ESG and Its Components: Impact on Stock Returns Across Firm Sizes in Europe and the United States

Abstract

A longstanding debate in finance concerns the impact of social responsibility actions on firms’ long-term profitability. This study provides a broad analysis on the relationship between ESG, its components, and stock returns. Using a dataset that spans from December 2014 to December 2023, this research analyzes an annual average of around 2260 publicly traded companies from Europe and the United States. The findings consistently show a negative link between ESG ratings, their components, and stock returns, a result that is possibly explainable by the mixed effect of a reduction of risk (lower risk premium) from social responsibility, and lower profitability from associated costs. The coefficients for ESG and its pillars in explaining stock returns are generally consistent, with a few exceptions for the environmental and governance components. The environmental pillar has a stronger influence in Europe, across firm sizes, while in the US, the effect is limited to larger companies. For governance, variations align with differing ownership structures across regions and changing investor priorities as firms grow, with stronger influence in Midcaps of both regions and in U.S. Large Caps. The effects of overall ESG scores and individual pillars on stock returns across regions, firm sizes, and their interaction, provide a more comprehensive perspective on their relationship.

1. Introduction

First discussed by Bowen (1953), the issue of Corporate Social Responsibility (CSR) has been extensively examined by academic researchers and finance practitioners alike (e.g., Baid and Jayaraman 2022; Berk et al. 2023; de Vincentiis 2023; Friede et al. 2015; Galema and Gerritsen 2025; Potharla et al. 2024; Riyadh et al. 2024). The central debate has focused on whether those firms that prioritize social responsibility— that is, those that deliberately allocate resources toward meeting the interests of stakeholders, employees, and the broader public—ultimately achieve superior long-term financial performance, or whether such ethical commitments potentially constrain profitability. Empirical evidence suggests that CSR considerations do influence investment decisions, and that this area of inquiry continues to attract sustained academic and practitioner interest. With the increasing availability of CSR ratings data, particularly those based on the Environmental, Social, and Governance (ESG) dimensions (United Nations 2004), investors have become more informed and attentive to corporate sustainability practices. While some studies suggest that firms embracing CSR are rewarded by investors and other stakeholders, others provide contradictory evidence, leaving the debate ongoing and unresolved. Furthermore, recent research concludes that ESG-related signals may be interpreted very differently across markets and contexts, sometimes producing opposite market reactions (de Vincentiis 2023), while CSR practices themselves can generate contrasting outcomes that complicate the link between sustainability and financial performance (Riyadh et al. 2024).

Socially Responsible Investing (SRI) investors typically rely on ESG scores as a central criterion for asset selection. This preferential allocation of capital may contribute to an improvement in the Corporate Financial Performance (CFP) of socially responsible firms, as they tend to attract greater investment flows from large financial institutions and asset managers employing SRI strategies. A United Nations report indicates that, as of June 2024, more than USD 128 trillion in assets were managed by organizations that align their investment approaches with CSR- and ESG-based standards (Principles for Responsible Investment Association 2024). The latest biennial Global Sustainable Investment Review 2024 finds that the value of fund assets reporting the use of responsible or sustainable investment approaches has reached USD 16.7 trillion, increasing by almost 49% over the last two years, and representing approximately 27% of the fund market of funds domiciliated in Europe, the United States, Japan, Canada, and Australasia (Global Sustainable Investment Alliance 2025). This increasing investor interest on ESG is consistent with recent evidence that sustainability-related rating changes and ESG momentum can have measurable effects on stock valuations across different markets (Berk et al. 2023; Escobar-Saldívar et al. 2025; Galema and Gerritsen 2025).

Corporate Social Responsibility (CSR) encompasses a range of internal and external factors that affect an organization’s activities and strategic decisions. It represents a critical source of information, particularly for long-term investors, regarding the risks faced by a firm and the way management responds to them (Dahlsrud 2008). Environmental, Social, and Governance (ESG) criteria, which Welford and Frost (2006) note are typically assessed by third parties, constitute a central and more clearly defined component of CSR, with greater conceptual consensus regarding their elements than for CSR (Widyawati 2020). ESG scores provide an external evaluation of the regulatory, legislative, operational, and reputational opportunities and risks that are relevant to investors. Boffo and Patalano (2020) assert that these scores carry both reputational and legal implications and may also serve as an indicator of a firm’s capacity to develop a sustainable competitive advantage. According to de Vincentiis (2023), ESG scores are interpreted differently in different regional contexts. Specifically, the author finds that ESG-related information does not generate uniform market reactions across major regions: European markets typically penalize negative ESG news, while in the U.S. market good news is more relevant than bad news and cause a negative price impact, and Asia-Pacific markets show no reaction at all.

Several empirical studies have reported that portfolios composed of assets with high ESG ratings often outperform their benchmarks across a range of market conditions (e.g., Friede et al. 2015; Nagy et al. 2016; Pisani and Russo 2021). In some instances, the performance of ESG-focused investments has been sufficiently strong to offset hypothetical transaction costs of up to 50 basis points per trade (Hoepner 2013). However, other research documents an inverse relationship between ESG ratings and returns, and still other studies report no statistically significant association at all (e.g., Girerd-Potin et al. 2014; Kumar 2023; Pedersen et al. 2021). In the absence of consistency in the results, it is possible to say that, so far, this relationship remains subject to ongoing empirical analyses and debate. Recent papers reinforce this mixed picture. While ESG momentum appears to generate positive abnormal returns in European markets (Berk et al. 2023). Galema and Gerritsen (2025) find that ESG downgrades are associated with negative abnormal returns in the United States, while Escobar-Saldívar et al. (2025) report an inverse relationship between ESG scores and stock returns in that market. These findings underline the context-dependent nature of ESG effects and their potential to vary substantially across regions. Additionally, beginning in 2017, the European Union’s Non-Financial Reporting Directive (NFRD) required large firms to disclose information on environmental, social, and employee-related matters (Carnini Pulino et al. 2022), whereas in the U.S. the advancement of ESG-related regulation intensified public debate and contributed to an evident anti-ESG backlash (Harper Ho 2024). Remarkably, the decline in excess demand for ESG assets appears to have preceded this political reaction and the subsequent introduction of state-level anti-ESG legislation (Curtis 2025).

