When ESG Starts to Pay Off: Nonlinear PSTR Evidence on Bank Performance and Stability in Europe and the USA

Abstract

This paper investigates the impact of Environmental, Social, and Governance (ESG) performance on the financial outcomes of 68 European and 60 U.S. banks over the period 2010–2022 using a Panel Smooth Transition Regression (PSTR) framework. Unlike traditional linear models, the PSTR approach captures the nonlinear, regime-dependent effects of ESG engagement on bank profitability, measured by ROA and ROE, and financial stability, measured by the Z-score. Our empirical findings reveal a critical ESG threshold in both regions, above which banks experience substantial improvements in profitability and resilience. Comparative analysis indicates that while ESG enhances stability slightly more in European banks, U.S. banks tend to achieve marginally higher profitability gains. Control variables, including bank size, capital adequacy, leverage, and macroeconomic conditions, also play a significant role in shaping performance. These results underscore the importance for banks to attain a minimum ESG maturity to fully realize the benefits of sustainable practices. The study provides valuable insights for bank managers, investors, and policymakers seeking to promote a sustainable and resilient banking sector across Europe and the United States.

1. Introduction

In recent years, Environmental, Social and Governance (ESG) criteria have become a decisive component of strategic planning and financial decision-making in the banking industry. Investors, regulators, and the public are increasingly attentive to how banks manage sustainability issues, pushing institutions on both sides of the Atlantic to reconsider the way they operate (Busch & Friede, 2018). In Europe, the transition toward sustainable finance has been strongly shaped by an integrated regulatory framework that includes the Sustainable Finance Disclosure Regulation (SFDR), the EU Taxonomy, and the Corporate Sustainability Reporting Directive (CSRD). These initiatives require banks to provide more transparent information on their ESG exposure and to align their practices with clearly defined sustainability objectives (Refinitiv, 2021).

U.S. banks, however, evolve within a distinct institutional setting. Although the Federal Reserve, the Office of the Comptroller of the Currency (OCC), and the Securities and Exchange Commission (SEC) have increasingly emphasized climate-related risks and ESG reporting, the U.S. regulatory landscape remains less uniform and more politically contested than the European one. Without a harmonized national taxonomy or mandatory sustainability classification system, U.S. banks often rely on voluntary reporting standards such as SASB or TCFD, and their ESG strategies are shaped more by market forces, institutional investors, and shareholder activism than by regulatory mandates. This divergence in regulatory frameworks has been shown to generate significant differences in the scope, intensity, and credibility of ESG integration between European and U.S. financial institutions (European Commission, 2023; OECD, 2024). Consequently, ESG performance may affect financial outcomes differently across the two regions.

Prior studies have reported mixed evidence regarding the impact of ESG performance on banks’ financial outcomes, suggesting that the relationship remains inconclusive and warrants further investigation. On the one hand, engaging in ESG activities can reduce exposure to legal, operational, and reputational risks (Goss & Roberts, 2011; Cek & Eyupoglu, 2020), help build stronger relationships with stakeholders (Cheng et al., 2014), lower funding costs by improving a bank’s perceived risk profile (Cornett et al., 2016), and reinforce risk management systems, particularly in relation to environmental or social vulnerabilities (Weber, 2017). These effects may be especially pronounced for large U.S. institutions that are under continuous scrutiny from global investors and are increasingly expected to demonstrate credible climate and sustainability commitments. On the other hand, ESG initiatives may require substantial financial resources and organizational adjustments, potentially weighing on short-term earnings—especially when banks pursue ambitious sustainability goals without having adequate internal capabilities. Banks also differ from non-financial firms in their heavy reliance on regulatory oversight, their dependence on public trust, and their exposure to systemic risks. These distinctive features suggest that the ESG–performance–stability nexus may follow nonlinear patterns that depend on the bank’s level of ESG engagement.

