ESG Ratings and Financial Performance: An Empirical Analysis

Abstract

In light of the growing interest in sustainable finance among investors and academics, in this study, we present an empirical analysis designed to understand whether sustainable investments outperform, underperform, or perform neutrally relative to conventional investments. The literature presents a spectrum of often-opposed conclusions, precluding the establishment of a definitive, consensus-driven judgment. Therefore, our analysis examines the behavior of sustainable investments within the Eurozone equity market from January 2019 to December 2023. Twenty portfolios are constructed to simulate sustainable investment strategies differentiated by environmental, social, and governance (ESG) strategy; stock inclusion/exclusion thresholds; and the type of ESG rating employed in the selection process. The analysis reveals that sustainable investments do not statistically significantly outperform or underperform traditional investments. This finding is significant for investors committed to ESG principles, as it suggests that they can align their investment choices with their ethical convictions without sacrificing performance.

1. Introduction

The emergence of sustainable investing, which incorporates environmental, social, and governance (ESG) considerations into asset allocation choices, has significantly changed the financial landscape. Unlike traditional investments, which prioritize measures of risk-adjusted performance, sustainable investments aim to balance financial performance with concerns about environmental stewardship, social responsibility, and sound governance practices. The performance of these investments has become a major concern for financial practitioners and scholars as investors shift their capital toward sustainable assets.

The sustainability of investments is commonly evaluated by their ESG ratings, which are assigned by specialized agencies. High ESG ratings are a sign of lower exposure to regulatory and reputational risks, which could improve portfolio resilience and yield superior returns. In contrast, traditional investments, which are not constrained by these criteria, may offer greater flexibility in portfolio construction but could be subject to unmitigated negative externalities. The relationship between ESG ratings and financial performance thus raises the question of whether sustainable investing can achieve parity or superiority over traditional approaches without sacrificing risk-adjusted performance. This study seeks to address this research question by comparing the financial performance of sustainable and traditional investments; we account for the role of ESG ratings as a determinant of the results. By analyzing performance metrics, we explore how ESG ratings affect returns, risk profiles, and portfolio efficiency, which should offer greater clarity on whether ESG-driven strategies represent a neutral alternative to traditional investing.

The remainder of this paper is organized as follows: Section 2 reviews the scientific literature on the financial performance of sustainable investments; Section 3 describes the data employed and outlines the methodology; Section 4 presents the empirical findings; and Section 5 discusses the implications for investors in a sustainable financial system.

2. Literature Review

The financial implications of ESG investment strategies have been examined in the scientific literature from diverse perspectives. This review primarily seeks to identify studies that investigate the nexus between ESG ratings and financial performance, thereby situating our analysis within the extant body of literature from both theoretical and empirical points of view.

2.1. Meta-Analyses

The extensive corpus of literature is marked by a lack of convergence of findings and has prompted scholars to conduct meta-analyses with the aim of synthesizing and comparing prior empirical findings to identify the prevailing position within the literature. However, these efforts reveal persistent points of contention, particularly regarding whether environmental, social, and governance (ESG) criteria confer a performance advantage, neutrality, or penalty relative to conventional investments.

Several meta-analyses agree on the performance neutrality of sustainable investments. Through independent meta-analytical investigations, Revelli and Viviani (2015), von Wallis and Klein (2015), Jedynak (2017), Kim (2019), AitElMekki (2020), Atz et al. (2023), and Hornuf and Yüksel (2024) concur by finding no statistically significant differences between the performance of sustainable and conventional investments. Their results suggest that investors may incorporate ESG criteria into their decision-making processes without incurring a systematic opportunity cost.

In contrast, other meta-analyses present different findings. Some, such as those by Friede et al. (2015) and Kumar (2023), suggest that sustainable investments outperform traditional investments, while others, like Akala et al. (2022), report evidence of a modest underperformance relative to traditional assets.

The article by Wang et al. (2025) synthesizes 46 papers from a pool of 387 identified studies from 2004 to 2024. The methodology combines a systematic literature review with meta-analysis, categorizing the impacts into seven areas (e.g., financial performance, sustainability, and firm risk). According to this analysis, ESG disclosure has a positive overall impact on financial performance, increasing profitability, reducing performance volatility, and lowering debt costs. It has the strongest effects on sustainability and investment efficiency, acting as a risk mitigation tool (e.g., reducing the stock price crash risk and default risk). However, the literature emphasizes firm-level economic benefits over broader social/environmental outcomes. The impacts vary by entity type, market environment, policies, investor preferences, disclosure voluntariness, and media oversight. Country-specific factors lead to regional differences, with calls for more research on ESG sub-dimensions (E, S, and G) and international standardization to address inconsistencies.

2.2. Theoretical Approaches

Beyond the aggregate findings of meta-analyses, several authors advance further considerations that shed light on the heterogeneity of the results. Revelli and Viviani (2015) contend that sector-level assessments cannot disregard analyses of non-financial performance, given the sustainability objectives typically pursued by ESG-oriented investors; the absence of this dimension would leave a critical gap in evaluation. Similarly, Atz et al. (2023) raise the question of why the positive effects of ESG factors on firms do not automatically translate into superior performance of ESG investment vehicles. They attribute this paradox to several factors, including the efficiency of markets in pricing ESG-related information, heterogeneity of sustainable investment approaches, diversity of methodologies for measuring risk and ancillary benefits, and constraints and costs inherent in portfolio construction. Kim (2019) further observes that evidence of ESG outperformance is more frequently reported in earlier studies, attributing this pattern to the progressive refinement of analytical techniques over time, and notes that ESG integration appears to yield stronger financial outcomes than either positive or negative screening approaches.

Finally, Akala et al. (2022) identify the lack of clear and consistent definitions, together with terminological overlaps, as additional sources of divergence in empirical results. They also call for the adoption of more advanced analytical methodologies, implying the necessity of moving beyond traditional approaches.

The study by Edmans (2023) theorizes that ESG investments are financially neutral in equilibrium, as they represent intangible assets yielding long-term financial and social returns comparable to conventional investments. Applying economic principles, any apparent alpha stems from mispricing or investor preferences, but efficient markets price ESG factors into expected cash flows without altering discount rates systematically. This perspective challenges views of the persistent outperformance of sustainable investments, revealing a gap in research on equilibrium dynamics amid shifting investor behaviors.

Hornuf and Yüksel (2024) emphasize that, consistent with Markowitz’s portfolio theory, the findings that point to underperformance in the ESG investment sector are the most theoretically consistent. This is because the shrinking of the investable universe, resulting from ESG screening strategies, reduces portfolio diversification, thereby generating an efficient frontier that lies below that of conventional investments, given the higher volatility and lower risk-adjusted returns. Consequently, analyses yielding different outcomes are likely influenced by the heightened demand for ESG products observed in recent years. This increasing interest of investors is unlikely to persist over the long term; as markets stabilize, the price effects are expected to dissipate, and the abnormal returns associated with sustainable investments may materialize only in the short term. Similarly, Derwall et al. (2011) observe significantly positive alphas for sustainable investments only over short horizons, attributing them to excess demand and the temporary mispricing of corporate social responsibility practices. If the analysis is extended to longer periods, these effects dissipate, leading to a performance that is indistinguishable from the market (i.e., non-significant alphas). In contrast, the sectors excluded from sustainable investment strategies, the so-called “sin” industries, are shown to sustain a consistent performance advantage relative to the market, even in the long term.

