by Sustainability Category
		Diff. Raw Return			Diff. Sharpe			Diff. Treynor
Portfolio	Period	Coeff.	t-Stat.		Coeff.	t-Stat.		Coeff.	t-Stat.
ESG-Aligned	Pre-COVID	1.6970	0.970		0.0086	0.659		2.6690	0.984
	COVID	7.6702	1.365		−0.0120	−0.705		−3.1625	−0.556
	Post-COVID	−4.0752	−1.664	*	−0.0260	−2.408	**	−5.5074	−1.986	**
ESG-Neutral	Pre-COVID	1.5728	0.636		0.0044	0.256		2.0768	0.588
	COVID	9.1641	1.235		−0.0091	−0.396		0.4990	0.062
	Post-COVID	−11.0890	−2.801	***	−0.0560	−3.637	***	−12.9863	−3.261	***
ESG-Unaligned	Pre-COVID	7.3403	0.820		0.0129	0.352		5.0679	0.608
	COVID	46.8479	1.524		−0.0072	−0.160		20.1022	0.974
	Post-COVID	−34.5438	−2.580	***	−0.0878	−3.147	***	−26.5695	−2.905	***
Non-ESG	Pre-COVID	−25.0921	−2.589	***	−0.0715	−2.554	***	−18.2290	−2.519	***
	COVID	15.9439	0.709		−0.0416	−1.127		−6.2488	−0.338
	Post-COVID	5.6289	0.614		0.0050	0.150		6.7934	0.580

Source: Authors’ own. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows, the raw and risk-adjusted excess return differentials of the equally weighted ETF portfolios relative to the S&P 500 are broadly consistent with the patterns observed at the individual fund level. The ESG-aligned, ESG-neutral, and ESG-unaligned portfolios, which primarily capture ESG-screened and thematic sustainability exposures, exhibit excess return differentials that are statistically indistinguishable from those of the S&P 500 during the pre-COVID and COVID subperiods but record significantly negative raw and risk-adjusted excess return differentials in the post-COVID subperiod. By contrast, the non-ESG portfolio underperforms the benchmark in the pre-COVID subperiod, while its performance during the COVID and post-COVID periods is not statistically different from that of the index. The results in Panel B of Table 4, which present the raw and risk-adjusted excess return differentials of the equally weighted ETF portfolios relative to the MSCI World Index, are very similar to the outcomes in Panel A. These portfolio-level results are therefore consistent with the regime-sensitive performance dynamics documented in the individual ETF analysis.

Additional analyses based on the mandate-oriented portfolio classifications produce qualitatively consistent raw and risk-adjusted results under both the S&P 500 and MSCI World benchmarks. Specifically, thematic sustainability portfolios exhibit significantly weaker benchmark-relative performance during the post-COVID period, whereas energy-related portfolios underperform primarily during the pre-COVID subperiod. Moreover, defense-related portfolios generally do not display statistically significant abnormal performance across subperiods. However, the ESG-screened portfolio shows some evidence of positive benchmark-relative performance during the COVID episode relative to the S&P 500 and based on Sharpe- and Treynor-adjusted measures. This result suggests that broadly diversified ESG-screened equity strategies exhibited greater resilience during the pandemic shock. The evidence is less consistent when the MSCI World Index serves as the benchmark.

Table A1. Raw and Risk-Adjusted Return Differentials Relative to the S&P 500 for Equally Weighted ETF Portfolios Classified by Mandate Category.