Despite the growing body of research on ESG and corporate outcomes, empirical evidence remains inconclusive regarding the impact of ESG engagement on bank profitability and financial stability. Several studies report a positive relationship, arguing that ESG practices improve stakeholder trust, risk management quality, and long-term financial performance (Cornett et al., 2016; Gangi et al., 2018; Weber, 2017). In contrast, other studies emphasize the substantial implementation and compliance costs associated with ESG initiatives, suggesting that sustainability investments may reduce profitability, particularly in the short run (Buallay, 2019). More importantly, recent research has begun to suggest that the ESG–performance relationship may not be linear. The financial benefits of ESG may only emerge after institutions reach a sufficient level of sustainability maturity, governance quality, or organizational capability (Nollet et al., 2016; Bitar et al., 2018). However, empirical evidence on such nonlinear effects remains limited in the banking sector and is particularly scarce in comparative studies between European and U.S. banks, which operate under markedly different regulatory and institutional environments. This gap in the literature motivates the present study.

Motivated by these considerations, this paper examines the impact of ESG performance on the profitability and stability of European and U.S. banks through a Panel Smooth Transition Regression (PSTR) framework. This method is particularly suited to capturing gradual transitions between different regimes, allowing the effect of ESG to vary depending on the degree of sustainability engagement. Such an approach is highly relevant for banks, as ESG practices tend to evolve progressively, and their financial consequences may only become evident once a certain threshold of maturity or credibility is reached.

Accordingly, this study contributes to the literature in three important ways. First, it extends the growing ESG–banking literature by explicitly examining whether the impact of ESG on bank profitability and stability is nonlinear. Second, it provides one of the first comparative analyses between European and U.S. banks using a Panel Smooth Transition Regression (PSTR) framework, thereby accounting for potential threshold effects and regime changes in the ESG–performance nexus. Third, it offers practical implications for regulators, investors, and bank managers by identifying the ESG thresholds above which sustainability engagement generates significant improvements in profitability and financial resilience.

The institutional environment in which banks operate strongly shapes ESG threshold dynamics. Capelle-Blancard and Petit (2019) show that ESG news triggers significantly stronger equity market reactions in Europe than in the U.S., a difference they attribute to Europe’s more credible mandatory disclosure standards. For banks, higher ESG scores are associated with stronger profitability through several channels: enhanced customer trust, improved deposit stability, better governance quality, and more efficient credit monitoring (Gangi et al., 2018; Cornett et al., 2016). By estimating region-specific ESG thresholds for Europe and the U.S., our comparative PSTR analysis directly tackles these institutional differences—an analytical approach that remains scarce in the existing literature.

The remainder of the paper is structured as follows. Section 2 reviews the theoretical and empirical literature on ESG, bank performance, and financial stability. Section 3 presents the empirical framework, including the PSTR model, the data used in the analysis and discusses the results, with emphasis on the nonlinear relationships detected. Section 4 concludes and suggests policy recommendations.

2. Theoretical Background

2.1. Theoretical Foundations for Nonlinear ESG–Performance Dynamics

The theoretical foundation for a nonlinear ESG–performance relationship rests on several interconnected mechanisms.

First, from a resource-based view (RBV), ESG investments require the development of specialized organizational capabilities, including data collection systems, reporting infrastructure, staff expertise, and governance processes. These capabilities are characterized by significant fixed costs and learning curves, implying that the marginal benefits of ESG engagement may be limited until a critical mass of organizational competence is achieved (Barney, 1991; Hart, 1995). In the banking context, this implies that smaller or less sophisticated institutions may struggle to capture ESG benefits until they have developed sufficient internal capacity.

Second, stakeholder theory suggests that the benefits of ESG engagement in terms of enhanced trust, improved reputation, and stronger stakeholder relationships accumulate gradually. Stakeholders—including investors, customers, regulators, and employees—require consistent evidence of sustainability commitment before adjusting their perceptions and behaviors. This accumulation process implies a threshold beyond which stakeholder trust generates tangible financial benefits (Freeman, 1984; Donaldson & Preston, 1995). For banks, this is particularly relevant as customer loyalty, deposit stability, and investor confidence are built over time through credible and sustained ESG engagement.