Behavioral finance provides a compelling explanation for the absence of persistent ESG outperformance, i.e., the presence of non-pecuniary preferences among investors. In a foundational equilibrium model, Pástor et al. (2021) demonstrate that when investors obtain direct utility from possessing sustainable assets, beyond their financial returns, these assets are traded at a premium, leading to diminished expected returns in equilibrium compared to otherwise comparable non-sustainable assets. Their framework accounts for taste-based demand for ESG attributes and illustrates that, while sustainable assets may excel after favorable shifts in investor preferences for ESG themes, this does not inherently produce enduring risk-adjusted alpha. If enough investors are willing to pay for the non-financial benefits of ESG exposure, the market prices will change, lowering the expected returns in the future. Thus, what may appear as modest or null outperformance does not indicate irrational pricing but rather reflects the internalization of non-pecuniary utilities into equilibrium asset values.

A more specific application of the behavioral finance approach is provided by Avramov et al. (2022). Building on the regret theory by Loomes and Sugden (1982), they posit that highly loss-averse investors exhibit a preference toward lower-risk strategies and are consequently inclined to avoid ESG funds, partly due to the perceived uncertainty and absence of standardized performance metrics in the ESG domain.

2.3. Empirical Studies on Risk and Performance

A broad consensus among scholars suggests that sustainable investments may experience a reduction in the risk diversification benefit (Barnett & Salomon, 2006; Renneboog et al., 2008a; Renneboog et al., 2008b; Climent & Soriano, 2011; Kuzmina et al., 2023). However, Hoepner (2010) highlights that two distinct effects of screening on diversification should be taken into account: a negative effect arising from the reduced pool of investable assets and a positive effect stemming from the lower average idiosyncratic risk associated with assets exhibiting high ESG ratings. Indeed, there should exist an inverse relationship between ESG ratings and firm-specific risk, as companies with higher ESG standards are better equipped to manage risks and are less exposed to ESG-related threats (e.g., reputational or regulatory threats). Consequently, the overall effect of ESG screening on diversification requires careful assessments to determine which of these two effects predominates in different scenarios, as there may be instances in which neither a reduction in diversification nor a negative effect on risk occurs.

This line of reasoning is echoed by Verheyden et al. (2016), who, by comparing conventional portfolios with ESG-screened portfolios, conclude that the degree of diversification is not systematically different across sectors. Where differences do emerge, any potential disadvantage appears to be adequately compensated through the selection of securities that exhibit superior volatility and return characteristics. As a result, sustainable strategies may prove particularly advantageous for investors due to their capacity to outperform traditional portfolios.

In accordance with this finding, Barnett and Salomon (2006) argue that, while a reduction in diversification is indeed present in the ESG sector, the degree of screening intensity is what ultimately drives the financial implications observed in the studies. Specifically, only funds employing an intermediate level of screening appear to be financially disadvantaged, whereas those with either a low or high screening intensity achieve favorable risk-adjusted performance. In the former case, minimal exclusion preserves diversification, while in the latter case, extensive exclusion tends to eliminate firms that frequently underperform, thereby offsetting the loss of diversification benefits.

Kempf and Osthoff (2007) compare different sustainable investment strategies applied to portfolios of U.S. companies and, using the Carhart four-factor model, highlight that best-in-class portfolios or those based on positive screening can generate significant excess returns relative to the market, with the former being particularly effective. This outperformance is especially pronounced when screening is based on social factors, such as community relations and employee engagement, or combined with other strategies. In contrast, negative screening appears to underperform. Long–short strategies based on ESG scores are also shown to be advantageous for investors. When used as the basis for stock selection, social factors are likewise identified by Barnett and Salomon (2006) as strong drivers of fund performance, whereas environmental factors appear to exert a negative effect relative to the market.

The phase of the economic cycle also plays a critical role in the research findings. Nofsinger and Varma (2014), employing factor models (capital asset pricing model, Fama–French three-factor model, and Carhart four-factor model), find that the relative performance of ESG versus conventional funds in the United States varies across market conditions. ESG investments outperform during crisis periods but exhibit weaker performance in normal economic phases. This hedge does not extend to ESG funds employing negative screening, but holds for other sustainable strategies that do not rely on negative screening, such as best-in-class, positive screening, and active ownership approaches. The influence of the business cycle is not limited to the U.S. market. Dopierała et al. (2020) corroborate these findings in their analysis of northern European funds, and primarily focused on climate-change mitigation and environmental themes. Their results indicate that alpha varies in magnitude across different phases of the economic cycle, depending particularly on the primary geographical investment focus of the funds.

Yue et al.’s (2020) analysis of European funds finds that sustainable funds exhibit lower return volatility and a reduced risk compared with their conventional counterparts. However, factor model assessments yield results that vary by the benchmark employed, ranging from evidence of superior performance to outcomes suggesting neutrality between sustainable and conventional investments. Notably, the analysis is limited to data available up to 2018, as the authors, echoing Hornuf and Yüksel (2024), argue that the surge in demand and interest in sustainable investments after 2018 is not a long-term constant and could distort analytical outcomes.

In contrast, Renneboog et al. (2008a, 2008b) report no significant differences in performance between ESG and conventional funds. In some European countries, ESG investments display mild signs of underperformance, but when factor model alphas are evaluated against local market indices, sustainable funds appear disadvantaged across all regions of the world. Here again, ESG strategies based on exclusions are the most penalized, though this may be considered an intentional compromise on the part of sustainable investors and accepted to align with their values. However, positive ESG screening appears to have a relatively positive effect on performance. Moreover, portfolio choices between sustainable and conventional funds are converging over time, suggesting a progressive alignment of investment practices across the two sectors.

Auer and Schuhmacher (2016), analyzing the Sharpe ratios of 600 portfolios, find that high-ESG-rated portfolios in the United States and Asia generally show no significant performance differences compared with either the benchmark or low-ESG portfolios. European investors still need to exercise greater caution because certain combinations of asset classes and sustainability screening can result in underperformance. Such effects are particularly evident when screening is centered on environmental or social factors, or within consumer and financial sectors. In any case, no evidence of excess returns is found.

Auer (2016) conducted further research based on European companies, constructing portfolios via negative screening with varying thresholds and score types (E, S, G, and composite): a passive replication of the traditional index can be outperformed by ESG portfolios with low exclusion thresholds, whereas those with very high thresholds tend to underperform. Among the ESG dimensions, governance contributes most to improving the Sharpe ratio, while environmental and social factors display neutral or inconsistent effects, depending on the threshold applied.

Building on this line of research, Bertelli and Torricelli (2024) conducted a similar analysis and highlight two key findings: exclusion-based ESG portfolios only outperform over long horizons and under stable economic conditions; during periods of stress, no significant performance differences are observed between sustainable and conventional investment sectors. Their results thus challenge the view of ESG investments as resilient to instability, a conclusion previously advanced by Nofsinger and Varma (2014) and Dopierała et al. (2020).