		Diff. Raw Return			Diff. Sharpe			Diff. Treynor
Portfolio	Period	Coeff.	t-Stat.		Coeff.	t-Stat.		Coeff.	t-Stat.
ESG-Screened	Pre-COVID	1.6509	2.024	**	0.0133	1.904	*	2.1130	1.524
	COVID	2.0980	1.683	*	0.0095	2.272	**	4.2922	3.075	***
	Post-COVID	−0.5933	−0.877		−0.0038	−1.276		−0.6299	−0.901
Thematic	Pre-COVID	2.8825	0.537		0.0090	0.337		2.9322	0.441
	COVID	24.4514	1.425		0.0135	0.352		19.8964	1.297
	Post-COVID	−20.1971	−2.566	***	−0.0807	−3.425	***	−20.3118	−2.880	***
Energy	Pre-COVID	−36.7423	−2.725	***	−0.0913	−2.767	***	−25.7612	−2.247	**
	COVID	23.3729	0.773		−0.0265	−0.627		7.4917	0.309
	Post-COVID	7.7772	0.600		−0.0023	−0.060		9.5858	0.503
Defense/Leisure	Pre-COVID	−5.4972	−0.806		−0.0117	−0.455		−5.4245	−0.720
	COVID	3.8001	0.226		−0.0196	−0.624		−0.2624	−0.017
	Post-COVID	−0.9427	−0.163		−0.0098	−0.418		−0.0645	−0.010

Source: Authors’ own. *** p < 0.01, ** p < 0.05, * p < 0.1.

and

Table A2. Raw and Risk-Adjusted Return Differentials Relative to the MSCI World Index for Equally Weighted ETF Portfolios Classified by Mandate Category.

		Diff. Raw Return			Diff. Sharpe			Diff. Treynor
Portfolio	Period	Coeff.	t-Stat.		Coeff.	t-Stat.		Coeff.	t-Stat.
ESG-Screened	Pre-COVID	1.9391	1.221		0.0135	1.253		2.7726	0.945
	COVID	1.7864	0.699		−0.0127	−1.527		−7.0649	−1.979	**
	Post-COVID	0.4293	0.290		0.0017	0.243		−0.0183	−0.009
Thematic	Pre-COVID	3.1708	0.669		0.0091	0.340		3.5918	0.647
	COVID	24.1398	1.447		−0.0087	−0.238		8.5393	0.591
	Post-COVID	−19.1745	−2.595	***	−0.0752	−3.337	***	−19.7003	−3.014	***
Energy	Pre-COVID	−36.4540	−2.803	***	−0.0912	−2.691	***	−25.1016	−2.476	***
	COVID	23.0612	0.776		−0.0487	−1.156		−3.8653	−0.162
	Post-COVID	8.7998	0.692		0.0032	0.081		10.1974	0.540
Defense/Leisure	Pre-COVID	−5.2089	−0.785		−0.0115	−0.409		−4.7649	−0.729
	COVID	3.4885	0.222		−0.0418	−1.468		−11.6195	−0.876
	Post-COVID	0.0799	0.014		−0.0043	−0.179		0.5471	0.085

Source: Authors’ own. *** p < 0.01, ** p < 0.05.

in Appendix A report the detailed results of these analyses.

In short, the benchmark-relative performance analysis reinforces the regime-sensitive nature of ESG-related and sector-specific investment performance. The consistency of the evidence from factor-based models and excess return differentials suggests that sectoral exposures contribute to differences in relative performance across market environments, although ESG orientation and sector composition remain closely intertwined.

5. Discussion and Conclusions

The findings suggest that the performance of ESG and non-ESG exchange-traded funds varies across the pre-COVID, COVID, and post-COVID market regimes. Across factor-model specifications and benchmark-relative performance measures, ESG-related ETFs generally do not exhibit statistically significant abnormal performance during the pre-COVID and COVID subperiods but exhibit underperformance during the post-COVID period. By contrast, non-ESG ETFs underperformed primarily before the pandemic, while their performance is largely indistinguishable from that of the benchmark indices thereafter. Additional robustness analyses based on alternative ETF classifications and benchmark indices are consistent with these findings and suggest that performance differences are closely associated with the interaction between sectoral exposures and changing macro-financial conditions.

