Asymmetric impact of ESG performance on systematic risk: exploring beta plus and beta minus dynamics

Systematic risk, often measured by Beta, has been a fundamental concept in finance since the introduction of the Capital Asset Pricing Model (CAPM) by Sharpe (1964), Lintner (1965), and Mossin (1966). The CAPM posits that Beta captures the sensitivity of a firm's returns to market movements, making it a crucial determinant of expected returns and portfolio management. However, recent empirical evidence suggests that risk is not symmetrically distributed, with firms responding differently to market upswings and downturns. This asymmetry has led researchers to explore the decomposition of Beta into upward (Beta+) and downward (Beta-) market sensitivities, as shown in studies by Ang et al. (2006) and Estrada (2002). Understanding this asymmetric behavior is essential for accurately assessing risk and designing effective investment strategies.

Asymmetric risk assessment offers a more precise understanding of how firms respond to different market conditions. Downside beta (β⁻) measures exposure during negative market returns, while upside beta (β⁺) captures sensitivity to market gains. Traditional asset pricing models typically assume symmetry, but empirical evidence highlights the limitations of this assumption. Ang et al. (2006), for example, find that distinguishing between downside and upside betas improves risk assessment, as investors place greater weight on losses than gains. More recently, Bollerslev et al. (2022) show that semi-betas associated with negative market variation represent significantly priced risk factors in the cross-section of stock returns.

Parallel to the advances in Beta decomposition, environmental, social, and governance (ESG) performance has gained significant attention from corporations and investors in recent years. Companies increasingly integrate ESG considerations into their business strategies to meet stakeholder expectations, improve sustainability, and enhance long-term value creation. As ESG practices have become central to corporate strategy, understanding their impact on financial risk has become a critical topic for academics, investors, and policymakers (Gallego-Nicholls et al., 2025).

ESG investing is often associated with reduced firm-level risk and long-term resilience, as firms adopting strong ESG practices are expected to maintain stable stakeholder relationships, avoid reputational damage, and mitigate environmental and regulatory risks. Investors and policymakers have embraced this view, contributing to the rapid expansion of ESG-related funds globally. According to recent studies, firms with high ESG scores are often considered to have a lower cost of capital and greater access to financing (Giese et al., 2019). However, the link between ESG performance and systematic risk remains an area of debate.

An important way to interpret this debate is through the lens of social capital, which refers to the trust, networks, and cooperative norms that strengthen stakeholder relationships. ESG—particularly its Environmental and Social dimensions—can be viewed as a proxy for social capital, since strong ESG practices foster durable ties with employees, customers, suppliers, and communities (Amiraslani et al., 2023). This connection helps explain why the literature often links social capital to systematic risk. For example, Lins et al. (2017) and Albuquerque et al. (2018) show that firms with higher social capital exhibit greater resilience during financial crises, as stakeholder trust cushions them against adverse shocks. Consequently, firms with robust ESG performance, especially in the social dimension, are theorized to benefit from the same mechanism—lower Beta and reduced sensitivity to systematic risk—because ESG-driven social capital provides a buffer during market downturns.

Despite the growing body of research, the relationship between ESG performance and risk remains insufficiently understood. Prior studies provide mixed evidence: some find that ESG reduces risk by strengthening stakeholder trust and improving access to finance (e.g. Lins et al., 2017), while others report limited or even contradictory effects (Giese et al., 2019). Much of this inconsistency arises because most existing work either examines ESG and aggregate Beta without distinguishing between upside and downside risk, or relies on single-country datasets that limit generalizability. As a result, the mechanisms through which ESG influences systematic risk—and whether its impact differs across market states—remain underexplored. Our study addresses this gap by combining a large international sample with a decomposition of Beta into its asymmetric components, providing new evidence on the directional effects of ESG on systematic risk.

Our study builds on this theoretical framework by presenting novel empirical evidence using a dataset of 9,643 firms from 89 countries, offering one of the most comprehensive international analyses of ESG and systematic risk to date. Whereas prior empirical research has primarily focused on single-country samples—for instance, Ang et al. (2006) use data from the United States—we extend the scope by examining a broad cross-country setting. Specifically, we analyze the relationship between the overall ESG score, its three distinct pillars (ESG), and the asymmetric components of Beta: Beta⁺, which measures sensitivity to market upswings, and Beta⁻, which captures exposure to market downturns. To address potential endogeneity concerns, we further employ the First-Difference Regression method, ensuring more robust estimates of the ESG–Beta relationship.

Our results challenge the theoretical predictions of Albuquerque et al. (2018) and related studies. Rather than mitigating risk, higher ESG scores are associated with greater systematic risk. This effect is particularly evident in the positive and statistically significant relationship between ESG scores and both upside and downside betas (β⁺ and β⁻). Importantly, the asymmetry in our findings shows that ESG is more effective in amplifying gains during market upswings than in shielding firms during downturns. The stronger positive association with β⁺ suggests that high-ESG firms are especially sensitive to investor optimism and pro-cyclical market dynamics, although this heightened exposure extends to downturns as well.