Abate et al. (2021) compare European funds with high ESG ratings, i.e., defined by Morningstar as those ranked within the top 10% of their category, and those with low ESG ratings, employing data envelopment analysis (DEA) to evaluate risk-adjusted performance. In their model, cumulative returns, ongoing expenses, and three measures of risk (standard deviation, downside deviation, and kurtosis) are used as inputs. The methodology then computes DEA scores, ranging from zero to one, for each fund by maximizing net returns subject to specified risk constraints, where a score of one indicates the best-performing fund in the sample and zero the worst-performing fund. High-ESG funds significantly outperform their low-ESG counterparts, achieving a substantially higher average DEA score.

Pokou et al. (2024) constructed three portfolios, each comprising stocks of companies with specific Morgan Stanley Capital International (MSCI) ESG ratings: Leader, Average, and Laggard. They find that the Average portfolio delivers the best risk-adjusted returns, followed by the Leader portfolio and, lastly, the Laggard portfolio. Only the Average portfolio exhibits a superior performance relative to the benchmark comprising all the stocks of the sample. These findings suggest that ESG screening can generate positive effects, but if applied too restrictively, it may disadvantage investors by causing them to forgo potential investment opportunities. Similarly, Ouchen (2022) applies Markov-switching generalized autoregressive conditional heteroskedasticity models to show that a passive replication of an ESG index proves more resilient during crises, outperforming and exhibiting a lower risk than a conventional portfolio replicating the S&P 500. These findings, derived from observing the behaviors of both portfolios, including during the COVID-19 period, are attributed to the differing weights of sectors (e.g., technology, finance, and energy) in the two indices.

Several researchers agree that sustainable investments offer enhanced protection against downside risks. This observation is documented, for instance, by Gupta and Chaudhary (2023) in a cross-country analysis comparing MSCI ESG indices with their conventional counterparts. Their results highlight not only the outperformance of ESG indices relative to the market but also a significantly higher Sharpe ratio for ESG indices in most of the countries studied. Similarly, Lööf et al. (2022) find that in both the normal and crisis market phases, securities from issuers with high ESG scores provide better loss protection than those with low scores, making them more appealing to risk-averse investors. This protective quality is attributed to the greater resilience of high-ESG firms to crises and negative news, which stems from stronger long-term risk management practices and the sustained trust of value-driven stakeholders. The benefit is subsequently transmitted to ESG funds, enabling them to outperform conventional funds during periods of instability (Nofsinger & Varma, 2014; Albuquerque et al., 2020; Ouchen, 2022). Such hedging characteristics are especially valued by investors, who, in the aftermath of major financial crises, appear to have become more conservative, therefore rebalancing their portfolios toward sustainable products.

Returning to the analyses of financial performance, it is worth noting that the neutrality, which is understood as the absence of statistically significant differences between ESG and conventional funds, is supported by studies such as Mill (2006), Humphrey and Tan (2014), and Kuzmina et al. (2023). Mill (2006) conducts his analysis on UK funds over the period of 1982–2004. Humphrey and Tan (2014) examine the years 1996–2010, focusing on ESG portfolios (using exclusionary and positive screening) and conventional portfolios of U.S. stocks drawn from the S&P 500 constructed to replicate as closely as possible the actual funds. Their results show that the alpha relative to the market is not statistically significant for any portfolio, while the Sharpe ratio indicates no performance differences either between the two sectors or between the two ESG strategies. Similarly, Kuzmina et al. (2023) investigates European energy funds from 2017 to 2022, finding that the four-factor model (both in its standard version and in an adjusted form incorporating a fifth ESG factor) produces no significant difference in the alphas of ESG and conventional funds.

Climent and Soriano (2011), in contrast, report an underperformance of sustainable investments. Climent and Soriano (2011) examine a sample of U.S. funds (ESG, environmentally focused, and conventional funds) and find that, prior to 2001, ESG and especially environmental funds significantly underperformed both conventional funds and the market, whereas no meaningful differences emerged between the three segments in the subsequent period. Hence, the choice of the time horizon proves to be a critical determinant of the results.

Numerous studies also report evidence of sustainable investments outperforming their conventional counterparts, such as Sherwood and Pollard (2018), Giese et al. (2019), and Brzeszczyński and McIntosh (2014). The first two contributions argue that a passively managed portfolio or fund would benefit from adopting an ESG index rather than a standard one as its benchmark. Sherwood and Pollard (2018) find that ESG indices in emerging markets outperform traditional indices, both in terms of higher returns and lower volatility, particularly when evaluated over the long term. Giese et al. (2019) report that, during the period of 2010–2017, MSCI ESG Leader indices achieved superior risk-adjusted performance compared with conventional indices across all regions of the world, with the exception of the United States. This finding is particularly significant, as many funds incorporate ESG criteria into their investment strategies by relying precisely on such sustainability-linked indices. Similarly, Brzeszczyński and McIntosh (2014) show that in the first decade of the 21st century, an equally weighted sustainable portfolio composed of U.K. stocks with high ESG ratings, specifically those included in the Global 100 ranking compiled by Corporate Knights, which identifies the 100 most sustainable companies worldwide, frequently outperformed the market, especially over longer horizons. These positive findings contrast the neutrality or underperformance reported in earlier studies, like Renneboog et al. (2008a, 2008b) and Climent and Soriano (2011), underscoring how methodological choices, such as an index-based versus fund-level analysis, may influence the outcomes.

The relationship between ESG investment performance and the recent financial and economic turmoil related to COVID-19 and geopolitical issues has been investigated in particular by Broadstock et al. (2021), Yahya (2023), and Binesh et al. (2025). These studies concur that sustainability provides resilience for firms, providing a hedge against downside risks. However, this resilience during crises contrasts with Bertelli and Torricelli’s (2024) findings of no significant differences in stressed periods, highlighting ongoing debates about the consistency of ESG hedging across diverse economic shocks.

A further and recent strand of the literature suggests that the relationship between ESG scores and financial performance may be non-linear, adding complexity to the analysis. Some analyses propose a U-shaped curve, where both low and high financial performers tend to have better ESG metrics, as highlighted by research by Nuber et al. (2020), Agarwala et al. (2024), and Lohmann et al. (2025). In contrast, other studies (Teng et al., 2022; El Khoury et al., 2023; Pu, 2023) suggest an inverted U-shape, implying that moderate ESG performance is linked to stronger financial performance. These differing perspectives highlight the need for a more nuanced understanding of the connection between ESG and financial outcomes.

Taken together, these results underscore the variability and subjectivity inherent in empirical choices across the literature, making a uniform and definitive assessment of the financial performance of ESG investments difficult to achieve. Key points of contention include the effect of screening intensity (e.g., low vs. high thresholds and positive vs. negative approaches), the roles of economic cycles (e.g., outperformance in crises vs. neutrality in stable periods), and the differential effects of individual ESG pillars. Research gaps persist in region-specific analyses, particularly for the Eurozone, where studies like those by Auer (2016) and Kuzmina et al. (2023) reveal inconsistencies across asset classes and pillars. Moreover, few studies disaggregate performance across multiple ESG categories (E, S, G, ESG, and ESGC) in a unified framework, leaving uncertainty about financial neutrality with conventional investments. It is therefore likely that such diversity and inconsistency will continue to characterize future research in this domain.