The regime-dependent results should be interpreted in light of the distinct macro-financial conditions prevailing across the sample. The pre-COVID period was characterized by generally favorable equity-market conditions despite episodes of trade-related uncertainty and monetary-policy adjustment (Board of Governors of the Federal Reserve System 2018, 2019, 2020). The COVID subperiod combined an unprecedented market shock with exceptional fiscal and monetary support, a rapid financial-market recovery, and persistently low interest rates (Board of Governors of the Federal Reserve System 2021). By contrast, the post-COVID period was characterized by rising inflation, monetary tightening, higher discount rates, elevated energy prices, and geopolitical tensions following Russia’s invasion of Ukraine (Bernanke and Blanchard 2023). The period also coincided with increased political and regulatory scrutiny of ESG investing in the United States, including changes in the interpretation of fiduciary responsibilities under ERISA (Department of Labor 2020) and broader debates regarding the role of ESG considerations in investment decision-making. These conditions help explain the weak performance of thematic sustainability portfolios during the post-COVID period, particularly those with substantial exposure to clean energy and renewable sectors, while energy-related non-ESG portfolios recovered from their earlier underperformance.

Viewed through the theoretical perspectives discussed in this study, the findings provide mixed support for competing explanations of ESG-related investment performance. Stakeholder, signaling, and agency theories suggest that ESG-oriented investments may benefit from stronger stakeholder relationships, enhanced transparency, and superior governance practices during periods of increased uncertainty. Consistent with these arguments, the mandate-based robustness analysis provides limited evidence that broadly diversified ESG-screened portfolios exhibited greater resilience during the COVID episode. However, the absence of widespread abnormal performance among ESG-related ETFs during the pre-COVID and COVID subperiods, together with the post-COVID underperformance observed across several sustainability-oriented portfolios, suggests that ESG orientation alone does not fully explain performance differences. Instead, the results suggest that ETF performance across market regimes is influenced by macro-financial conditions and sectoral exposures, particularly for portfolios concentrated in thematic sustainability sectors.

From a practical investment perspective, the findings suggest that investors may benefit from distinguishing among ESG-related ETFs rather than treating them as a homogeneous asset class. The results indicate that broad ESG-screened portfolios and thematic sustainability portfolios exhibit different performance patterns across market regimes. While broad ESG-screened portfolios display some evidence of resilience during the COVID episode, particularly under benchmark-relative measures based on the S&P 500 Index, thematic sustainability portfolios appear more sensitive to changes in macro-financial conditions, including rising interest rates and higher discount rates. More generally, the evidence suggests that portfolio composition and sectoral exposures are important factors shaping benchmark-relative performance, alongside ESG classification. Consequently, investors evaluating ESG-related investment strategies should consider the underlying economic exposures of ETF portfolios in addition to sustainability classifications and ratings.

From a policy perspective, the findings indicate that ESG labels alone do not necessarily identify investment strategies that exhibit superior financial performance. Rather, the results highlight the importance of underlying sectoral exposures and investment mandates. At the same time, the absence of systematic abnormal returns should not be interpreted as evidence against the broader relevance of sustainability considerations in financial markets. Instead, the findings suggest that the financial implications of ESG investing depend on market regimes, sectoral exposures, and investor objectives.

This study contributes to the literature on sustainable investing by examining the performance of ESG and non-ESG ETFs across distinct market regimes using factor-model and benchmark-relative performance analyses. By combining sustainability-based classifications with mandate-oriented robustness tests and alternative benchmark indices, the analysis provides additional evidence on the determinants of ETF performance under different economic and financial conditions.

Several limitations should be acknowledged. The analysis focuses on U.S.-listed ETFs and relies, in part, on end-of-sample sustainability classifications, although the robustness analyses indicate that the main conclusions do not depend on this choice. In addition, the empirical models do not directly incorporate macro-financial variables such as interest rates, oil prices, or market uncertainty. Consequently, the discussion of these factors is intended to provide economic context for the observed performance patterns rather than to establish direct causal relationships. Future research could extend the analysis to other markets, actively managed ETFs, alternative ESG classification methodologies, and additional crisis episodes.