This study contributes to the literature by examining the relationship between CSR, reflected by ESG scores, and CFP, measured by stock returns, offering a broader perspective on the subject. More precisely, this study differs from prior work by integrating several layers—regional context, individual ESG pillars, and firm size—within a unified econometric framework. Previous studies typically examine one of these dimensions in isolation, which may help explain their divergent findings. The mixed evidence documented across markets suggests that ESG performance is priced differently depending on institutional and cultural environments, as illustrated by recent work showing that both rating changes and ESG-related news elicit different market reactions across regions. Our framework allows the joint evaluation of these mechanisms.

The study includes a total of 22,627 observations over 10 years, corresponding to an annual average of approximately 2260 companies from the thirteen European countries (EC) and the U.S. The analyses are conducted for all firms together and separately for EC and U.S. companies, contributing to uncovering regional differences in the relationship between ESG performance and corporate financial performance. These two regions were selected because of their central role in the global economy and capital markets. The United States accounts for approximately 46% of global stock market capitalization and 24% of world GDP, while the thirteen European countries included in our study represent roughly 13% of global market capitalization and more than 17% of global GDP (International Monetary Fund 2024; World Bank 2025). This research also considers firm size to explore relevant variations in the relationship and to provide a clearer picture of how these factors interact. To examine the relationship between stock returns and ESG scores, we employ panel regression models. The Hausman (1978) test results indicated that fixed-effects models were suitable for analyzing this relationship in most cases, with only a few exceptions. This regional and firm-size perspective is consistent with emerging evidence that legal, regulatory, and cultural factors, particularly the growing divergence between U.S. and European ESG frameworks (Kuntz 2024), shape how markets evaluate sustainability information. Li et al. (2024) suggest that future research should examine both the independent and interactive effects of each ESG pillar on firm value—and corresponding stock returns—while paying particular attention to how these relationships vary across regions and industries.

Thus, the primary contribution of this study lies in clarifying whether ESG performance exerts a positive or negative influence on corporate financial performance through a comprehensive, multi-layered empirical design that simultaneously accounts for geographic context, firm size, and the individual ESG dimensions. By doing so, the study offers a more nuanced and integrated assessment of the ESG–CFP relationship than has been provided in prior research.

The remainder of the paper is organized as follows: Section 2 provides a brief review of the relevant literature, Section 3 outlines the data and explains the research methodology, Section 4 presents the empirical findings, and Section 5 discusses these results and concludes the paper.

2. Literature Review

2.1. Theoretical Basis

The relationship between Corporate Social Responsibility (CSR), Environmental, Social, and Governance (ESG) performance, and corporate financial performance (CFP) can be viewed from the standpoint of various theoretical perspectives. Stakeholder theory suggests that firms create value not only for shareholders but also for a broader set of stakeholders, including employees, customers, suppliers, and the community (Freeman 1984). From this perspective, ESG activities may enhance firm value by improving reputation, strengthening stakeholder relationships, and reducing conflicts and non-financial risks, which contribute to supporting long-term financial performance. Agency theory focuses on potential conflicts of interest between managers and shareholders (Jensen and Meckling 1976). Nevertheless, greater ESG disclosure tends to reduce information asymmetry by improving transparency and may contribute to lower agency costs, as strengthened governance mechanisms help mitigate conflicts of interest between principals and agents (Carnini Pulino et al. 2022).

Legitimacy theory provides an additional perspective by focusing on firms’ efforts to align their behavior with prevailing social norms and expectations to maintain societal approval (Suchman 1995). Because these norms and expectations differ across cultural and institutional contexts, the financial consequences of ESG performance are also likely to vary across regions. Finally, signaling theory suggests that firms use ESG disclosures and ratings to convey information about their quality, risk profile, and long-term orientation in settings characterized by information asymmetries (Spence 1973). In this context, investors may interpret ESG scores or rating changes as informative signals and adjust their valuation of firms accordingly, which can generate the heterogeneous market reactions documented in recent empirical studies.

Taken together, these theoretical perspectives indicate that the ESG–CFP relationship is unlikely to be uniform across firms, regions, or ESG dimensions. They provide a conceptual basis for the present study’s focus on regional context (United States vs. Europe), firm size, and the separate effects of the environmental, social, and governance pillars on stock returns.

2.2. The Environmental, Social, and Governance Pillars of Ethical Finance

While the acronyms CSR and ESG are often used interchangeably, in recent years there has been a noticeable shift in financial economics research toward placing more emphasis on environmental and climate-related issues, a trend reinforced by evolving disclosure standards and institutional ESG frameworks across countries (Singhania and Saini 2023). Chowdhury et al. (2021) find that foreign firms in the U.S., represented by ADRs, provide better CSR disclosure than U.S. firms, which results in lower volatility, better liquidity and higher institutional ownership. The Environmental (ENV) pillar often receives the most attention as climate change has become a primary concern for shareholders, institutional investors, and money managers (Breitz and Partapuoli 2020), predominantly in regions with firmer normative and regulatory ESG commitments such as Europe (Li et al. 2024). The environmental (ENV) pillar considers key factors that can significantly affect a firm’s long-term financial health and sustainability, including its dependence on fossil fuels; water and other resource management practices; pollution levels; exposure to climate change; hazardous waste generation and disposal; and overall carbon footprint. For example, Senadheera et al. (2021) find that a stronger focus on the environmental pillar can translate into competitive advantages, particularly through the development of eco-friendly products and services.

The social (SOC) pillar centers on how an organization manages its relationships with employees, suppliers, customers, and the community, reflecting its broader impact on the social systems in which it operates (Baid and Jayaraman 2022). Examples of issues associated with this pillar include, among others, customer satisfaction; data protection and privacy; gender and diversity; employee engagement; community relations; human rights; and labor standards (CFA Institute 2015).

The governance (GOV) pillar defines how decision-making authority is distributed among various stakeholder groups within a corporation (Gherghina 2024). This pillar reflects how a company manages leadership structures, executive compensation, internal control mechanisms, and the protection of shareholders’ rights. Issues associated with the GOV pillar include board composition, audit committee structure, bribery and corruption, executive pay, and lobbying activities (CFA Institute 2015). Accordingly, unlike the environmental and social pillars of ESG, the governance pillar primarily concerns the firm’s internal organizational structure and control processes, rather than its external environmental or societal impact (Breitz and Partapuoli 2020).