3. Empirical Analysis

3.1. Introduction

Despite the extensive body of literature on sustainable investing, a notable gap remains: limited attention to recent Eurozone data (after 2018), particularly amidst the growing demand for ESG investments. Additionally, as noted by Wang et al. (2025), few studies have simulated diverse portfolio strategies across multiple ESG score categories (E, S, G, ESG, and ESGC). The need to ensure comprehensive coverage of specific E, S, G and controversy scores for Eurozone companies has been a key factor in selecting the time window for this analysis, which spans from 1 January 2019, to 31 December 2023.

Consistent with prior empirical research by Auer (2016), this analysis aims to address this gap by evaluating the performance and risk characteristics of sustainable investments within the European financial market, with a particular focus on the Eurozone. Moreover, our analysis is grounded in the theoretical framework advanced by Pástor et al. (2021), which posits the absence of inherent outperformance in ESG assets. Building on this foundation, the present study undertakes an empirical assessment of whether sustainable investments systematically outperform, underperform, or remain performance-neutral relative to their conventional counterparts. Accordingly, we formulate the following hypothesis:

The financial performance and risk profiles of Eurozone investments across multiple ESG score categories (E, S, G, ESG, and ESGC) are not statistically different from those of conventional investments.

In the analysis, sustainable investments are simulated by constructing 20 portfolios, referred to as sustainable portfolios, which differ in the following aspects:

• Strategy employed—negative screening/exclusion or positive screening/best-in-universe;
• Threshold applied—in the case of positive screening, the top 10% or 20% of securities based on sustainability scores are included in the portfolios, whereas in negative screening, the bottom 10% or 20% of securities are excluded;
• Type of underlying sustainability score used for selection—five variants are considered, namely, E, S, G, ESG, or ESG combined (ESGC).

Subsequently, to evaluate the performance of each of the 20 portfolios relative to the benchmark, the following metrics and models are employed: the information ratio, Sharpe ratio, CAPM (Sharpe, 1964), Fama–French three-factor model (Fama & French, 1993), and Carhart four-factor model (Carhart, 1997). These measures and models have been selected coherently with an ample body of literature on both ESG (Kempf & Osthoff, 2007; Nofsinger & Varma, 2014; Auer & Schuhmacher, 2016; Dopierała et al., 2020) and traditional investments (Bacon, 2023).

3.2. Data Sample

The sample employed as input for the analysis comprises the stocks of companies that were included at least once in the Euro Stoxx index (symbol: SXXGT) during the period from 1 January 2019, to 31 December 2023, totaling 385 listed companies. This equity index, provided by Stoxx, was selected because its constituents consist exclusively of companies from the 11 largest Eurozone countries and are all denominated in EUR, thereby allowing the exclusion of the currency risk from the analysis, and, at the same time, it is the benchmark for Eurozone stock markets, representing the most relevant sample of companies for investors in this asset class.

For each stock in the sample, the following financial data were obtained from the LSEG (formerly known as Refinitiv) database: monthly total returns over the period from January 2019 to December 2023, market capitalization for each month between January 2019 and December 2023, the book-to-market ratio at the beginning of each of the five years from 2019 to 2023, and end-of-month prices from December 2017 to October 2023.

The time window required for the monthly price data differs from that of the other input data because, in order to calculate the momentum factor (MOM) used in the Carhart four-factor model, the requisite variable is the price change occurring between the twelfth month and the second month prior to $t$. Therefore, for the MOM in the first month of the analysis (January 2019), we must calculate the price change of each stock between the end of December 2017 and the end of November 2018. For the second month (February 2019), the price change between the end of January 2018 and the end of December 2018 is used. This procedure continues to the last month of the analysis (December 2023), for which the required price change to calculate the MOM spans from the end of November 2022 to the end of October 2023.

As a proxy for the risk-free rate, we employ the 1-month Euribor rate, sourced from the European Central Bank’s website.

The implementation of sustainable investment strategies for the construction of the 20 simulated portfolios also requires non-financial data. Accordingly, five different types of sustainability scores provided by LSEG were downloaded for all stocks in the sample:

• E score, representing the assessment assigned to the environmental pillar;
• S score, representing the assessment assigned to the social pillar;
• G score, representing the assessment assigned to the governance pillar;
• ESG score, representing the aggregated assessment of the three-dimensional pillars;
• ESG combined score, representing the combined assessment of the ESG score and the controversy score.

These scores are computed by LSEG using its proprietary methodology (LSEG, 2024). For the analysis, the scores assigned at the end of each year from 2018 to 2022 are required, as the scores from year $t-1$ are used to construct the sustainable portfolios at the beginning of each year $t$.

The choice of the LSEG scoring system stems from its methodology compared to those of the two main ESG rating providers, i.e., MSCI and Sustainalytics. While MSCI employes letters for its ratings (MSCI, 2024), LSEG has created a data-driven, percentile-based scoring system. Due to its inherently quantitative nature, the LSEG system is better suited for portfolio construction based on objective criteria. In contrast, Sustainalytics provides an absolute ESG risk assessment, measuring the extent to which a company’s enterprise value is exposed to unmanaged ESG factors (Sustainalytics, 2024). This approach, therefore, is focused on potential downsides, while the LSEG scores are designed to benchmark how a company stacks up against competitors in terms of ESG efforts. Furthermore, MSCI and Sustainalytics update their scores annually, whereas LSEG provides weekly updates. This higher frequency is more suitable for investors and analysts requiring timely data to respond to changes and aligns more closely with the monthly data frequency of our sample.

3.3. Methodology

The 20 sustainable portfolios are equally weighted, and their composition is updated at the beginning of each year based on the sustainability scores obtained at the end of the preceding year. Specifically, at the beginning of each year $t$, we considered only those securities for which LSEG was able to assign sustainability scores for year $t-1$ and that possess all monthly returns for year t. This ensures that the selected equities have been continuously listed throughout the period, as the sample includes companies that entered the stock market at different times relative to others.

Following this initial selection, four types of sustainable investment screening methods are applied to the remaining stocks for each of the five LSEG scores to identify the securities that will constitute each of the 20 portfolios for year $t$. The four sustainable investment strategies and thresholds considered are as follows:

• Best-in-universe or positive screening at 10%;
• Best-in-universe or positive screening at 20%;
• Negative screening or exclusion at 10%;
• Negative screening or exclusion at 20%.

In the first two cases, corresponding to positive screening, only the top x% of equities based on the relevant sustainability score are included in the portfolio. In the latter two cases, corresponding to negative screening, the bottom x% of equities based on the score are excluded, and only the remaining equities (1–x%) are considered for portfolio construction. Consequently, there is no fixed number of equities in the portfolios; the composition varies each year by how many securities satisfy the two initial conditions. A stock with a very high score in a given metric may be included in all four portfolios constructed according to that particular type of sustainability evaluation.

The portfolio employed as a benchmark is also equally weighted and rebalanced annually. At the beginning of each year $t$, only stocks that are listed for the entire year $t$ and that possess all monthly total returns for that period are included in this benchmark portfolio.

Table 1. Descriptive statistics and stationarity tests for the time series of monthly total returns of the sustainable portfolios and the benchmark (2019–2023).