We argue that this apparent contradiction stems primarily from two reinforcing mechanisms: demand-driven crowding and valuation premia. The rapid growth of ESG-focused funds has concentrated flows into a relatively narrow set of firms, amplifying their covariance with the market as inflows and outflows move in tandem with investor sentiment. At the same time, investors' willingness to pay a “greenium” elevates valuations and lengthens effective duration, which increases sensitivity to discount-rate shocks. Together, these mechanisms magnify systematic risk without necessarily increasing firms' fundamental cash-flow volatility. As a result, high-ESG firms, despite being promoted as safer or more resilient, empirically behave more like high-beta growth equities. By offering new insights into the asymmetric impact of ESG performance on systematic risk, our study contributes to the growing literature on ESG investing and risk management. The implications of these findings extend to investors, firms, and policymakers, who must carefully evaluate the potential trade-offs between ESG-driven strategies and their impact on financial risk. While ESG performance may enhance firm value during favorable market conditions, its association with greater downside risk underscores the importance of addressing potential vulnerabilities related to investor behavior and overvaluation concerns.

The remainder of the paper is structured as follows. Section 2 reviews the literature on risk, the determinants of beta, the limitations of symmetric risk measures, and the relationship between ESG and systematic risk, and outlines the proposed hypotheses. Section 3 describes the methodology, while Section 4 presents the empirical results. Section 5 discusses these findings, and Section 6 concludes by highlighting the study's main contributions, limitations, and directions for future research.

Risk in financial markets refers to the uncertainty surrounding investment returns. It can be categorized into systematic risk and unsystematic risk. Systematic risk, or market risk, arises from macroeconomic factors that affect the entire market, such as inflation, interest rate changes, geopolitical events, or global financial crises. This type of risk is unavoidable and cannot be mitigated through diversification, as it impacts all market participants simultaneously. In contrast, unsystematic risk is specific to individual companies or industries. It includes risks such as operational failures, management decisions, product recalls, or regulatory changes. Unlike systematic risk, unsystematic risk can be minimized or eliminated by holding a diversified portfolio of investments.

The CAPM, introduced by Sharpe (1964), Lintner (1965), and Mossin (1966), focuses exclusively on systematic risk, under the assumption that investors maintain diversified portfolios that negate the effects of unsystematic risk. The model provides a framework for quantifying the relationship between an asset's expected return and its systematic risk, offering a simple yet powerful tool for asset pricing and portfolio management.

The CAPM builds on the foundation of Markowitz (1952) mean-variance optimization framework. It asserts that the expected return of an asset is a function of its exposure to systematic risk, as measured by the beta coefficient (β). Mathematically, the model is expressed as (Eq. 1):

where E(⁠$Ri$⁠) is the expected return of the asset, $Rf$ is the risk-free rate of return, $E(Rm)$ is the expected return of the market portfolio, and $βi$ quantifies the sensitivity of an asset's returns to changes in the market returns, reflecting its exposure to market risk.

The beta coefficient β can be calculated using the single-index regression model, which establishes a linear relationship between the excess return of an asset and that of the market. This relationship is modeled as:

where $Ri$ is the return of asset i, $Rm$ is the return on the market index, $αi$ is the intercept, $βi$ is the sensitivity of the asset's returns to the market's returns (market beta), and $ϵi$ is the residual (idiosyncratic or firm-specific component). Under this model, $E(ϵi)=0$ and it is assumed that $Cov(Rm,ϵi)=0$ and $Var(ϵi)=σϵi2$⁠. In this framework the beta can be computed as:

and the asset i return variance $Var(Ri)=Var(αi+βiRm+ϵi)$ or:

where $βi2Var(Rm)$ represents the asset i systematic variance and $σϵi2$ idiosyncratic (or specific) variance.

The CAPM formalized the link between beta and the cost of equity, sparking extensive research into why some firms exhibit higher betas than others. Scholars have examined how managerial, accounting, and macroeconomic factors shape this risk parameter over time. The literature on stock betas identifies eight key determinants: earnings characteristics, accounting quality, financial leverage, dividend policy, profitability, firm size and industry effects, macroeconomic conditions, and market liquidity.

Earnings volatility is a core determinant of beta. Beaver et al. (1970) found a high correlation between earnings-to-price volatility and market beta. Christie (1982) confirmed this relationship, breaking down beta into value- and leverage-adjusted components. At the industry level, Brimble and Hodgson (2007) and Parthasarathy (2019) found similar links between earnings variability and beta across different regions, underscoring the importance of earnings volatility as a driver of systematic risk.

Earnings quality, which reflects the accuracy of reported profits as proxies for cash flows, also influences beta. Francis et al. (2005) and Hribar and Jenkins (2004) demonstrated that poor accrual quality and low earnings persistence predict higher betas. Transparency reduces systematic risk, as higher-quality earnings are associated with lower betas and smaller equity risk premiums (Francis et al., 2005).

Leverage increases beta by amplifying equity return volatility. Following Modigliani and Miller (1958, 1963), Hamada (1972) showed that leveraging the balance sheet raises beta. Bhandari (1988) confirmed this, and subsequent studies (Korteweg, 2010; Alcock et al., 2014) corroborated the positive relationship between leverage and beta after accounting for taxes and bankruptcy costs.