2.3. The Relationship Between ESG and Financial Performance

The literature presents arguments both for and against the favorable impact of ESG observance on a firm’s stock returns, which are commonly used as a proxy for financial performance, and empirical studies continue to report mixed results. Li et al. (2024) systematic review validates that findings in this area are heterogeneous, varying notably across countries and industries. A key argument supporting a negative relationship between stock returns and ESG activities is that investments in ESG-related initiatives may reduce short-term earnings and increase operating costs (Jyoti and Khanna 2021). By contrast, studies reporting a positive relationship argue that ESG engagement may enhance long-term profitability through innovation, stronger customer loyalty, higher employee productivity, improved governance, regulatory compliance, and more robust stakeholder relationships (Carnini Pulino et al. 2022; Xie et al. 2017). Additionally, some studies conclude that the impact of ESG activities on stock returns is largely neutral, suggesting that the associated costs and benefits tend to offset one another over time (Billio et al. 2021; Fiskerstrand et al. 2020; Naffa and Fain 2022).

The panel data study by Ye et al. (2022), involving publicly listed companies across multiple EU member states, found that firms with higher levels of ESG disclosure in their annual reports were associated with superior stock return performance. Likewise, Giese et al. (2019) conclude that firms’ ESG information influences both valuation and financial performance through its effects on systematic and idiosyncratic risk profiles. This is reflected in lower costs of capital, higher firm valuations, increased profitability, and reduced exposure to tail risks.

In contrast, Hsu et al. (2023) found that, even after controlling for various risk factors, firms with higher toxic emissions tend to generate a higher annual return of approximately 4.42%. Menicucci and Paolucci (2023) examined how do environmental performance, social responsibility and corporate governance actions influence bank performance on 105 Italian banks and conclude that ESG policies negatively affect operational and market performance. Feng et al. (2022) argue that, in the long run, ESG factors exert a negative impact on the stock returns of most Chinese-listed companies included in their study. Villarreal-Samaniego et al. (2022) suggest that, while large automotive firms experienced improved stock return performance associated with higher ESG scores during the COVID-19 pandemic, sustainability-oriented initiatives during this period were generally related to lower stock returns across the broader industry.

Some studies suggest there is no functional relationship between ESG and stock returns. La Torre et al. (2020), for example, grouped companies included in the Eurostoxx50 index over the 2010–2018 period according to their ESG commitment, measured by combining different ESG indicators (quantitative ratings, scorings and qualitative opinions). Their results indicated that performance does not seem to be affected by the companies’ ESG efforts. (Kaur and Singh 2021) found no connection between stock returns and CSR among Indian steel industry companies, proposing that more active CSR involvement may lead to higher returns, but only over the long term. Amanova et al. (2025) studied whether global official institutions (the World Bank, IMF, OECD, and European Commission) defined Sustainable Development Goals (SDGs) benefit the banking sector profitability. With an unbalanced panel of 143 countries over 2000–2024 they confirmed that the SDG Index Score has a weak and inconsistent effect on profitability. Also, Billio et al. (2021) reported no significant difference in the stock price behavior between firms with ESG investments and their non-ESG counterparts. Naffa and Fain (2022) concluded that ESG portfolios did not produce significant portfolio alphas during the 2015–2019 period, supporting the argument in the literature that ESG-related investments have a neutral effect on returns.

Kumar (2023) reviewed more than 30 empirical studies that applied Fama–French-type factor models to assess the attribution of portfolio alpha to ESG factors. The author found that most studies reported a positive alpha–ESG relationship, followed closely by those indicating a neutral association, with only a few concluding a negative relationship between these variables. Similarly, based on their analysis of more than 2200 studies examining the link between corporate financial performance and ESG, Friede et al. (2015) report that approximately 90% find a non-negative relationship, with the majority showing statistically positive outcomes. They therefore conclude that the business case for ESG investing is strongly supported by the available empirical evidence.

Finally, it is relevant to mention that recent developments in the examination of CSR-ESG and financial performance include sector-specific analyses (Affes and Jarboui 2023; Alamoudi 2025; Utami and Damayanti 2024), research focused on periods defined by financial crisis or other extraordinary events, such as the COVID-19 pandemic (Rubbaniy et al. 2022; Villarreal-Samaniego et al. 2022; Zhang et al. 2023), and studies examining different geographic regions (Lisin et al. 2022; Meles et al. 2023). The present study is framed within this latter strand of the literature, as the central objective is to test whether ESG actions have a positive or a negative influence on corporate financial performance, measured by the stock returns of the U.S. and EC sample firms, using subsamples built according to size (small, medium, large), as well as by each one of the ESG pillars.

2.4. Corporate Financial Performance and the ESG Pillars

There is a growing number of studies that investigate the connection between each of the ESG components and CFP, a direction explicitly advocated in recent papers suggesting independent analysis of each of the ESG pillars (Li et al. 2024). For instance, in terms of the environmental aspect, the study by de Haan et al. (2012) suggests a negative relationship between Corporate Environmental Performance (CEP) and stock returns, which they partly attribute to risks commonly associated with CEP. However, the authors also acknowledge that investor preferences may play a role in this outcome. (Prashar 2023) meta-analysis results on 60 studies on Sustainability Reporting (SR) indicated that, for large, mature firms, those with institutional investors as board members, or active participants in SR quality awards, SR is reflected in firms’ market, accounting and operational-based measures performance. The findings of Ma et al. (2024) reveal that companies with stronger environmental performance generate significantly higher excess returns in the Chinese stock market, implying the existence of “green returns”.

Potharla et al. (2024) find a positive association between social sustainability and the alignment of stock prices with overall market movements and conclude that investors place a high value on ethical management practices, commitments to human rights, and levels of employee satisfaction. By contrast, Liagkouras et al. (2022) suggest that, in the context of the UK equity market, investors who prioritize environmental and social considerations in their investment decisions may face lower expected returns and different risk-adjusted performance profiles.

Affes and Jarboui (2023) report that the implementation of effective corporate governance practices in UK firms has a positive impact on their CFP, as measured by return on equity. Pérez Ortega and Briano Turrent (2024) found that certain governance elements, such as the separation of the Chair of the Board and Chief Executive Officer positions, as well as the inclusion of women on boards, positively influence profitability, liquidity, and stock prices in industrial companies listed in the Mexican stock market. By contrast, Guney et al. (2020) find a strong negative relationship between company performance and the quality of corporate governance. Menicucci and Paolucci (2023) examined the influence of ten dimensions of ESG constructs over the performance indicators of a sample of Italian banks and found that they negatively affect operational and market performance. In the context of the Johannesburg Stock Exchange, (Chawarura et al. 2025) report that while total ESG performance was positively associated with firm financial performance, the effects of each ESG dimension were mixed and moderated by firm size.