Portfolio	Minimum %	Maximum %	Mean %	Standard Deviation %	Skewness	Kurtosis	PP-Test p-Value	ADF-Test p-Value
Benchmark	−18.4190	19.1388	1.0051	5.5502	−0.2970	5.6219	<0.01	0.0162
Best/Positive 10% score E	−23.9502	26.8031	1.2989	6.8685	−0.0511	7.1802	<0.01	0.0194
Best/Positive 20% score E	−23.8045	25.2589	1.1739	6.4897	−0.2031	7.7383	<0.01	0.0138
Negative 10% score E	−19.3187	20.1347	0.9965	5.6352	−0.2939	6.1993	<0.01	0.0188
Negative 20% score E	−20.1138	20.8593	0.9696	5.7774	−0.2786	6.4159	<0.01	0.0190
Best/Positive 10% score S	−20.0213	23.0231	1.0716	5.8962	−0.0038	6.9022	<0.01	0.0138
Best/Positive 20% score S	−19.4787	21.9443	0.9528	5.7458	−0.0616	6.6300	<0.01	<0.0100
Negative 10% score S	−19.1076	19.9616	0.9808	5.6067	−0.2698	6.0855	<0.01	0.0146
Negative 20% score S	−20.0580	20.8519	0.9625	5.7796	−0.2668	6.3858	<0.01	0.0168
Best/Positive 10% score G	−24.9859	23.7268	0.8702	6.4396	−0.4100	7.9354	<0.01	<0.0100
Best/Positive 20% score G	−22.1675	21.8946	0.9574	6.2131	−0.3221	6.5033	<0.01	<0.0100
Negative 10% score G	−19.2373	19.6832	0.9579	5.6595	−0.3014	5.8953	<0.01	0.0166
Negative 20% score G	−19.1470	19.7438	0.9814	5.6669	−0.2799	5.8513	<0.01	0.0172
Best/Positive 10% score ESG	−22.7717	24.7748	1.0153	6.6244	−0.0822	6.5887	<0.01	<0.0100
Best/Positive 20% score ESG	−22.8584	24.8002	1.0225	6.3771	−0.1268	7.4955	<0.01	<0.0100
Negative 10% score ESG	−19.3853	20.0702	0.9603	5.6345	−0.2874	6.2022	<0.01	0.0177
Negative 20% score ESG	−20.1996	20.9305	0.9910	5.7697	−0.2763	6.5345	<0.01	0.0177
Best/Positive 10% score ESGC	−22.4461	22.4712	0.8768	6.2287	−0.3037	6.7309	<0.01	0.0121
Best/Positive 20% score ESGC	−21.9973	23.0270	0.9322	6.1174	−0.2055	7.0713	<0.01	0.0114
Negative 10% score ESGC	−18.5292	19.8258	1.0125	5.5449	−0.2326	5.9653	<0.01	0.0164
Negative 20% score ESGC	−18.5299	19.8248	0.9806	5.5598	−0.2241	5.9176	<0.01	0.0179

All calculations are based on LSEG data. ADF, augmented Dickey–Fuller; PP, Phillips–Perron.

presents the descriptive statistics of the monthly total returns for the 20 sustainable portfolios and the benchmark over the years 2019–2023. In addition to the minimum, maximum, and first four moments of the return distribution for each portfolio, the table also reports the p-values resulting from the Phillips–Perron test and the augmented Dickey–Fuller test. Both tests are designed to verify the presence of a unit root in the time series used (with the null hypothesis being the existence of a unit root, which would imply non-stationarity), as a robust analysis is best conducted with stationary data series. Upon performing the tests and considering a pre-specified threshold of 5%, we observe that the null hypothesis is rejected (p-values are very low and below the threshold) for all the time series, thereby indicating that all series are stationary.

Following the calculation of the monthly returns of the portfolios, two risk-adjusted performance measures are computed for each portfolio over the entire period (2019–2023): the information and Sharpe ratios.

The information ratio ($IR$) is calculated as follows:

$$I{R}_i={\displaystyle \frac{{\mu }_i-{\mu }_{Benchmark}}{{\sigma }_{TE}}}$$

The subscript $i$ refers to the 20 sustainable portfolios, ${\mu }_i$ denotes the mean return of sustainable portfolio i, ${\mu }_{Benchmark}$ refers to the mean return of the benchmark portfolio, and ${\sigma }_{TE}$ is the standard deviation of the tracking error (i.e., the difference over period $t$ between the return of sustainable portfolio $i$ and that of the benchmark).

The Sharpe ratio ($SR$) is calculated as follows:

$$S{R}_i={\displaystyle \frac{{\mu }_i-r_f}{{\sigma }_i}}$$

The subscript $i$ refers to the 21 portfolios subject to our analysis, i.e., the 20 sustainable portfolios and the benchmark. ${\mu }_i$ and ${\sigma }_i$ denote, respectively, the mean and standard deviation of the monthly returns of portfolio $i$, and $r_f$ represents the mean risk-free rate. Accordingly, the numerator represents the average excess return of portfolio $i$.

After calculating the Sharpe ratio for each portfolio, the test developed by Ledoit and Wolf (2008) is employed to determine whether the difference between the Sharpe ratios of the 20 sustainable portfolios ($S{R}_i$) and that of the benchmark ($S{R}_{Benchmark}$) is statistically significant. The test is based on the following hypotheses:

H0. 

$S{R}_i$ − $S{R}_{Benchmark}$ = 0

H1. 

$S{R}_i$ − $S{R}_{Benchmark}$ ≠ 0

The null hypothesis tested is that the difference between the two Sharpe ratios is not statistically significant.

Subsequently, for each of the 20 sustainable portfolios, a t-test is conducted to assess whether a statistically significant difference exists between the mean return of sustainable portfolio $i$ and that of the benchmark. The hypotheses of the t-test are as follows:

H0. 

${\mu }_i$ − ${\mu }_{Benchmark}$ = 0

H1. 

${\mu }_i$ − ${\mu }_{Benchmark}$ ≠ 0

The subsequent models are based on the premise that only the non-diversifiable (systematic) risk is priced and rewarded by the market and that specific factors exist that contribute to understanding and explaining the formation of stock and portfolio returns. Any return exceeding these is defined as an abnormal return, which, according to theory, should not be significantly different from zero. The models applied are the CAPM, Fama–French three-factor model, and Carhart four-factor model. These models are employed to detect the presence or absence of such excess return relative to the model benchmark through linear regression.

The CAPM is based on the following univariate linear regression equation:

$$r_{i,t}-r_{f,t}={\alpha }_i+{\beta }_i(r_{MKT,t}-r_{f,t}),$$

where $r_{i,t}$ is the return of the $i$^th portfolio in month $t$, $r_{f,t}$ is the risk-free rate in month $t$, and $r_{MKT,t}$ is the market portfolio return in month $t$.

The Fama–French three-factor model (1993) requires the identification of two additional elements beyond the market factor ($MKT$), namely, the size factor, or small minus big (SMB), and the value factor, or high minus low (HML), which allow for a better explanation of stock and portfolio returns.