Dividend policy can influence beta as well. Stable, high dividends mitigate risk by reducing information asymmetry and agency conflicts. Litzenberger and Ramaswamy (1979) and Rozeff (1982) found that dividend-paying firms had lower betas. DeAngelo et al. (2006) suggested that firms with excess free cash flow face higher betas unless they commit to generous payouts, a finding confirmed by studies on dividend cuts during the 2008–09 crisis (e.g. Graham et al., 2014).

Profitability also affects beta. Gebhardt et al. (2001) found that robust future profit expectations were associated with lower betas. Hou et al. (2015) showed that high operating profitability led to higher returns without increasing betas, suggesting that profitability provides better-than-proportional compensation for systematic risk.

Firm size is a well-known correlate of beta. Chan et al. (1985) and Fama and French (1992) found that small firms typically have higher betas. Fama and French (2015) confirmed that the smallest decile of global stocks had betas around 1.2, compared to 0.9 for the largest decile. Industry characteristics also influence betas, with sectors like biotechnology and mineral exploration exhibiting higher betas (Lakonishok et al., 1994). Gormley and Matsa (2016) found that diversified firms tend to have lower betas due to reduced cash flow volatility.

Furthermore, macroeconomic conditions affect beta. Chen et al. (1986) demonstrated that changes in inflation and industrial production influence beta, increasing it during periods of macroeconomic volatility, while Lettau and Ludvigson (2001) found that stocks become more market-sensitive when the consumption-wealth ratio declines.

Liquidity also impacts beta. Pastor and Stambaugh (2003) found that betas were higher when market liquidity was low. Lamont (1997) confirmed that macroeconomic news and policy surprises rapidly shift betas, particularly for highly leveraged or low-profitability firms.

In sum, factors such as earnings volatility, poor accrual quality, high leverage, dividend policy, profitability, firm size, industry effects, macroeconomic shocks, and liquidity contribute to beta's variability. Stable earnings, high-quality accounting, conservative balance sheets, and deep market liquidity reduce beta, while volatile earnings, high leverage, low profitability, and adverse macroeconomic conditions increase it.

The traditional CAPM is derived under mean–variance preferences (Markowitz, 1952; Sharpe, 1964; Lintner, 1965), where risk is summarized by variance, a symmetric measure that treats upside and downside deviations equally. In this framework, beta reflects an asset's covariance with the market, without distinguishing between positive and negative market states. While this simplification aids tractability, it does not align well with empirical evidence on how investors actually perceive risk. Prospect Theory (Kahneman and Tversky, 1979) demonstrates that investors are loss averse: the value function is steeper for losses than for gains, meaning the disutility of losing a given amount outweighs the utility of gaining the same amount. This asymmetry directly challenges the CAPM's reliance on symmetric variance as the sole measure of risk.

A stream of research in behavioral and empirical finance highlights the importance of downside risk. Ang et al. (2006) show that downside beta—defined as an asset's sensitivity to the market when the market is declining—earns a significant risk premium, whereas upside beta receives little or even negative compensation. Estrada (2002, 2007) further demonstrates that the downside CAPM explains cross-sectional returns more effectively than the standard CAPM, particularly in emerging markets where volatility and return asymmetry are more pronounced.

This emphasis on downside exposure has theoretical foundations in lower partial moment (LPM) frameworks. Hogan and Warren (1974) introduced semivariance-based measures of risk, and Bawa and Lindenberg (1977) extended this into a mean–LPM equilibrium model. Later, Harlow and Rao (1989) generalized asset pricing under LPM preferences, reinforcing the idea that investors care disproportionately about adverse returns.

More recent work connects tail risk and downside sensitivity to return premia. Bali et al. (2014) introduce a hybrid tail covariance risk measure that strongly predicts stock returns, consistent with the pricing of left-tail co-movements even beyond downside beta. Meanwhile, Bali et al. (2009) document a positive intertemporal relation between downside risk—measured through value-at-risk and expected shortfall—and expected returns.

While downside sensitivity dominates the pricing story, related evidence highlights how firm characteristics can mitigate exposure to adverse market conditions. Lins et al. (2017) show that firms with high social capital and trust outperformed during the 2008–2009 financial crisis, suggesting that corporate policies and stakeholder relationships can buffer firms against negative shocks, even if this is not formally expressed through downside beta estimates.

Taken together, the decomposition of beta into upside (β⁺) and downside (β⁻) components better reflects investor behavior under loss aversion. The evidence consistently shows that β⁻ is what investors price, while β⁺ attracts little premium. This framework not only reconciles asset pricing with behavioral insights but also opens the door to examining how firm-level factors such as ESG or Corporate Social Responsibility (CSR) engagement influence exposure to downside risk.