Third, the integration of ESG considerations into risk management systems involves organizational adaptation that follows an S-shaped learning curve. Early stages of ESG implementation may be characterized by compliance costs and organizational friction, while mature ESG practices generate risk mitigation benefits through improved credit assessment, reduced reputational exposure, and enhanced regulatory alignment (Weber, 2017; D’Orazio & Popoyan, 2019). This dynamic is especially pronounced in banking, where risk management systems are deeply embedded in organizational routines and require substantial adaptation to incorporate sustainability factors.

These theoretical mechanisms collectively suggest that the ESG–performance relationship is unlikely to be linear. Instead, we expect a smooth transition from a low-ESG regime, where the costs of ESG implementation may outweigh the benefits, to a high-ESG regime, where banks capture substantial financial and stability gains. This S-shaped pattern is precisely what the Panel Smooth Transition Regression (PSTR) framework is designed to capture, as it allows coefficients to vary gradually according to the level of ESG engagement, without imposing an arbitrary or discrete split of the sample.

2.2. Empirical Literature on ESG and Bank Performance

Research examining the link between ESG engagement and firm performance has grown substantially over the past two decades. Early surveys of the literature generally concluded that corporate social and environmental initiatives were associated with improved financial outcomes, reflecting reputational gains, enhanced stakeholder relations, or operational efficiencies (Margolis & Walsh, 2003; Friede et al., 2015). An important contribution in the banking literature is provided by Wu and Shen (2013), who examine the relationship between corporate social responsibility (CSR) and bank financial performance. Their findings indicate that CSR engagement positively affects bank profitability and asset quality. However, unlike the present study, Wu and Shen (2013) reject nonlinear specifications and conclude that the CSR–performance relationship is predominantly linear.

Contrary to the early conclusion of Wu and Shen (2013) that the CSR–performance link in banking is essentially linear, a growing strand of the recent literature points to more complex dynamics. Brooks and Oikonomou (2023), in a meta-analysis of more than 1200 studies, demonstrate that the ESG–financial performance relationship varies considerably across sectors, regions, and ESG maturity levels. Within the banking sector, Chiaramonte et al. (2022) find that ESG components only reduce tail risk beyond a minimum threshold, while La Torre et al. (2024) identify governance-driven nonlinearities in bank stability. Moreover, Ciner et al. (2023) show that ESG profitability gains in European banks are amplified under strict regulations like the EU Taxonomy but remain limited in loosely regulated environments. Overall, this evidence implies that linear models risk being misspecified and that nonlinear, threshold-based approaches, including the PSTR framework used here, are increasingly necessary.

Although much of this work initially focused on non-financial corporations, recent studies have increasingly analyzed the banking sector, where ESG considerations intersect closely with regulatory compliance, prudential supervision, and risk management. For banks, higher ESG scores have been linked to stronger profitability through channels such as increased customer trust, improved deposit stability, enhanced governance quality, and more efficient monitoring of credit exposures (Gangi et al., 2018; Cornett et al., 2016). Environmental policies can help banks reduce the likelihood of stranded assets, while social initiatives may strengthen borrower relationships and reduce credit risk (Weber, 2017).

Despite this encouraging evidence, empirical findings remain far from unanimous. Several studies emphasize that sustainability initiatives often entail significant upfront costs related to data collection, reporting systems, staff training, or compliance programs that can temporarily depress returns (Buallay, 2019). Others argue that the benefits of ESG materialize only once institutions achieve a certain degree of sophistication in their sustainability frameworks, making the gains neither immediate nor linear. This is particularly relevant in the banking industry, where institutions differ widely in their internal ESG capabilities, the maturity of their risk-governance structures, and their exposure to sustainability-driven regulatory requirements. As a result, studies relying solely on linear models may fail to identify threshold effects or phases in which ESG produces diminishing, neutral, or even adverse financial impacts.