In summary, ESG and its pillars have become a topic of increasing interest in finance for both practitioners and academics. Despite significant empirical research efforts to determine what is the nature of the relationship between the CSR-related variables, as measured by ESG scores, and financial performance, no consensus has been reached so far. Therefore, considering previous studies, we proposed the following hypotheses for our research:

Hypothesis 1.

ESG scores are negatively related to firm stock returns.

Hypothesis 2.

The ESG–CFP relationship differs between the United States and Europe.

Hypothesis 3.

The ESG–CFP relationship varies across firm size categories.

Hypothesis 4.

Each ESG pillar affects returns differently across regions and firm sizes.

3. Research Method

To examine whether ESG and its components are equally important across the U.S. and Europe, and among Large, Mid and Small companies, the statistical analysis employs a panel regression model to the broad sample and its subsamples. Clearly, the main variables of interest are the ESG score and its individual components (ENV, SOC, and GOV). We used Python version 2.12 for all the data gathering, data preparation and econometric analysis.

The data used in this study was obtained by LSEG’s Eikon platform via its Python API, including the ESG scores which are now an evolution of the original Asset4 ESG ratings (Refinitiv 2023). Asset4’s scores have a range of 0 to 100 and are frequently represented with letter-ratings from D− to A+ based on percentile ranges of equal size. The model represented by Equation (1) is used to carry out all the analyses. VarInt represents either ESG, ENV, SOC or GOV, since the variables of interest were included only one at a time given the high correlations among them. All the stocks in the study have ESG scores as well as annual data for all the variables listed in

Table 1. Variables used in the multifactor model of stock returns.

Type	Variable	Description
Dependent variable	ri,t	1-year total return of stock i at time t
Control variables	Sectori,t	Global Industry Classification Standard (GICS) sector of stock i at time t
	Betai,t	Beta of stock i to its corresponding market (U.S. or EU) at time t
	Sizei,t	Natural logarithm of market capitalization of stock i at time t
	Valuei,t	Book-to-Price ratio of stock i at time t
	Profi,t	Profitability ratio of stock i at time t (Operating Income/Equity)
	Invi,t	Asset growth of stock i at time t (Assetsi,t/Assetsi,t−1 − 1)
	Momi,t	Momentum of stock i at time t (1-month lagged, 6-month price change)
Variables of interest	ESGi,t	ESG score of stock i on time t
	Envi,t	Environmental pillar score of stock i on time t
	Soci,t	Social pillar score of stock i on time t
	Govi,t	Governance pillar score of stock i on time t
Error term	εi,t	Error term, i.i.d.

Source: Authors’ own.

, covering the entire period under examination. The control variables listed in Table 1 were selected based on the common multifactor models used to explain stock returns (Fama and French 1993, 2015).

$$\begin{gathered} {\mathit{r}}_{\mathit{i},\mathit{t}}={\mathsf{\beta }}_0{\mathit{S}\mathit{e}\mathit{c}\mathit{t}\mathit{o}\mathit{r}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_1{\mathit{B}\mathit{e}\mathit{t}\mathit{a}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_2\mathit{S}\mathit{i}\mathit{z}{\mathit{e}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_3\mathit{V}\mathit{a}\mathit{l}\mathit{u}{\mathit{e}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_4\mathit{P}\mathit{r}\mathit{o}{\mathit{f}}_{\mathit{i},\mathit{t}} \\ +{\mathsf{\beta }}_5\mathit{I}\mathit{n}{\mathit{v}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_6\mathit{M}\mathit{o}{\mathit{m}}_{\mathit{i},\mathit{t}}+{\mathsf{\beta }}_7{\mathit{V}\mathit{a}\mathit{r}\mathit{I}\mathit{n}\mathit{t}}_{\mathit{i},\mathit{t}}+{\mathit{\epsilon }}_{\mathit{i},\mathit{t}} \end{gathered}$$

(1)

The dataset includes 22,627 entity–time observations spanning a ten-year period from December 2014 to December 2023. These observations cover two regions, with 13,624 corresponding to the United States and 9003 to European countries. The firms in the sample can also be grouped by size, with 6865 observations classified as Large-Caps (market capitalization above $10 billion USD), 8644 observations as Mid-Caps (market capitalization between $2 and $10 billion USD) and 7118 as Small-Caps (market capitalization between $250 million and $2 billion USD).

Table 2. Firms per year observations for the six Region-Size samples from 2014–2023.

Region-Size		2014	2015	2016	2017	2018	2019	2020	2021	2022	2023	Total
EC	Large	188	186	178	231	195	235	262	289	225	254	2243
	Mid	257	271	287	286	280	292	404	474	447	453	3451
	Small	147	146	156	148	202	173	410	476	710	741	3309
US	Large	346	329	366	427	387	468	534	609	550	606	4622
	Mid	313	310	305	335	399	599	674	742	740	776	5193
	Small	39	66	52	84	180	374	653	694	845	822	3809
	Total	1290	1308	1344	1511	1643	2141	2937	3284	3517	3652	22,627

Source: Authors’ own.

presents the observations of firms by Region-Size combination through the 10 years analyzed and

Table 3. Distribution of stocks per GICS sector in the total sample for 2014–2023.1

Sector	2014	2015	2016	2017	2018	2019	2020	2021	2022	2023
Comm. Services	71	74	74	83	89	112	148	164	182	190
Cons. Discret.	181	186	188	213	247	312	414	446	459	476
Cons. Staples	88	90	91	99	113	131	169	180	182	199
Energy	75	75	78	89	90	107	124	149	171	174
Financials	151	153	155	167	171	217	309	346	361	379
Health Care	99	99	106	135	141	215	362	426	463	473
Industrials	248	252	261	288	321	417	582	648	711	734
Info. Tech.	108	109	114	126	150	210	306	354	390	411
Materials	120	119	124	130	136	158	194	210	225	230
Real Estate	82	85	86	109	112	170	220	248	260	266
Utilities	67	66	67	72	73	92	109	113	113	120
Total	1290	1308	1344	1511	1643	2141	2937	3284	3517	3652

Source: Authors’ own.

shows the distribution of stocks per GICS sector in the combined U.S. and EC sample. The countries included in the European region subsample are, with total observations in parentheses: Belgium (287), Denmark (256), Finland (299), France (1060), Germany (1133), Italy (463), the Netherlands (334), Norway (330), Poland (189), Spain (407), Sweden (822), Switzerland (759), and the UK (2664). There are other countries in the European region, but we included the ones that had at least 15 observations per year. Austria and Greece meet this requirement only since 2020 and other countries like Portugal or Ireland did not meet this constraint.