Social capital refers to the trust, norms, and networks that enable cooperation and reciprocity among individuals and organizations. At the firm level, social capital reflects the ability to build durable, trust-based relationships with stakeholders such as employees, customers, suppliers, and communities, thereby reducing coordination costs and strengthening resilience (Coleman, 1988; Putnam, 1995). Because social capital is inherently difficult to observe directly, researchers often rely on ESG metrics as proxies.

In particular, the environmental and social (E&S) dimensions of ESG capture a firm's stakeholder orientation—through measures such as employee relations, community engagement, product responsibility, and environmental stewardship—making them natural indicators of corporate social capital. Lins et al. (2017) demonstrate that firms with higher CSR scores, interpreted as higher levels of social capital, outperformed during the 2008–2009 financial crisis and enjoyed superior access to external finance. Extending this approach, Amiraslani et al. (2023) explicitly employ ESG performance as a proxy for social capital in the bond market and show that, while social capital has little effect in normal times, it proved valuable during the crisis by lowering bond spreads, lengthening maturities, and increasing debt issuance. Similarly, Goss and Roberts (2011) find that socially responsible firms face lower loan spreads, consistent with creditor trust, while El Ghoul et al. (2011) show that stronger CSR (proxied by Morgan Stanley Capital International (MSCI) ESG Stats) reduces firms' cost of equity capital.

Empirical research provides evidence suggesting that ESG reduces systematic risk by strengthening stakeholder trust. For instance, Lins et al. (2017) show that firms with high CSR outperformed their peers during the 2008–2009 financial crisis, as stakeholders were more willing to support firms perceived as socially responsible during a period of eroded market trust. This relationship highlights the role of ESG in building intangible assets such as social capital, which act as a buffer against systematic shocks. By maintaining stakeholder confidence, ESG-oriented firms are better positioned to sustain demand, avoid costly disruptions, and access external financing during periods of uncertainty.

From a theoretical perspective, the relationship between ESG and systematic risk operates through several channels. First, ESG enhances customer loyalty and reduces price elasticity, as argued by Albuquerque et al. (2018). This stability in revenue streams insulates firms from market downturns, reducing their exposure to negative market shocks. Second, ESG initiatives often improve employee satisfaction and retention, which strengthens operational resilience and reduces costs associated with labor turnover. Third, socially responsible firms tend to have better relationships with suppliers and regulators, providing flexibility and support during adverse conditions, further mitigating systematic risk.

In terms of the systematic risk components, ESG is particularly relevant for Beta Minus, which measures a firm's sensitivity to negative market movements. Firms with strong ESG practices are less likely to experience significant declines during market downturns because their stakeholder relationships provide a stabilizing effect. For example, customers may continue to support socially responsible firms even in economic contractions, while employees and suppliers may exhibit greater commitment, reducing the impact of external shocks on firm performance. Conversely, the effect of ESG on Beta Plus, or sensitivity to positive market movements, is less clear. While ESG may enhance growth opportunities by fostering innovation and stakeholder collaboration, its stabilizing nature implies that its effect on upside risk is likely weaker than its impact on downside risk. ESG-oriented firms may prioritize long-term sustainability over short-term gains, which could temper their responsiveness to bullish market conditions (Pineda Perez and Grijalvo, 2025).

By integrating ESG into the analysis of systematic risk, this perspective extends the traditional CAPM framework, which assumes symmetric sensitivity to market movements. ESG's role in strengthening stakeholder relationships highlights the importance of non-financial factors in determining a firm's risk profile, providing a richer understanding of how intangible assets influence systematic risk under different market conditions. This framework underscores the value of ESG not only as an ethical or reputational consideration but also as a strategic tool for risk management.

We conceptualize ESG—especially the Environmental and Social (E&S) pillars—as a firm-level proxy for social capital, understood as trust-based, cooperative relationships with stakeholders that lower coordination frictions and stabilize operations (Coleman, 1988; Putnam, 1995). This interpretation is consistent with evidence that ESG performance functions as social capital and becomes most valuable when trust is scarce, such as during crises (Lins et al., 2017; Amiraslani et al., 2023). Mechanistically, ESG links to reduced information asymmetry and improved financing conditions (Cheng et al., 2014; El Ghoul et al., 2011; Goss and Roberts, 2011; López-Moreno et al., 2024), stronger customer loyalty and lower demand elasticity (Albuquerque et al., 2018), and greater operational resilience through employees, suppliers, and communities (Lins et al., 2017). Because market beta is the covariance of firm returns with the market (Sharpe, 1964; Lintner, 1965), any channel that stabilizes cash flows and relaxes financing constraints—particularly in adverse market states—should reduce that covariance. Lower-partial-moment asset-pricing theory and downside-sensitivity evidence reinforce the asymmetric expectation that these effects are stronger in downturns than in upturns (Bawa and Lindenberg, 1977; Harlow and Rao, 1989; Ang et al., 2006; Estrada, 2007).