An expanding strand of research investigates the implications of ESG for banking stability. Several works find that banks with stronger ESG performance undertake more prudent risk-taking, manage their loan portfolios more responsibly, and display higher resilience during periods of economic turbulence (Nollet et al., 2016; Capelle-Blancard & Petit, 2019; Liaqat et al., 2026). Governance quality, in particular, plays a central role by constraining managerial opportunism, improving operational oversight, and reducing the incidence of misconduct risks (Saidi, 2020). Environmental screening contributes to limiting banks’ exposure to sectors vulnerable to transition or physical climate risks, while social practices such as financial inclusion and community engagement can mitigate reputational risks and enhance long-term deposit stability (D’Orazio & Popoyan, 2019).

Yet the relationship between ESG and stability is not uniformly positive. Several scholars have pointed out that some green or socially oriented credit policies may channel funds to emerging industries with limited historical data or uncertain creditworthiness (Krueger et al., 2020). Similarly, expanding ESG initiatives may increase organizational complexity, raise compliance costs, or encourage risk-shifting in competitive markets where banks attempt to differentiate themselves through sustainability branding (Neitzert & Petras, 2021). These concerns highlight that the stability benefits of ESG depend on the strategic alignment of sustainability practices with core risk-management frameworks and on the regulatory consistency surrounding ESG implementation.

2.3. Cross-Regional Differences and Institutional Context

The literature has also begun to recognize that the ESG–performance and ESG–stability relationships may differ substantially across institutional environments. European banks operate within a highly structured regulatory setting characterized by the EU Taxonomy, the SFDR, and the CSRD, which impose clear disclosure requirements and incentivize financial institutions to integrate sustainability into their business models. This regulatory harmonization contributes to reducing information asymmetry and may amplify the financial benefits of ESG by making sustainability commitments more credible to investors and supervisors. In contrast, U.S. banks face a more fragmented and politically polarized environment. While federal agencies such as the SEC and the Federal Reserve have intensified their focus on climate-related disclosure, the absence of unified national sustainability rules and the presence of anti-ESG political pressures create greater uncertainty around ESG adoption. These differences likely shape the conditions under which ESG contributes to bank profitability and financial resilience and may introduce additional sources of nonlinear or asymmetric effects.

The institutional setting within which banks operate is a key determinant of ESG threshold dynamics. Capelle-Blancard and Petit (2019) demonstrate that ESG news elicits significantly stronger equity market responses in Europe than in the U.S., a finding they trace to Europe’s more credible and mandatory disclosure standards. Meanwhile, Cornett et al. (2016) document that U.S. banks’ ESG engagement translated into improved financial performance only after the 2008 crisis, pointing to a regime shift triggered by an external shock. By contrast, Gangi et al. (2018) report that European banks experience relatively larger stability gains than profitability enhancements, which they attribute to the forward-looking nature of EU regulatory frameworks.

The European regulatory framework, characterized by the SFDR, EU Taxonomy, and CSRD, imposes clear disclosure requirements and incentivizes financial institutions to integrate sustainability into their business models. This regulatory harmonization contributes to reducing information asymmetry and may amplify the financial benefits of ESG by making sustainability commitments more credible to investors and supervisors. In contrast, U.S. banks face a more fragmented and politically polarized environment. While federal agencies such as the SEC and the Federal Reserve have intensified their focus on climate-related disclosure, the absence of unified national sustainability rules and the presence of anti-ESG political pressures create greater uncertainty around ESG adoption. These differences likely shape the conditions under which ESG contributes to bank profitability and financial resilience and may introduce additional sources of nonlinear or asymmetric effects.