To avoid survivorship bias, we must include data for firms that existed for only part of the study period. This inherently results in an unbalanced panel. Creating a balanced panel by only including stocks that existed for the entire period would intentionally re-introduce survivorship bias. The primary risk of using an unbalanced panel comes from the reason that causes the data to be missing. Mainly if this is a non-random mechanism there could be some type of sample selection bias. In our case, we are using a very large dataset that only filters stocks based on their country of exchange and minimum market capitalization (above micro caps). Given that the number of stocks increased naturally over time due to more stocks being listed than the ones being delisted, standard panel data regression models like FE and RE are robust and can be applied to our data.

As a first step, the analysis tested the pair-wise correlations among the variables to identify any potential multicollinearity issues. The correlations are shown in

Table 4. Pair-wise correlations among the variables in our regression models.

	mktret	beta	size	value	inv	prof	mom	esg	env	soc	gov
mktret	1
beta	0.02	1
size	0.05	−0.04	1
value	−0.03	−0.01	−0.13	1
inv	−0.01	0.03	0.02	−0.02	1
prof	0.01	−0.01	0.03	−0.01	0.00	1
mom	0.10	0.06	0.11	−0.08	−0.01	0.01	1
esg	−0.07	−0.09	0.45	−0.01	−0.07	0.00	−0.04	1
env	−0.10	−0.09	0.44	0.02	−0.09	0.00	−0.05	0.86	1
soc	−0.07	−0.05	0.43	−0.02	−0.06	0.00	−0.03	0.89	0.74	1
gov	−0.02	−0.10	0.24	0.00	−0.05	−0.01	−0.02	0.67	0.38	0.38	1

Correlations > 0.20 are shown in bold. Source: Authors’ own.

below, with magnitudes larger than 0.20 in bold. Correlations between ESG components are high, especially between the Environmental (ENV) and Social (SOC) components (0.74), whereas the correlation with the Governance (GOV) component is also considerable (0.38). In view of these results, the study examines each of the three pillars independently in the regressions rather than simultaneously. Another relevant correlation between variables is that of Size and the ESG components (0.45), an expected result since larger companies have greater slack to invest in ESG initiatives. Nevertheless, the literature identifies Size as an important control variable (Fama and French 1993, 2015). Consequently, we include it in the model, as its correlation with the ESG variables is below 0.5 and variance inflation factor tests yielded values lower than 4.

4. Results

The analysis divides the sample by region and size categories to compare their differences, resulting in 12 different sub-samples: one full sample that includes both regions (U.S. and EC) and all firm sizes (Large-, Mid-, and Small-Cap), two sub-samples based on region, three sub-samples based on firm size, and six sub-samples based on pairwise combinations of region and firm size.

We performed VIF tests for all subsamples to test for multicollinearity. We found, as expected given the pair-wise correlations, that the largest VIF occurs with the Size factor in all cases. However, we consider that the VIF is acceptable in all the AllCap, SmallCap and MidCap models (Max VIF < 3.5). In the LargeCap models we do find larger VIF values with relation to the Size variable (Max VIF > 5). However, given the importance of the Size factor in the literature we decided to keep it and maintain the same variables in all cases, for comparison.

Table 5. Residuals diagnostics tests from Pooled OLS regressions.

Model	Jarque–Bera	White	Breusch-Pagan	Durbin-Watson
US&EC_AllCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
EC_AllCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US_AllCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US&EC_SmallCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US&EC_MidCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US&EC_LargeCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
EC_SmallCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US_SmallCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
EC_MidCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US_MidCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
EC_LargeCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation
US_LargeCap	Non-Normal	Heteroskedasticity	Heteroskedasticity	No Serial Correlation

Identical for ESG, ENV, SOC and GOV models. Source: Authors’ own.

shows a summary of the residual diagnostics tests that were also performed for all models.

In view of the above econometric considerations, the methodological approach leaned toward Panel regression models, with both Fixed Effects (FE) and Random Effects (RE). Each of these models has its advantages and limitations due to their underlying assumptions. The FE model removes the effect of time-invariant characteristics of the stocks and other unobserved factors, which addresses this type of potential endogeneity and allows the evaluation of the net effect of the exogenous variables on the dependent variable. Alternatively, the RE model incorporates time-invariant variables (e.g., Sector), by modelling constant terms as randomly distributed across cross-sectional units. However, this holds true only if the individual effects are strictly uncorrelated with the regressors. A disadvantage of the RE model is the difficulty in specifying all the individual characteristics that may influence the dependent variable, leading to a potential omitted variable bias that could cause endogeneity. To determine whether the FE or RE models are more appropriate, we conducted Hausman tests for each case. The null hypothesis of the Hausman test asserts that unique errors are uncorrelated with the regressors, and if it holds true then the RE model is consistent and efficient. However, if the null hypothesis is rejected, the FE model is preferred as its coefficients are consistent.

Table 6. Panel model selection based on Hausman Test results (p-values).

Model	ESG	ENV	SOC	GOV
US&EC_AllCap	0.0000	0.0000	0.0000	0.0000
EC_AllCap	0.0000	0.0000	0.0000	0.0000
US_AllCap	0.0000	0.0000	0.3705	0.0000
US&EC_SmallCap	0.0000	0.0000	0.0000	0.0000
US&EC_MidCap	0.0000	0.0000	0.0000	0.0000
US&EC_LargeCap	0.1724	0.7439	0.2418	0.1375
EC_SmallCap	0.0000	0.0000	0.0000	0.0000
US_SmallCap	0.0000	0.0000	0.0000	0.0000
EC_MidCap	0.0000	0.0000	0.0000	0.0000
US_MidCap	0.0000	0.0000	0.0000	0.0000
EC_LargeCap	0.0000	0.0000	0.0000	0.0000
US_LargeCap	0.9913	0.9956	0.9979	0.9844

Of the total 48 models, the 9 RE models are shown in bold, the rest are FE. Source: Authors’ own.

shows that the Hausman test rejected the null hypothesis in 39 cases, leading to the selection of the FE model in such instances, while selecting the RE model in 9 of the cases.