In the CAPM framework, systematic risk is captured by beta, defined as the covariance of firm returns with market returns relative to market variance (Sharpe, 1964; Lintner, 1965). Any factor that stabilizes firm cash flows or eases financing frictions should therefore reduce this covariance. ESG—particularly the Environmental and Social (E&S) pillars, conceptualized as a form of social capital—operates through several such channels: it dampens cyclical revenue swings by strengthening customer loyalty and lowering demand elasticity (Albuquerque et al., 2018), mitigates financial stress by improving access to capital on more favorable terms (Cheng et al., 2014; El Ghoul et al., 2011; Goss and Roberts, 2011), and reduces operational disruptions through more resilient relationships with employees, suppliers, and communities (Lins et al., 2017). Evidence that high-ESG firms raised more capital and performed better under stress (Lins et al., 2017; Amiraslani et al., 2023) further supports this stabilizing interpretation. Taken together, ESG as social capital is expected to reduce firms' exposure to aggregate fluctuations, lowering overall systematic risk.

Downside exposure is particularly salient because losses weigh more heavily on investor utility than equivalent gains, and downside covariance with the market is empirically priced in expected returns (Bawa and Lindenberg, 1977; Harlow and Rao, 1989; Ang et al., 2006; Estrada, 2007). Firms with stronger social capital should be especially resilient in adverse states: customer loyalty and lower demand elasticity sustain revenues (Albuquerque et al., 2018), trusted stakeholder relationships stabilize labor and supply chains, and superior access to financing mitigates stress when credit is scarce (Cheng et al., 2014; El Ghoul et al., 2011). These mechanisms collectively dampen left-tail outcomes, consistent with evidence that high-CSR/ESG firms performed better and attracted external capital during the 2008–2009 crisis (Lins et al., 2017), that ESG reduced bond spreads and extended debt maturity under stress (Amiraslani et al., 2023), and that CSR attenuates stock-price crash risk (Kim et al., 2014). Thus, ESG's value as “trust capital” is expected to manifest most strongly in downturns, reducing firms' sensitivity to negative market shocks.

If ESG/social capital functions primarily as a form of “downside insurance,” then its stabilizing benefits should be concentrated in adverse states rather than during expansions. In good times, revenues and financing are less constrained, trust is abundant, and market sentiment is broadly favorable, reducing the marginal value of stakeholder capital. Empirical asset-pricing research indicates that investors demand compensation mainly for downside covariance, while upside covariance is weakly priced, if at all (Ang et al., 2006; Estrada, 2007). Moreover, the product-market loyalty and lower demand elasticity linked to ESG (Albuquerque et al., 2018) may even attenuate firms' sensitivity to upswings relative to more opportunistic, short-term-oriented peers that aggressively chase booming demand. Therefore, the expected ESG–beta link should be asymmetric, with weaker effects on upside beta than on downside beta:

Figure 1 presents a compact theoretical framework: ESG (E&S) as social capital operates through stakeholder trust and coordination, customer-loyalty–driven demand stability, improved employee and supplier relations, and lower information asymmetry with better access to finance; these channels reduce β and, more strongly, β⁻, with a comparatively small effect on β⁺.

This study examines whether firms' ESG performance is associated with their systematic risk. Financial and ESG data are sourced from Refinitiv Eikon. The working sample comprises 9,643 firms from 89 countries. ESG variables (and other firm-level controls) are available annually from 2002–2022. To mitigate simultaneity, all firm characteristics are aligned to pre-2023 information and related to cross-sectional beta measures observed at the beginning of December 2023.

We use Refinitiv Eikon's reported market betas as of December 2023. These are static, trailing-window estimates based on firm-level returns regressed on a country-level market index selected by Refinitiv as the firm's primary benchmark (e.g. S&P 500 for U.S. firms; analogous flagship indices for other countries). We do not re-estimate or adjust these betas; all values are taken as reported by Refinitiv at the stated date.

Concerning the baseline beta (β). For each firm i, Refinitiv estimates the slope coefficient from an ordinary least squares (OLS) market model over the specified trailing window:

So that

For the 5-year beta, Refinitiv uses monthly returns from December 2018 to December 2023 (approximately 60 observations). For the 3-year and 2-year betas, Refinitiv uses weekly returns over the corresponding trailing windows ending in December 2023 (roughly 156 and 104 observations, respectively). Returns are those provided by Refinitiv for each frequency; we rely on Refinitiv's standard construction and do not alter the periodicity, compounding, or dividend treatment.

With respect to asymmetric betas (β⁺ and β^-), Refinitiv also reports “upside” and “downside” betas that condition on the sign of the benchmark return. Let $T+={t:rm,t>0}$ and $T−={t:rm,t<0}$ denote the sets of up- and down-market observations within the same trailing window. Upside and downside betas are estimated as OLS slopes on the corresponding subsamples:

which are equivalent to conditional covariance–variance ratios:

Refinitiv applies the same return frequency and trailing window as for the corresponding overall beta (i.e. monthly for the 5-year measure; weekly for the 3- and 2-year measures), using the firm's country benchmark index to define up and down states.

ESG and control variables are annual panel data from 2002–2022. We merge these with the December 2023 cross-section of (β, β⁺, β⁻) so that all firm characteristics used in the regressions are measured strictly prior to the beta snapshot. In our baseline specification, each firm's ESG and controls are taken from fiscal year 2022 (or the latest available year ≤2022). Where noted in robustness checks, we also consider other lag years. This alignment ensures that explanatory variables predate the dependent variables, helping to limit simultaneity concerns.