2.4. Theoretical Synthesis and Research Hypotheses

Building on the theoretical mechanisms outlined above, namely, the organizational capabilities required for ESG implementation, the gradual accumulation of stakeholder trust, and the S-shaped learning curve associated with ESG integration into risk management, we argue that the ESG–performance relationship is inherently nonlinear. Below a certain threshold of ESG maturity, banks bear the costs of sustainability investments without capturing significant financial or stability benefits. Beyond this threshold, however, ESG engagement generates substantial improvements in profitability and resilience. To empirically capture these anticipated regime-dependent dynamics, we employ the Panel Smooth Transition Regression (PSTR) framework, which allows coefficients to change gradually across regimes without imposing an arbitrary discrete split of the sample.

Based on the foregoing theoretical and empirical discussion, we formulate the following hypotheses:

H1. 

The relationship between ESG engagement and bank financial performance is nonlinear, characterized by a smooth transition from a low-ESG regime to a high-ESG regime.

H2. 

The threshold level of ESG maturity beyond which financial benefits materialize differs between European and U.S. banks, reflecting distinct institutional and regulatory environments.

H3. 

Crossing the ESG threshold is associated with economically significant improvements in both profitability (ROA, ROE) and financial stability (Z-score).

3. Materials and Methods

3.1. Model, Data and Variables

The PSTR model introduced by Gonzalez et al. (2005) and extended by Colletaz and Hurlin (2006) is well suited to this purpose. The PSTR framework allows coefficients to change gradually across different regimes, making it possible to identify how ESG impacts evolve as banks progress from low to high sustainability engagement. This approach is particularly relevant for banks, where ESG implementation is incremental and shaped by learning effects, regulatory adaptation, and evolving market expectations. Despite its methodological advantages, the PSTR model remains underutilized in empirical ESG–banking research, which has largely relied on linear panel models or static nonlinear techniques. Although the PSTR framework is particularly suitable for capturing nonlinear and regime-dependent dynamics, potential endogeneity concerns may still arise in the ESG-bank performance relationship. In particular, reverse causality may exist because more profitable and financially stable banks may possess greater financial resources and stronger incentives to invest in ESG activities. In addition, omitted stakeholder-related factors such as institutional ownership, analyst coverage, or governance quality may simultaneously influence both ESG engagement and financial performance.

The present study addresses this gap by applying a PSTR approach to examine how ESG affects profitability and financial stability across different stages of ESG maturity in both European and U.S. banking systems over the period 2010–2022. The sample selection process was conducted in several stages. We initially identified all publicly listed commercial banks available in the Refinitiv Eikon and Bankscope/Orbis Bank Focus databases over the period 2010–2022. Banks with insufficient ESG disclosure, incomplete balance-sheet information, or substantial missing data were excluded to ensure the reliability and consistency of the panel estimations. For the remaining banks, missing observations for control variables were handled as follows: (i) if fewer than 5% of observations for a given variable were missing, linear interpolation was applied; (ii) if 5–10% of observations were missing, the bank was retained but missing values were replaced with the bank-specific mean; (iii) if more than 10% of observations were missing for any key variable, the bank was excluded from the sample. This approach ensured that our panel data are balanced and complete for the final estimation sample. After applying these filters, the final sample consisted of 68 European banks and 60 U.S. banks (The list of European and US banks is provided in Appendix A). Banks with insufficient ESG disclosure, incomplete balance-sheet information, or substantial missing data were excluded to ensure the reliability and consistency of the panel estimations. Financial statement variables for the post-2014 period were obtained from Bureau van Dijk’s Orbis Bank Focus database, which represents the continuation and successor database of Bankscope. Formally, the model can be expressed as

$$y_{it}={\alpha }_i+{\beta }_0^{\prime }{X}_{it}+{\beta }_1^{\prime }{X}_{it}\ast g\left({ESG}_{it}; \gamma ;c\right)+{\epsilon }_{it}$$

(1)