The results of the 48 models regarding the magnitude and significance of the variables of interest are summarized below in Figure 1, Figure 2, Figure 3 and Figure 4. The detailed coefficients and significance of all variables are included in

Table A1. ESG Variable.

Size	All Market Caps			Large-Caps
Region	All	EU	US	All	EU	US
beta	0.6340 **	0.3232	0.7058 **	0.9797	1.1592 **	0.9379
size	0.4150 ***	0.2541 ***	0.4769 ***	0.1459 ***	0.1610 ***	0.1411 ***
value	0.0040	−0.1237 ***	0.0099	−0.4134 ***	−0.2102 **	−0.5489 ***
investment	−0.0577 **	−0.0214	−0.0852 ***	−0.0359 ***	−0.0459 **	−0.0266 **
profitability	−0.0005	0.0018	−0.0006	−0.0001	−0.0818 ***	−0.0001
momentum	0.9210 ***	1.0238 ***	0.8802 ***	1.1505 ***	1.0652 ***	1.1848 ***
esg	−0.1878 ***	−0.2216 ***	−0.1485 **	−0.2446 ***	−0.2550 ***	−0.2013 ***
const	−0.0591 *	0.1054 ***	−0.1636 ***	−0.0055	−0.0276	−0.0446
obs	22,627	9003	13,624	6865	2243	4622
Size	Mid-Caps			Small-Caps
Region	All	EU	US	All	EU	US
beta	0.6618 ***	0.7826 **	0.7345 ***	−0.1079	−0.1902	−0.0307
size	0.3476 ***	0.3238 ***	0.3457 ***	0.6700 ***	0.4683 ***	0.7627 ***
value	−0.3732 ***	−0.2841 ***	−0.4380 ***	0.0083	−0.0892	0.0109
investment	−0.0212 *	−0.0103	−0.0388 **	−0.0975 ***	−0.1157 ***	−0.0898 ***
profitability	−0.0005	0.0058 *	−0.0014	0.0000	0.0007	0.0016
momentum	0.9007 ***	1.0065 ***	0.8646 ***	0.6884 ***	0.9361 ***	0.5825 ***
esg	−0.1762 ***	−0.2333 ***	−0.1263 **	−0.3186 ***	−0.3338 ***	−0.2407 *
const	−0.0307	0.0461	−0.0843 ***	0.6012 ***	0.5339 ***	0.5956 ***
obs	8644	3451	5193	7118	3309	3809

Coefficients significant at ≤0.01, ≤0.05 and ≤0.10 are represented by ***, ** and * respectively. Source: Authors’ own.

,

Table A2. Environmental Variable (ENV).

Size	All Market Caps			Large-Caps
Region	All	EU	US	All	EU	US
beta	0.6390 **	0.3133	0.7091 **	0.9812	1.1870 **	0.9345
size	0.4143 ***	0.2543 ***	0.4761 ***	0.1363 ***	0.1560 ***	0.1357 ***
value	0.0041	−0.1233 ***	0.0099	−0.4213 ***	−0.2164 ***	−0.5465 ***
investment	−0.0576 ***	−0.0217	−0.0850 ***	−0.0349 ***	−0.0446 **	−0.0261 **
profitability	−0.0005	0.0017	−0.0006	0.0000	−0.0786 ***	0.0000
momentum	0.9215 ***	1.0235 ***	0.8807 ***	1.1537 ***	1.0674 ***	1.1875 ***
env	−0.0683 **	−0.1467 ***	−0.0291	−0.1543 ***	−0.1348 **	−0.1335 ***
const	−0.1249 ***	0.0580 **	−0.2209 ***	−0.0548	−0.1072 *	−0.0844
obs	22,627	9003	13,624	6865	2243	4622
Size	Mid-Caps			Small-Caps
Region	All	EU	US	All	EU	US
beta	0.6683 ***	0.7681 **	0.7372 ***	−0.1087	−0.2176	−0.0362
size	0.3445 ***	0.3222 ***	0.3438 ***	0.6677 ***	0.4671 ***	0.7601 ***
value	−0.3734 ***	−0.2846 ***	−0.4380 ***	0.0085	−0.0863	0.0109
investment	−0.0212 *	−0.0107	−0.0385 **	−0.0979 ***	−0.1157 ***	−0.0902 ***
profitability	−0.0005	0.0056	−0.0013	0.0000	0.0006	0.0016
momentum	0.9021 ***	1.0081 ***	0.8655 ***	0.6881 ***	0.9335 ***	0.5828 ***
env	−0.0543 *	−0.1379 **	−0.0174	−0.1387 **	−0.2518 ***	−0.0151
const	−0.0937 ***	−0.0110	−0.1320 ***	0.5085 ***	0.4786 ***	0.5133 ***
obs	8644	3451	5193	7118	3309	3809

Coefficients significant at ≤0.01, ≤0.05 and ≤0.10 are represented by ***, ** and * respectively. Source: Authors’ own.

,

Table A3. Social Variable (SOC).

Size	All Market Caps			Large-Caps
Region	All	EU	US	All	EU	US
beta	0.6362 **	0.3234	0.3019	0.9942	1.1484 **	0.9521
size	0.4154 ***	0.2544 ***	0.1291 ***	0.1420 ***	0.1637 ***	0.1370 ***
value	0.0039	−0.1242 ***	−0.0127	−0.4171 ***	−0.2074 **	−0.5502 ***
investment	−0.0576 ***	−0.0214	−0.0584 ***	−0.0352 ***	−0.0447 **	−0.0262 **
profitability	−0.0005	0.0020	−0.0001	0.0000	−0.0820 ***	0.0000
momentum	0.9210 ***	1.0247 ***	0.8868 ***	1.1523 ***	1.0671 ***	1.1863 ***
soc	−0.1543 ***	−0.1459 ***	−0.2890 ***	−0.2045 ***	−0.2281 ***	−0.1639 ***
const	−0.0719 **	0.0667 **	0.1425 ***	−0.0165	−0.0401	−0.0538
obs	22,627	9003	13,624	6865	2243	4622
Size	Mid-Caps			Small-Caps
Region	All	EU	US	All	EU	US
beta	0.6646 ***	0.7827 **	0.7364 ***	−0.1102	−0.1945	−0.0276
size	0.3472 ***	0.3228 ***	0.3451 ***	0.6709 ***	0.4678 ***	0.7633 ***
value	−0.3724 ***	−0.2820 ***	−0.4391 ***	0.0082	−0.0913	0.0110
investment	−0.0211 *	−0.0105	−0.0384 **	−0.0974 ***	−0.1138 ***	−0.0897 ***
profitability	−0.0004	0.0062 *	−0.0013	0.0002	0.0009	0.0016
momentum	0.9011 ***	1.0075 ***	0.8650 ***	0.6888 ***	0.9379 ***	0.5827 ***
soc	−0.1239 ***	−0.1218**	−0.0927 *	−0.2584 ***	−0.1804 *	−0.2776 **
const	−0.0529 **	−0.0137	−0.0963 ***	0.5835 ***	0.4657 ***	0.6145 ***
obs	8644	3451	5193	7118	3309	3809

Coefficients significant at ≤0.01, ≤0.05 and ≤0.10 are represented by ***, ** and * respectively. Source: Authors’ own.

and

Table A4. Governance Variable (GOV).