Descriptive statistics for the beta measures are reported in Table 1, and the correlation matrix is presented in Table 2. As expected, the ESG pillars are highly correlated.

The key explanatory variable is the ESG Score, which provides an aggregate measure of a firm's performance across ESG dimensions. Refinitiv's ESG database further breaks this score into three pillar components: ESG Scores. The Social Pillar Score, particularly relevant to stakeholder management, captures factors such as labor practices, diversity, and community engagement. The Environmental Pillar Score includes metrics on emissions, resource use, and innovation, while the Governance Pillar Score focuses on board structure, shareholder rights, and transparency. ESG scores were extracted from Refinitiv for all available years up to the most recent fiscal years, 2022 and 2023.

To control for firm-specific factors that might influence both risk and ESG performance, several financial and operational variables are included. These controls capture dimensions such as profitability (Return on Assets), liquidity (Quick Ratio), leverage (Debt-to-Equity Ratio), valuation (Price-to-Book Ratio), and firm size (Market Cap and Employees). The dataset also accounts for firm age (log-transformed), cash holdings as a percentage of total assets, free float, and R&D expenditures. Market Cap and Company Age are included to capture firm maturity and size, which may influence risk exposure. All financial and firm-level variables were extracted from Refinitiv for all available years up to 2022 and 2023.

To test the proposed hypotheses, we define the following cross-sectional regression models (Eqs. 10 -12):

where $ESGi,t$ represents our independent variable measured by one of the three ESG pillars or by the composite ESG Score. Beta, Beta+ and Beta-are the dependent variables. Here, i denotes a firm and Control represents the set of control variables included in the study. The parameters $α$⁠, $γ1$⁠, $γ2$⁠, are to be estimated and $εi$ represents the error term.

For parameter estimation, we employ two approaches. First, we use a pooled OLS model with sector and country fixed effects to address potential omitted variable bias from unobserved characteristics that vary across sectors and countries. To ensure robustness, we also implement a first-difference regression analysis to tackle potential endogeneity by taking differences between the 3-year and 5-year Betas, as well as differences in ESG scores.

The first-difference regression addresses endogeneity by transforming the data into changes between consecutive time periods, effectively removing time-invariant unobserved variables that could bias the estimates. Instead of estimating relationships between levels of the variables, the regression is performed on their changes over time. This differencing eliminates fixed firm-specific factors, reducing omitted variable bias, and can also help mitigate reverse causality if persistent feedback loops drive endogeneity.

We expect a negative relationship between ESG scores and Beta⁻, supporting the hypothesis that higher ESG performance reduces downside risk by stabilizing stakeholder relationships during downturns. Conversely, the relationship between ESG scores and Beta⁺ is expected to be weaker or non-significant, consistent with the idea that ESG primarily mitigates risk rather than amplifying returns during market upswings.

In Table 3, we present the results of our main baseline model, which examines the impact of ESG scores on the 3-year and 5-year Betas. The first column shows a positive and statistically significant relationship between the ESG Score and the 3-year Beta. This finding contradicts our initial expectation, which anticipated a negative relationship between ESG performance and systematic risk. Similarly, column (4), where the dependent variable is the 5-year Beta, confirms this result, showing a positive and statistically significant association at the 1% level. Columns (2) and (3) disaggregate this relationship into Beta⁺ and Beta⁻ over the 3-year period. Here, the results continue to contradict our forecast: the ESG Score is positively related to both Beta⁺ and Beta⁻. Collectively, these findings lead us to reject the hypotheses initially derived from the theoretical model of Albuquerque et al. (2018), which predicted that firms with higher social capital would mitigate risk through strong stakeholder relationships and trust, providing resilience during adverse market conditions.

Our empirical evidence, however, suggests the opposite. Higher ESG scores are associated with increased systematic risk, with the effect more pronounced during positive market movements. The coefficient on the ESG Score is notably larger and more positive for Beta⁺ in both the 3-year and 5-year models, indicating greater sensitivity to market upswings. At the same time, the positive coefficient for Beta⁻ implies that firms with stronger ESG performance also experience higher downside systematic risk, contrary to the expected “risk-buffering” role of ESG.

Table 4 reports the relationship between the ESG Score's individual pillars and the overall Beta, Beta⁺, and Beta⁻. The results indicate no statistically significant association between the Environmental Pillar and any of the Beta measures. In contrast, the Social and Governance Pillars exhibit positive and statistically significant relationships with all Betas. This pattern may reflect heightened investor attention to Social and Governance practices, which amplifies the systematic movements in the stock prices of firms excelling in these areas, while Environmental aspects receive comparatively less market focus.

Consistent with the findings in Table 3, the positive impact of the Social and Governance Pillars is stronger for Beta⁺ than for Beta⁻. This suggests that these scores provide greater benefits during market upswings than protection during downturns, as stocks with high Social or Governance scores tend to experience larger gains in rising markets relative to losses in declining markets.