Although the PSTR framework is particularly suitable for capturing nonlinear and regime-dependent dynamics, potential endogeneity concerns may still arise in the ESG–bank performance relationship. In particular, reverse causality may exist because more profitable and financially stable banks may possess greater financial resources and stronger incentives to invest in ESG activities. In addition, omitted stakeholder-related factors such as institutional ownership, analyst coverage, or governance quality may simultaneously influence both ESG engagement and financial performance. The aggregate ESG score is employed as the primary threshold variable in our main analysis for several reasons. First, from a regulatory and investor perspective, ESG is increasingly assessed as a composite measure, with rating agencies and investors using aggregate scores for benchmarking, portfolio allocation, and risk assessment (Berg et al., 2022). Second, the three pillars (E, S, and G) are conceptually interconnected: strong governance often facilitates better environmental and social performance, while social practices can reinforce governance mechanisms (Eccles et al., 2014). Third, using an aggregate measure provides a parsimonious and interpretable threshold that reflects overall sustainability maturity, which is relevant for bank managers and policymakers seeking practical guidance. However, we recognize that aggregation may conceal heterogeneous effects across pillars. For instance, governance improvements may yield more immediate financial benefits, while environmental practices may primarily affect long-term risk exposure rather than short-term profitability. More explicitly, the model to estimate is

$$\begin{array}{c}{y}_{it}={\alpha }_i+{\beta }_0^{ESG}{ESG}_{it}{ + \beta }_0^{SIZE}{SIZE}_{it}+{\beta }_0^{CAP}{CAP}_{it}+{\beta }_0^{LTD}{LTD}_{it}+{\beta }_0^{GDP}{GDP}_{it} +\\ {\beta }_0^{IR}{IR}_{it}{ + (\beta }_1^{ESG}{ESG}_{it}{ + \beta }_1^{SIZE}{SIZE}_{it}+{\beta }_1^{\prime }{CAP}_{it}+{\beta }_1^{LTD}{LTD}_{it}+{\beta }_1^{GDP}{GDP}_{it}+ {\beta }_1^{IR}{IR}_{it}{)\ast g\left(ES{G}_{it};\gamma ,c\right) + \epsilon }_{it}\end{array}$$

(2)

with

$$g(ES{G}_{it};\gamma ,c)={\displaystyle \frac{1}{1+\mathrm{e}\mathrm{x}\mathrm{p}[-\gamma (ES{G}_{it}-c\left)\right]}}$$

The threshold parameter (c) identifies the ESG score level at which the relationship between ESG engagement and bank performance changes significantly. Meanwhile, the smoothness parameter (γ) determines the speed of transition between the low-ESG and high-ESG regimes. Higher values of γ indicate sharper transitions between regimes. Unlike conventional linear panel models that impose constant coefficients across all observations, the PSTR framework allows the impact of explanatory variables to vary smoothly according to the level of ESG engagement. This characteristic is particularly relevant in the banking industry where ESG implementation generally evolves progressively rather than instantaneously.

• Regime 1: Low ESG

  $$g\left(ES{G}_{it}\right)\approx 0$$

  $$y_{it}={\alpha }_i+{\beta }_0^{\prime }{X}_{it}+{\epsilon }_{it}$$
• Regime 2: High ESG

  $$g\left(ES{G}_{it}\right)\approx 1$$

  $$y_{it}={{\alpha }_i+({\beta }_0+{\beta }_1)}^{\prime }{X}_{it}+{\epsilon }_{it}$$