Size	All Market Caps			Large-Caps
Region	All	EU	US	All	EU	US
beta	0.6333 **	0.3276	0.7025 **	0.9901	1.1938 **	0.9372
size	0.4125 ***	0.2510 ***	0.4757 ***	0.1177 ***	0.1479 ***	0.1172 **
value	0.0038	−0.1258 ***	0.0098	−0.4459 ***	−0.2230 ***	−0.5765 ***
investment	−0.0573 **	−0.0210	−0.0849 ***	−0.0329 ***	−0.0428 **	−0.0243 *
profitability	−0.0005	0.0019	−0.0006	−0.0001	−0.0782 ***	−0.0001
momentum	0.9221 ***	1.0261 ***	0.8806 ***	1.1597 ***	1.0680 ***	1.1934 ***
gov	−0.0720 **	−0.0477 *	−0.0751 *	−0.0884 ***	−0.0213	−0.0838 **
const	−0.1130 ***	0.0062	−0.1904 ***	−0.0597	−0.1780 ***	−0.0736
obs	22,627	9003	13,624	6865	2243	4622
Size	Mid-Caps			Small-Caps
Region	All	EU	US	All	EU	US
beta	0.6589 ***	0.7862 **	0.7277 ***	−0.1124	−0.1633	−0.0377
size	0.3446 ***	0.3172 ***	0.3457 ***	0.6664 ***	0.4631 ***	0.7610 ***
value	−0.3693 ***	−0.2826 ***	−0.4355 ***	0.0081	−0.0905	0.0108
investment	−0.0208 *	−0.0100	−0.0384 **	−0.0979 ***	−0.1154 ***	−0.0900 ***
profitability	−0.0005	0.0059 *	−0.0014	0.0001	0.0008	0.0016
momentum	0.9017 ***	1.0091 ***	0.8645 ***	0.6897 ***	0.9402 ***	0.5827 ***
gov	−0.0895 ***	−0.0866 **	−0.0879 **	−0.0956 *	−0.0777	−0.1215
const	−0.0667 ***	−0.0363	−0.0921 ***	0.5157 ***	0.4090 ***	0.5687 ***
obs	8644	3451	5193	7118	3309	3809

Coefficients significant at ≤0.01, ≤0.05 and ≤0.10 are represented by ***, ** and * respectively. Source: Authors’ own.

from Appendix A.

The results in Figure 1 reveal that ESG is a relevant variable in all financial models, regardless of region and company size. In the U.S. Small-Cap sample, it shows slightly weaker significance but preserves a negative sign, consistent with the other samples (i.e., the higher the ESG score the lower the return on a company’s stock). The literature suggests that following ESG policies decreases returns because it implies added costs for companies. However, an advantage of ESG is risk reduction and higher-quality investments which, in turn, lead to lower required returns and higher stock valuations. Furthermore, since there is stronger public coverage of firms as they increase in market capitalization, it makes sense for Small-Cap stock returns to be less influenced by ESG ratings.

Figure 2 reveals that the Environmental pillar (ENV) is also relevant by itself in the combined European and U.S. sample. However, when examining the regions separately, ENV appears to be considerably more relevant in European countries across different firm sizes than in the US, where it is only relevant in Large-Caps. This regional difference is not surprising, given that European countries have introduced more rigorous policies for cleaner energy sources, either out of conviction or necessity, while the U.S. seems to have a more divided stance (Li et al. 2025). European firms are subject to a more requiring regulatory framework on ESG compared to the U.S. companies (Fox et al. 2024). Moreover, as of 2021, Europe was the largest climate-fund market, as implied by Ma et al. (2024). However, ENV is still significant in the U.S. Large-Caps sample, which is logical given that many of those companies are global firms and must adhere to international regulations. Additionally, these companies are more visible and closely monitored by the public and media, making it advantageous for them to be socially responsible in the three ESG pillars.

As shown in Figure 3, the Social pillar (SOC) is relevant across all regions and firm size combinations. In Europe, it is only slightly weaker in Small-Caps, which aligns with the public coverage argument mentioned earlier. In the case of the US, the SOC variable is slightly weaker in Mid-Caps but remains significant at the 10% level, while its negative sign and magnitude are consistent with its coefficient in other samples. Although Table 4 shows a high correlation between the ENV and SOC pillars (0.74) and a strong correlation between both pillars and the overall ESG scores (0.86 and 0.89 respectively), the SOC pillar results resemble the ESG results more closely, while the ENV results do show regional differences in Mid- and Small-Cap firms.

Figure 4 suggests that the Governance pillar (GOV) is not significant in Small-Caps stock returns, except at the 10% level in the combined sample. This finding may be a consequence of the increased sample size. Eun and Resnick (2018) assert that with a large managerial ownership, the interests of managers and outside investors become better aligned, thus reducing agency costs. We suggest that many Small-Caps still have founders as managers, making them less susceptible to agency problems compared to their larger counterparts.

In contrast, GOV is significant in both regions for the combined and Mid-Cap samples. The significance of Governance for Mid-Cap returns in both Europe, and the U.S. may stem from these firms being in a growth stage, which increases uncertainties about their future. Therefore, effective governance policies and practices may be particularly important for maintaining investor confidence in these companies. Eun and Resnick (2018) also signpost an interim range of managerial ownership share over which there is an “entrenchment effect”. If managerial ownership is still large but below a certain threshold, it might be able to extract larger private benefits at the expense of outside investors, while still being able to resist takeover bids.