Overall, the Social and Governance Pillars contribute positively to firm performance, despite the higher level of systematic risk indicated by symmetric Beta. The greater sensitivity to upside movements (Beta⁺) relative to downside movements (Beta⁻) implies that the benefits derived from investor interest and trust in these practices outweigh potential risks, producing a net positive effect on firm value.

In Table 5, we present the results of our first-difference regression, which addresses potential endogeneity by using changes in variables rather than their levels. The dependent variable is the difference between the 3-year and 5-year Betas, while the key independent variables are changes in ESG scores (ESG_t – ESG_t-4 and ESG_t – ESG_t-5). This approach mitigates the influence of time-invariant unobserved factors and reduces concerns related to reverse causality.

The results align with the baseline findings from Tables 3 and 4 Changes in ESG scores remain positively and significantly associated with changes in Beta, indicating that firms with greater increases in ESG performance experience higher systematic risk. The effect is more pronounced for Beta⁺ than for Beta⁻, suggesting that improvements in ESG scores are linked to greater sensitivity to positive market movements than to negative ones.

These results imply that the observed relationship between ESG and market sensitivity is not solely driven by omitted variable bias or reverse causality. Firms with higher ESG growth may attract greater investor attention, amplifying systematic risk. Despite the elevated Betas, the stronger impact on Beta⁺ relative to Beta⁻ indicates that the upside potential during market upswings outweighs the downside risk during downturns, consistent with previous conclusions.

Table 6 extends this analysis by examining the individual ESG pillars in the first-difference framework. The results are broadly consistent with earlier findings, although the effect of the Governance Pillar is somewhat weaker, with coefficients showing lower statistical significance compared to prior specifications.

This study documents a robust positive association between ESG performance and systematic risk, measured by overall beta (β) and its asymmetric components β⁺ and β⁻. Across our global dataset, higher ESG scores are linked to stronger co-movement with the market, particularly for upside sensitivity. These findings are consistent across baseline OLS regressions and first-difference specifications addressing endogeneity. Pillar-level analyses indicate that ESG scores all contribute positively to systematic risk, with Social and Governance components being especially influential. Contrary to the ex-ante expectations from social capital theory (Albuquerque et al., 2018; Lins et al., 2017), ESG does not buffer firms from adverse market states but instead amplifies exposure to both expansions and contractions.

Two complementary mechanisms help explain this paradox. First, demand-driven crowding increases common return variation via flow-induced trading pressure (Greenwood and Thesmar, 2011). Between 2018 and 2024, global sustainable funds absorbed over USD 650 billion in net inflows—roughly one-third of the free-float capitalization of the MSCI World Socially Responsible Investment (SRI) index (Morningstar Manager Research, 2024). Because ESG ratings and policies concentrate on a limited set of firms, new capital disproportionately targets these names, intensifying order-flow concentration. Benchmark design further amplifies the effect: the ten largest constituents of leading ESG indices account for over 40% of index weight (MSCI, 2024), so even broadly diversified funds generate collinear trades. Heinkel et al. (2001) formalized how a “green boycott” reduces free float, pushing prices above fundamentals, while Pastor et al. (2022) show that as the wealth share of sustainability-oriented investors rises, a distinct ESG demand factor emerges, raising β loadings on the market factor. Pedersen et al. (2021) similarly argue that sustainability preferences introduce a priced “green factor” that increases market covariance.

Micro-level evidence supports this demand-pressure channel. van der Beck (2021) finds that one euro of ESG fund inflow generates roughly EUR 0.80 of same-month abnormal return in top-decile ESG holdings, with effects reversing over subsequent quarters, consistent with flow rather than fundamentals. Ben-David et al. (2018) report similar patterns in Exchange-Traded Fund (ETF) rebalancing, while Barberis et al. (2005) show that index inclusions raise systematic correlations absent cash-flow links. Flow-induced crowding affects both upside and downside betas because ESG flows are pro-cyclical: inflows during periods of risk appetite boost covariance with rallies (raising β⁺), whereas outflows in downturns propagate declines (maintaining β⁻). Pastor and Vorsatz (2020) also document that funds with higher sustainability ratings performed better and attracted inflows during market stress. Because mutual fund and ETF demand is largely price-inelastic in the short term (Israeli et al., 2017), these flows generate transitory shocks and tilt the market portfolio permanently, mechanically increasing unconditional betas (Greenwood and Thesmar, 2011). Crowding also interacts with other priced risks: Bolton and Kacperczyk (2022) show that green stocks load positively on carbon-transition factors.

The second mechanism operates through valuation premia and effective duration. Investors deriving non-pecuniary utility from sustainable assets are willing to pay a “greenium,” raising valuations and compressing yields (Pastor et al., 2022; Pedersen et al., 2021). Higher valuations lengthen effective Macaulay duration, increasing sensitivity to discount-rate shocks (Campbell and Mei, 1993). Empirical evidence documents this effect: Zerbib (2019) finds a modest but persistent green premium in corporate bonds, while Pastor et al. (2022) estimate that by 2020, high-ESG firms traded at a 2% aggregate market-cap premium. At the equity level, ESG also carries incremental pricing power beyond traditional factors (Pedersen et al., 2021), whereas carbon-intensive firms trade at a 6–8% discount (Bolton and Kacperczyk, 2023). Intangible-rich, high-ESG firms—common in software, biotech, and renewables—thus resemble growth-duration stocks, with valuations more sensitive to discount-rate dynamics than to short-term cash flows (Peters and Taylor, 2017).