At the beginning of each year, six portfolios are constructed, which are rebalanced annually. Portfolios are value-weighted, i.e., the stocks within each portfolio are weighted by the market capitalization of their issuing companies. Each stock is allocated into the appropriate portfolio based on their market capitalization and book-to-market ratio. This allocation is carried out using a two-dimensional classification against predefined thresholds such that each of the six portfolios contains only stocks with specific characteristics. For market capitalization, the threshold is the 50th percentile: if a stock falls below this threshold, it is classified as small (referring to a smaller firm); otherwise, it is classified as large (a larger firm). For the book-to-market ratio, two thresholds are applied: the 30th and 70th percentiles. If the ratio is below the 30th percentile, the stock is classified as low (low value); if it falls between the 30th and 70th percentiles, it is classified as neutral; and if it is above the 70th percentile, it is classified as high (high value).

The intersection of these two classifications produces six portfolios into which stocks are sorted according to their characteristics. Once the monthly returns of all portfolios are calculated (by computing the value-weighted average of the returns of the constituent stocks), the following formulas are applied to compute the SMB and HML factors for each month:

$$SMB={\displaystyle \frac{r_{SH}+r_{SN}+r_{SL}}{3}}-{\displaystyle \frac{r_{BH}+r_{BN}+r_{BL}}{3}}$$

$$HML={\displaystyle \frac{r_{BH}+r_{SH}}{2}}-{\displaystyle \frac{r_{BL}+r_{SL}}{2}}$$

With these factors, it is possible to estimate the following multivariate linear regression equation:

$$r_{i,t}-r_{f,t}={\alpha }_i+{\beta }_{i,1}\left(r_{MKT,t}-r_{f,t}\right)+{\beta }_{i,2}{SMB}_t+{\beta }_{i,3}{HML}_t$$

In the four-factor Carhart model, an additional element is introduced, namely, the MOM, which represents the monthly return differential between winner and loser stocks. Its construction follows the methodology outlined by Carhart (1997).

At the beginning of each month $t$, the cumulative return of the previous eleven months, lagged by one period, is computed for all stocks in the sample. This corresponds to the price change from month $t-12$ to month $t-2$ (for instance, for January 2019, the price change from January 2018 to November 2018 is used). This metric serves as the second sorting variable for assigning assets to portfolios. As in the previous case of the Fama–French model, six portfolios are formed, weighted by market capitalization, but here they are recalculated and rebalanced at the beginning of every month since the MOM is measured on a monthly basis. These portfolios include only stocks that meet specific size and momentum characteristics, as determined at the beginning of month $t$.

The size threshold is the 50th percentile, while for the momentum, two cut-off levels are applied: the 30th and 70th percentiles. Stocks below the 30th percentile are defined as losers, those between the two thresholds are neutrals, and those above the 70th percentile are winners. Once the portfolios are constructed and stock weights are assigned proportionally to market capitalization, the monthly returns of the six portfolios are calculated. The MOM for month $t$ is then derived using the following formula:

$$MOM={\displaystyle \frac{r_{BW}+r_{SW}}{2}}-{\displaystyle \frac{r_{BL}+r_{SL}}{2}}$$

Once the momentum factor has been calculated for each month, the following multivariate linear regression equation is used to assess the level of alpha:

$$r_{i,t}-r_{f,t}={\alpha }_i+{\beta }_{i,1}\left(r_{MKT,t}-r_{f,t}\right)+{\beta }_{i,2}{SMB}_t+{\beta }_{i,3}{HML}_t+{\beta }_{i,4}{MOM}_t$$

These three econometric models allow us to determine whether an alpha that is significantly different from zero exists, which would imply either superior or inferior performance (depending on the sign of the intercept) of the sustainable portfolio over the period of 2019–2023:

H0. 

$\alpha$ = 0

H1. 

$\alpha$ ≠ 0

4. Results of the Analysis

The annualized information ratios of the 20 sustainable portfolios are reported in the first column of

Table 2. Risk-adjusted performance of the 20 sustainable portfolios and the benchmark.

	Annualized Information Ratios	Annualized Sharpe Ratios	Ledoit and Wolf Test (p-Values)	t-Test (p-Values)
Benchmark	-	0.60843	-	-
Best/Positive 10% score E	0.50699	0.63981	0.78454	0.79713
Best/Positive 20% score E	0.36523	0.61046	0.98414	0.87856
Negative 10% score E	−0.09413	0.59400	0.61193	0.99334
Negative 20% score E	−0.23625	0.56321	0.28944	0.97268
Best/Positive 10% score S	0.19177	0.61178	0.96794	0.94940
Best/Positive 20% score S	−0.18428	0.55617	0.49753	0.95962
Negative 10% score S	−0.32253	0.58729	0.37042	0.98101
Negative 20% score S	−0.31807	0.55871	0.17900	0.96722
Best/Positive 10% score G	−0.27834	0.45183	0.20484	0.90239
Best/Positive 20% score G	−0.13016	0.51694	0.30256	0.96473
Negative 10% score G	−0.72772	0.56781	0.04285 *	0.96332
Negative 20% score G	−0.26623	0.58142	0.30588	0.98159
Best/Positive 10% score ESG	0.01999	0.51513	0.37017	0.99269
Best/Positive 20% score ESG	0.03879	0.53897	0.49881	0.98733
Negative 10% score ESG	−0.48732	0.57175	0.21149	0.96504
Negative 20% score ESG	−0.09728	0.57680	0.43256	0.98913
Best/Positive 10% score ESGC	−0.33512	0.47081	0.13687	0.90540
Best/Positive 20% score ESGC	−0.23991	0.51075	0.18579	0.94563
Negative 10% score ESGC	0.09516	0.61360	0.79683	0.99422
Negative 20% score ESGC	−0.30527	0.59209	0.42189	0.98074

+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001. All calculations are based on LSEG data.

. This metric captures the excess return achieved by the sustainable portfolio relative to the market benchmark over the period, adjusted for the additional risk incurred due to deviations in investment choices from those of the benchmark. Accordingly, the information ratio indicates whether a sustainable portfolio was more or less efficient than the market over the analyzed period due to its active management.

Six sustainable portfolios outperformed the benchmark during the 2019–2023 period, as evidenced by a positive information ratio. The portfolio constructed with the top 10% of the E score (“best-in-universe”) is the top performer, with an information ratio of 0.50699. Among the six portfolios with a positive information ratio, five were built using a positive selection strategy specifically based on the E and ESG scores. This strategy does not consistently yield superior performance, as portfolios constructed using G and ESGC scores exhibit negative information ratios. Namely, even slight changes in portfolio construction criteria can lead to significantly different outcomes, complicating definitive assessments of the financial attractiveness of the sustainable portfolio sector.

Conversely, all portfolios employing a negative selection strategy, except the 10% ESGC portfolio, underperformed the benchmark on a risk-adjusted basis, as indicated by the negative information ratios. The portfolio constructed with a 10% negative screening strategy based on the G score performed particularly poorly, with a negative information ratio of –0.72772, making it the worst performer in the analyzed period. Notably, the G score appears suboptimal for portfolio construction from an investor’s perspective, as no strategy relying on governance scores generated a positive information ratio.