Interestingly, GOV is significant in the case of U.S. Large-Caps but non-significant in the case of European Large-Caps. This result aligns with the arguments proposed by Eiteman et al. (2013) who assert that in the case of US, management typically serves as a hired agent of widely held firms, owning only a small proportion of stock in their firms. In contrast, many firms in other global markets are characterized by controlling shareholders such as family-controlled firms (e.g., France and Italy), institutions like banks (e.g., Germany), governments or other consortiums of interests. These controlling owners tend to prioritize the long term, and being closer to management, they face fewer conflicts and agency problems. In this context, Kirchmaier and Grant (2005) found that governance contributes to performance differences between European and U.S. firms. Hence, having good governance practices might be more relevant in the U.S. to keep management in check than in Europe. Additionally, Li et al. (2025) find that there during this period there has been a stronger diversity, equity and inclusion (DEI) push in the USA than in Europe, which is typically captured in Governance measures. The sign of the GOV variable remains negative in all cases, however, consistent with other ESG variables.

5. Discussion and Conclusions

This study examines the relationship between overall ESG performance and its three individual pillars and the stock returns of large-, medium-, and small-capitalization firms in the United States and in a sample of thirteen European countries. A key contribution of the study lies in its multi-layered analytical design, which goes beyond an aggregate perspective by independently evaluating each combination of region and firm size, thereby providing a more in-depth and comparative analysis of how ESG-related factors are reflected in equity performance across different market and institutional contexts.

The results of the study consistently indicate a negative relationship between ESG, each of its components, and stock returns, aligning with findings from previous research (e.g., de Vincentiis 2023; Escobar-Saldívar et al. 2025; Villarreal-Samaniego et al. 2022). These outcomes line up with two well-established mechanisms in the literature, namely, the risk-mitigation channel, which suggests that stronger ESG practices reduce firms’ exposure to idiosyncratic and tail risks, thereby potentially lowering the required risk premium and average returns (Giese et al. 2019), and the cost channel, which claims that sustainability initiatives involve operational and compliance costs that may depress short-term profitability and stock performance (Jyoti and Khanna 2021; Menicucci and Paolucci 2023). While we acknowledge that the negative relationship between ESG and stock returns may reflect a mixture of reduced risk premia and higher ESG-related costs, further research is warranted to determine whether the negative coefficients in our results are driven primarily by one channel or a combination of both.

The influence of both ESG and the individual pillars on stock returns is generally robust across the full sample of European and U.S. companies included in this study, as summarized in

Table 7. Summary of the significance of results for ESG and its three pillars in the full sample and the European and U.S. sub-samples.

		ESG	ENV	SOC	GOV
All	All	**	**	**	**
	Large-Caps	**	**	**	**
	Mid-Caps	**	*	**	**
	Small-Caps	**	**	**
EC	All	**	**	**
	Large-Caps	**	**	**
	Mid-Caps	**	**	**	**
	Small-Caps	**	**	*
US	All	**		**
	Large-Caps	**	**	**	**
	Mid-Caps	**		*	**
	Small-Caps	*		**

Coefficients significant at ≤0.05 are represented by ** and coefficients significant at ≤0.10 are represented by *. Source: Authors’ own.

. An exception, however, is the governance pillar for small-capitalization firms. This finding also holds true when U.S. and European firms are examined separately. This result likely stems from the ownership structure of Small-Caps, where owners often act as managers, thereby reducing agency problems. On the other hand, the study found that GOV is significant for both European and U.S. medium-capitalization companies’ stock returns. Since these firms are in a growth stage, there are concerns regarding their future performance. Consequently, internal practices and policies, such as governance, may be particularly relevant to investors interested in these companies. Furthermore, while GOV proved significant for Large-Cap firms in the US, it was not for large European firms. Unlike many large U.S. companies, where managers frequently act as agents, managers and shareholders in large European firms are family members, banks, or governments. These divergent ownership structures may help explain the difference in the significance of GOV for U.S. and European Large-Caps.

Another regional difference can be observed in the environmental component of ESG, which appears to have a more significant impact on European companies’ stock returns than on U.S. companies. This is likely due to stricter European regulations on environmental sustainability and a larger European climate-fund market. While U.S. large-cap companies are also influenced by ENV, their global operations and increased public scrutiny may contribute to their ESG focus. The regional differences found in our study align with Legitimacy Theory, since firms operating in different institutional environments face distinct societal expectations and regulatory pressures. Specifically, European companies may be compelled to adopt stronger environmental practices to maintain legitimacy in markets where sustainability standards are more deeply embedded.

This study documents a consistent negative association between ESG scores, their individual pillars, and stock returns across regions and firm size groups. However, the analysis does not allow for causal inference, and the findings should be interpreted as associational. In addition, although the analyses were estimated separately for the U.S., Europe, and firm-size groups, we did not formally test whether the estimated coefficients differ statistically across these subgroups. Such tests represent a natural extension of the present study. Future research could incorporate these formal comparisons and explore identification strategies that more directly isolate the causal mechanisms underlying ESG–return relations.

The findings in this paper are relevant to both practitioners and policymakers. For practitioners, the consistent negative association between ESG scores and stock returns suggests that portfolio creation should carefully account for how ESG attributes alter expected risk–return trade-offs across regions and firm sizes. Specifically, investors who rely on ESG screens or integrate ESG-tilted strategies may need to adjust the weighting of ESG-intensive assets to maintain portfolio allocations that are compatible with their target risk-return profiles. The stronger pillar-specific effects observed in certain market segments further suggest that active managers should evaluate ESG dimensions individually rather than rely solely on aggregate scores. For policymakers, the cross-regional differences identified in this study underscore the importance of harmonizing disclosure standards and supporting regulatory mechanisms that enable pension funds and long-horizon institutional investors to incorporate ESG criteria without compromising their mandate to maximize beneficiaries’ long-term returns.

More research is required to examine whether these results apply in other contexts, such as emerging market firms or specific industrial sectors, and to clarify whether the negative association between ESG, its components, and stock returns is driven primarily by lower profitability or by reduced risk. Future work could incorporate dynamic panel specifications or time-varying models to capture adjustment processes in the ESG–return relationship, and explore potential nonlinearities, for example, threshold or quantile models that assess different effects at low versus high ESG levels. In addition, formal statistical tests for differences across regions and firm-size groups, based on pooled models with interaction terms, would complement the separate estimations presented here. Finally, studies that examine regulatory or institutional changes—such as shifts in ESG disclosure requirements or the introduction of pro- and anti-ESG policies—as quasi-natural experiments could help identify the causal mechanisms underlying ESG–CFP relations more precisely.