These two channels—flow-driven demand and valuation premia—interact to amplify systematic risk. ESG-driven inflows elevate prices and extend duration, which magnifies β⁺ during rallies and β⁻ in downturns. The greenium typically widens when discount rates fall and attention to sustainability rises (Pastor et al., 2022), reinforcing upside sensitivity, while stress episodes, rising discount rates, and ESG fund redemptions amplify downside betas (Pastor and Vorsatz, 2020), even when firms' fundamentals remain stable.

In sum, the positive relationship between ESG and systematic risk does not reflect higher fundamental cash-flow volatility. Rather, it reflects how market structures and investor preferences shape pricing: concentrated, pro-cyclical flows and duration-extended valuations increase co-movement with the market. Consequently, high-ESG firms, often marketed as safer or more resilient, empirically behave like high-beta growth equities.

This paper demonstrates that stronger ESG performance is associated with greater systematic risk, as measured by overall beta (β) and its asymmetric components, β⁺ (upside) and β⁻ (downside). Using a global sample of 9,643 firms across 89 countries, we document positive and statistically significant links between ESG and all three beta measures, with the association particularly pronounced for upside sensitivity. These findings are robust across pooled OLS models with country and sector fixed effects and first-difference specifications, and hold at the pillar level, with Social and Governance components especially influential. By focusing on β⁺ and β⁻ rather than a single symmetric beta, the study departs from prior work and directly examines state-contingent exposure, overturning the conventional “downside-insurance” interpretation of ESG.

Theoretically, the results reframe ESG's role in asset pricing from risk mitigation to covariance amplification. Two reinforcing channels explain this pattern. First, demand-driven crowding: pro-cyclical flows into a concentrated set of ESG-preferred names amplify common return variation. Second, valuation premia lengthen effective duration: greeniums raise valuations, increasing sensitivity to discount-rate shocks. Together, these mechanisms magnify β⁺ in rallies and sustain β⁻ in downturns, without necessarily increasing fundamental cash-flow volatility. By decomposing beta asymmetrically, the study clarifies that ESG primarily affects conditional covariance with the market rather than idiosyncratic risk.

These findings have important implications. For investors, portfolio construction and risk management should explicitly account for upside and downside betas when designing ESG-tilted mandates, stress-testing discount-rate exposures, and monitoring concentrated flows into popular ESG holdings. In expected-return budgeting and cost-of-capital assessments, ESG should not be assumed to reduce systematic risk; instead, it is associated with higher market covariance and greater sensitivity to discount-rate dynamics, particularly when valuations carry a greenium. For firms, the results caution against framing ESG as inherently resilient, as higher ESG scores appear to amplify exposure to broad market movements. At the policy level, supervisors could improve transparency by requiring ESG fund-labeling rules that disclose flow concentration, top-holding overlaps, and index weight distributions, while index providers could mitigate excessive co-movement by publishing concentration metrics and considering capping rules for heavily top-weighted ESG indices. Collectively, these measures would better align market practices with the state-contingent behavior of ESG-intensive firms, enhancing both information quality and financial stability in sustainable finance.

Several limitations qualify our inferences. We rely on Refinitiv's reported betas at a December 2023 snapshot and on annual ESG data through 2022; we do not re-estimate betas or harmonize methods across data providers, leaving room for measurement error and comparability concerns. While first-difference regressions mitigate time-invariant confounding, residual endogeneity from time-varying unobservables or feedback from risk to ESG may remain, so causal claims are not warranted. ESG pillar scores are correlated, and effects likely vary across sectors, regions, and firms' intangible intensity, aspects we do not fully model. Finally, although the demand- and duration-based mechanisms align with our findings, we do not directly observe firm-level flow elasticities or effective duration.

These limitations highlight several avenues for future research. Event-style or longitudinal studies could examine how β⁺ and β⁻ evolve around shocks such as index inclusions, fund reclassifications, or regulatory changes. Detailed data on fund ownership and flows would help clarify how investor demand translates into market co-movement, while direct tests of valuation duration could better capture sensitivity to discount-rate dynamics. Replicating the analysis with alternative ESG datasets and outcome-based sustainability measures would help distinguish genuine effects from greenwashing and account for differences across rating providers. Extending the analysis to credit markets or exploring sectoral and regional heterogeneity would provide a broader understanding of how ESG shapes systematic risk. Collectively, these steps would refine the central insight of this paper: ESG should be understood as a preference-driven source of covariance shaped by market structure and discount rates, rather than a simple form of downside protection.

The authors are grateful to the two anonymous reviewers for their constructive comments and insightful suggestions, which significantly improved the quality, clarity, and rigor of this paper.