The results for the Sharpe ratio can be found in the second column of Table 2. Consistent with the findings for the information ratio, the sustainable portfolio with the highest annualized Sharpe ratio of 0.63981 is the one constructed using a positive selection strategy at the top 10% of the E score. Conversely, the lowest Sharpe ratio of 0.45183 is observed for the portfolio employing the same positive selection strategy and threshold but based on the G score. The benchmark, represented by the Euro Stoxx index, has an annualized Sharpe ratio of 0.60843, and only four of the 20 sustainable portfolios were able to exceed this value. Three of these four portfolios belong to the positive selection (“best-in-universe”) category, with two of them constructed using the environmental (E) score, consistent with the pattern observed for the information ratio. Among the remaining 16 sustainable portfolios with Sharpe ratios below that of the benchmark, the worst risk-adjusted performance is generally observed for portfolios based on positive selection using the G and ESGC scores.

The disparities between the Sharpe ratios of the sustainable portfolios and that of the benchmark are not statistically significant in most cases, as indicated by the p-values from the Ledoit and Wolf tests. These tests yield high p-values exceeding the pre-specified 5% threshold, meaning that the null hypothesis cannot be rejected. The only exception is the portfolio constructed using a negative selection strategy at the top 10% of the G score, which exhibits a statistically significant difference (p-value less than 5%) and a lower Sharpe ratio than the benchmark (negative difference). This portfolio, which was previously identified as the worst performer in terms of the information ratio, appears to be the only sustainable portfolio that investors should approach with caution and potentially avoid due to its significant inefficiency.

With the sole exception of the 10% negative selection G score portfolio, there are no significant differences in performance between sustainable and conventional investments, supporting a conclusion of financial neutrality between the two segments when considering the Sharpe ratio and the results of the Ledoit and Wolf tests.

Further evidence supporting this notion of neutrality comes from the t-tests of the differences between the mean returns of sustainable portfolios and the benchmark, which show no significant divergences in any case. All t-tests yield high p-values, indicating that the null hypothesis cannot be rejected and therefore that the mean returns of the sustainable portfolio and of the benchmark are not significantly different.

We now examine the results of three financial models that account solely for non-diversifiable risk. Starting with the CAPM results (

Table 3. Capital asset pricing model.

	α (%)	β1 (MKT)
Best/Positive 10% score E	0.09899	1.19979 ***
Best/Positive 20% score E	0.03063	1.14175 ***
Negative 10% score E	−0.02193	1.01374 ***
Negative 20% score E	−0.07185	1.03729 ***
Best/Positive 10% score S	0.02685	1.04065 ***
Best/Positive 20% score S	−0.07147	1.01964 ***
Negative 10% score S	−0.03315	1.009085 ***
Negative 20% score S	−0.08030	1.03867 ***
Best/Positive 10% score G	−0.2586	1.12688 ***
Best/Positive 20% score G	−0.1455	1.10033 ***
Negative 10% score G	−0.06559 *	1.018902 ***
Negative 20% score G	−0.04254	1.019342 ***
Best/Positive 10% score ESG	−0.1465	1.1608 ***
Best/Positive 20% score ESG	−0.1001	1.12054 ***
Negative 10% score ESG	−0.05812	1.01362 ***
Negative 20% score ESG	−0.04939	1.03618 ***
Best/Positive 10% score ESGC	−0.2270	1.10132 ***
Best/Positive 20% score ESGC	−0.1601	1.08944 ***
Negative 10% score ESGC	0.009515	0.99779 ***
Negative 20% score ESGC	−0.02496	1.000441 ***

+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001. All calculations are based on LSEG data.

), we observe that exposure to the market factor ($MKT$), represented by the coefficient ${\beta }_1$, is significant for all sustainable portfolios. In contrast, the intercept alpha is significant at the 5% level only for the portfolio employing the 10% negative screening strategy based on the G score, which underperformed relative to the benchmark. For the remaining 19 sustainable portfolios, no relevant difference in returns relative to the benchmark is observed. This finding reinforces the notion of financial neutrality between sustainable and conventional investments.

A similar pattern emerges from the application of the Fama–French three-factor model, with results displayed in

Table 4. Fama–French three-factor model.

	α (%)	β1 (MKT)	β2 (SMB)	β3 (HML)
Best/Positive 10% score E	0.20012 *	1.1228 ***	−0.41016 ***	0.26555 ***
Best/Positive 20% score E	0.11688	1.07672 ***	−0.31675 ***	0.22064 ***
Negative 10% score E	−0.006796	1.001967 ***	−0.07301 ***	0.041803 ***
Negative 20% score E	−0.05465	1.022746 ***	−0.14081 ***	0.057722 ***
Best/Positive 10% score S	0.03495	1.02712 ***	−0.39559 ***	0.08537 *
Best/Positive 20% score S	−0.06595	1.0089 ***	−0.34348 ***	0.07121 **
Negative 10% score S	−0.01889	0.99849 ***	−0.04466 ***	0.035120 ***
Negative 20% score S	−0.05206 +	1.018195 ***	−0.06371 **	0.065173 ***
Best/Positive 10% score G	−0.1784	1.06769 ***	−0.23115 *	0.19399 ***
Best/Positive 20% score G	−0.06893	1.04625 ***	−0.10167	0.16412 ***
Negative 10% score G	−0.06102 *	1.014988 ***	−0.03969 **	0.015745 **
Negative 20% score G	−0.03333	1.011779 ***	−0.06431 ***	0.028945 ***
Best/Positive 10% score ESG	−0.07402	1.10410 ***	−0.3673 ***	0.20334 ***
Best/Positive 20% score ESG	−0.01109	1.05392 ***	−0.30115 ***	0.22322 ***
Negative 10% score ESG	−0.04247	1.001719 ***	−0.06247 ***	0.040923 ***
Negative 20% score ESG	−0.02183	1.015489 ***	−0.09713 ***	0.069774 ***
Best/Positive 10% score ESGC	−0.1867	1.07311 ***	−0.03998	0.08407 *
Best/Positive 20% score ESGC	−0.14449	1.07431 ***	−0.22419 **	0.06934 **
Negative 10% score ESGC	0.02100	0.989162 ***	−0.04040 **	0.029068 ***
Negative 20% score ESGC	−0.01915	0.995781 ***	−0.03488	0.017266 *

+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001. All calculations are based on LSEG data.

. For 18 sustainable portfolios, the regression alpha does not show any statistically significant difference. However, for the remaining two portfolios, alpha is significantly different from zero but in opposite directions. The positive selection (“best-in-universe”) 10% portfolio based on the E score was the only portfolio to outperform the market, consistent with the Sharpe ratio results, where it also had the highest ratio among the 21 portfolios, with an intercept alpha of 0.20012%. The 10% negative selection portfolio based on the G score, as seen previously with the CAPM, was the only portfolio to significantly underperform the market, with an alpha of –0.06102%.

The regression results from the Carhart four-factor model are shown in

Table 5. Carhart four-factor model.

	α (%)	β1 (MKT)	β2 (SMB)	β3 (HML)	β4 (MOM)
Best/Positive 10% score E	0.19942 *	1.12314 ***	−0.41059 ***	0.26589 ***	0.0009187
Best/Positive 20% score E	0.1068	1.08235 ***	−0.32305 ***	0.22566 ***	0.01340
Negative 10% score E	−0.005081	1.001011 ***	−0.07194 ***	0.040950 ***	−0.002275
Negative 20% score E	−0.04429	1.01696 ***	−0.13435 ***	0.05257 ***	−0.01375
Best/Positive 10% score S	0.05518	1.01583 ***	−0.3830 ***	0.07531 +	−0.02684
Best/Positive 20% score S	−0.04296	0.99609 ***	−0.32915 ***	0.05978 **	−0.03051
Negative 10% score S	−0.01683	0.997341 ***	−0.04337 ***	0.03410 ***	−0.002731
Negative 20% score S	−0.04400	1.013693 ***	−0.05868 **	0.061160 ***	−0.010706
Best/Positive 10% score G	−0.2360	1.09983 ***	−0.2670 *	0.22263 ***	0.07642
Best/Positive 20% score G	−0.104	1.06585 ***	−0.12356	0.18158 ***	0.04660
Negative 10% score G	−0.05839 *	1.013525 ***	−0.03806 *	0.014440 *	−0.003481
Negative 20% score G	−0.03615	1.013353 ***	−0.06607 ***	0.030348 ***	0.003741
Best/Positive 10% score ESG	−0.08758	1.11167 ***	−0.37575 ***	0.21008 ***	0.01799
Best/Positive 20% score ESG	0.01054	1.04185 ***	−0.28766 ***	0.21246 ***	−0.02870
Negative 10% score ESG	−0.0416	1.00123 ***	−0.06192 ***	0.040490 ***	−0.001155
Negative 20% score ESG	−0.02090	1.01497 ***	−0.09654 ***	0.06931 ***	−0.00124
Best/Positive 10% score ESGC	−0.1591	1.05769 ***	−0.02276	0.07033 +	−0.03666
Best/Positive 20% score ESGC	−0.1122	1.05628 ***	−0.20406 **	0.05327 *	−0.04286
Negative 10% score ESGC	0.02627	0.986222 ***	−0.03711 **	0.026447 ***	−0.006992
Negative 20% score ESGC	−0.01315	0.992438 ***	−0.03114	0.014286 +	−0.007951

+ p < 0.1; * p < 0.05; ** p < 0.01; *** p < 0.001. All calculations are based on LSEG data.

. The conclusions are essentially identical to those of the previous model. Once again, the same two sustainable portfolios are the only portfolios with a significant alpha, while for the other 18 portfolios, the intercept is not significantly different from zero. The portfolio based on the environmental (E) score continues to exhibit a positive alpha (0.19942%), whereas the portfolio based on the governance (G) score continues to show a negative alpha (–0.05839%), indicating underperformance.

5. Conclusions

This analysis indicates no significant evidence of either outperformance or underperformance of the sustainable investment sector relative to conventional investments in the Eurozone between 2019 and 2023, with the notable exception of the portfolio constructed using a negative selection strategy based on the top 10% of the governance (G) score. Conversely, the portfolio employing a positive/best-in-universe selection strategy at the top 10% of the environmental (E) score emerged as the top performer.

Both the information and Sharpe ratios indicate that the majority of sustainable portfolios exhibit performance that is broadly consistent with the market benchmark. The CAPM, Fama–French three-factor, and Carhart four-factor regression models corroborate the same pattern: most sustainable portfolios have an alpha that is not significantly different from zero. This implies that any excess returns observed are largely attributable to systematic market factors rather than to the sustainability characteristics themselves. The few exceptions reinforce the earlier finding that underperformance, rather than outperformance, is more likely to be linked to specific selection criteria. Overall, these findings underscore a degree of financial neutrality between the two market segments in terms of financial efficiency, coherently with the theoretical approach by Pástor et al. (2021). Thus, our study contributes to the literature on sustainable investing by confirming the financial neutrality of ESG investments (Revelli & Viviani, 2015; von Wallis & Klein, 2015; Jedynak, 2017; Kim, 2019; AitElMekki, 2020; Atz et al., 2023; Edmans, 2023; Hornuf & Yüksel, 2024). Additionally, our study accounts for variations in ESG scores and selection strategies in Eurozone stock portfolios, providing an innovative and up-to-date contribution to the existing literature.

This study contributes to the ongoing debate on the relationship between ESG ratings and financial performance by providing recent empirical evidence from the Eurozone, a market characterized by increasing integration of sustainability criteria into investment practices. By demonstrating that adherence to ESG principles does not entail a systematic performance penalty, this study supports the view that investors can align their portfolios with sustainability objectives without compromising financial efficiency. This result is particularly relevant for asset managers, institutional investors, and policymakers seeking to promote sustainable finance as a viable alternative to traditional investment approaches.

This conclusion is important for ESG-oriented investors, as it suggests that these investors can pursue their sustainability goals, favoring companies implementing superior environmental and social practices, without the concern of forgoing potentially superior returns compared with more traditional investment choices. In this context, demonstrating the clear financial superiority of sustainable investments to reassure investors is not necessary; establishing the absence of consistent underperformance suffices, as this analysis confirms.

Drawing on insights from behavioral finance, in particular regret theory (Loomes & Sugden, 1982) and its application to ESG investments (Avramov et al., 2022), the perception or belief of missed opportunities often drives economic agents to favor traditional assets. Consequently, for profit-oriented investors, the choice between conventional and sustainable strategies becomes largely indifferent, while sustainable investments still allow capital to flow toward companies committed to ESG principles and projects supporting ecosystem preservation and social improvement. This enables the “do better while doing good” outcome, provided there is no definitive underperformance. The present analysis confirms this neutrality or indifference in financial attractiveness between the two sectors.

Note that the variation in results across different sustainability scores requires a careful portfolio construction process. Positive screening based on environmental and aggregated ESG scores tends to produce higher risk-adjusted performance, whereas negative screening or governance-focused selection may lead to underperformance. From a practical standpoint, the results highlight that positive screening strategies, particularly those centered on environmental and composite ESG scores, tend to yield better risk-adjusted performance than exclusionary approaches. This insight can guide portfolio managers in calibrating sustainability thresholds and optimizing portfolio construction methods to balance ethical commitments with financial prudence. Moreover, the neutrality of financial outcomes between sustainable and conventional investments can encourage broader capital allocation toward ESG-oriented assets, fostering corporate transparency, long-term risk management, and social responsibility.

Despite the robustness of the analytical framework and the consistency of the results obtained, certain limitations should be acknowledged. First, the scope of this analysis is confined to Eurozone equity markets over the period of 2019–2023, which may restrict the generalizability of the findings to other geographical contexts or longer time horizons. Second, this study relies exclusively on LSEG sustainability scores for portfolio construction, but the use of alternative ESG ratings might produce different outcomes. These considerations suggest the need for caution in extrapolating the conclusions and indicate the need for future research extending the temporal, geographical, and methodological dimensions of the present analysis.

Future research could extend this work by incorporating a broader temporal window and cross-regional comparisons to test the persistence of ESG–performance neutrality under different economic regimes. Comparative analyses across ESG data providers could also help disentangle the influence of rating methodologies on portfolio performance (Berg et al., 2022). Such extensions would improve the understanding of the mechanisms by which ESG integration contributes to long-term market stability and value creation.

Since this analysis establishes financial neutrality between the sustainable and traditional investment sectors, there is significant potential to advance the sustainable investment sector. However, the absence of a unified and shared perspective among scholars on this topic currently poses challenges to its full development.