  where
  • ${\beta }_0$ is the coefficient in the low-ESG regime (when ESG is far below the threshold $c$). ${\beta }_0+{\beta }_1$ is the coefficient in the high-ESG regime (when ESG is above the threshold).
  • $g(ESG;\gamma ,c)$ is the transition function: close to 0 in low ESG, close to 1 in high ESG; $\gamma$ determines how steeply the transition occurs.
  • $y_{it}$ denotes the dependent variable bank performance measured by ROA, ROE, or Z-score; $X_{it}$ is the vector of explanatory variables including ESG score and control variables; ${\mu }_i$ captures bank-specific fixed effects; and ${\epsilon }_{it}$ is the idiosyncratic error term. The variables included in the analysis are defined as follows:
    • ROA (Return on Assets): Net income divided by total assets, measuring bank profitability. The data are extracted from Bankscope.
    • ROE (Return on Equity): Net income divided by shareholders’ equity, reflecting return to equity holders. The data are extracted from Bankscope.
    • Z-score: Indicator of bank stability, calculated as (ROA + CapitalRatio)/σ(ROA), where σ(ROA)is the standard deviation of ROA. (Author’s Calculations). Z-score: A widely used indicator of bank financial stability calculated as the ratio of return on assets plus the capital ratio divided by the standard deviation of return on assets. Higher Z-score values indicate greater banking stability and lower insolvency risk.
    • ESG Score: Composite score (0–100) capturing environmental, social, and governance performance. The data are extracted from Refinitiv Eikon/Refinitiv ESG database.
    • Size: Natural logarithm of total assets. The data is extracted from Bankscope.
    • Capital Ratio (CAP): Equity-to-Assets ratio (%). The data are extracted from Bankscope.
    • Loan-to-Deposit Ratio (LTD): Ratio of loans to deposits, measuring liquidity risk. The data are extracted from Bankscope.
    • GDP Growth (GDP): Annual growth rate of the country’s GDP, as a macroeconomic control. The data are extracted from World Bank World Development Indicators.
    • Interest Rate (IR): Annual benchmark interest rate of the bank’s home country. The data are extracted from World Bank World Development Indicators.

Table 1. Descriptive Statistics.

Variable	Region	Mean	Q1	Q2	Q3	Std. Dev.	Min	Max	Unit
ROA	Europe	0.62	0.32	0.58	0.89	0.48	−2.10	2.45	%
ROA	US	1.00	0.60	0.90	1.40	0.50	−0.50	2.50	%
ROE	Europe	7.98	4.21	7.85	11.20	5.61	−12.32	21.90	%
ROE	US	4.50	3.10	4.20	5.80	2.00	−1.00	8.00	%
Z-score	Europe	21.4	15.2	20.8	26.9	8.7	6.1	45.8	Index
Z-score	US	20.5	14.8	19.9	25.5	8.2	8.0	35.0	Index
ESG Score	Europe	55.7	48.3	56.2	63.5	13.4	21	88	0–100
ESG Score	US	55.1	47.9	54.8	62.4	10.0	30	75	0–100
Capital Ratio	Europe	14.9	11.2	14.5	18.2	3.8	8.5	26.1	%
Capital Ratio	US	12.0	9.3	11.8	14.4	4.0	6.0	25.0	%
Loan-to-Deposit Ratio	Europe	102.1	85.2	101.5	118.4	21.4	55	162	%
Loan-to-Deposit Ratio	US	75.0	68.5	74.2	80.8	10.5	50	95	%
GDP Growth	Europe	1.65	0.25	1.80	2.95	2.06	−7.80	5.90	%
GDP Growth	US	2.0	1.2	2.1	2.8	1.0	−1.0	4.50	%

Note: All financial variables are expressed in percentage points except where otherwise indicated. ROA and ROE are reported in percentage points (e.g., 0.62% and 7.98%). The Z-score is a unitless index. ESG scores range from 0 to 100. GDP growth is expressed in percentage points. This uniform presentation ensures comparability across regions.

reports the descriptive statistics for the European and U.S. banking samples. To ensure methodological consistency, identical data sources, variable definitions, and measurement procedures were applied across both samples. However, differences in descriptive statistics are expected due to structural, institutional, and regulatory differences between the two banking systems. Therefore, the objective of Table 1 is to describe the characteristics of each sample rather than to demonstrate statistical similarity between European and U.S. banks. To provide a more comprehensive description of the sample, we report the mean, median, standard deviation, minimum, maximum, and interquartile range for all variables. The inclusion of median and quartile statistics facilitates the identification of potential asymmetries and extreme observations and improves the transparency of the comparison between the European and U.S. banking samples.

Results indicate that European and U.S. banks exhibit relatively comparable ESG scores, with average ESG values exceeding 55 in both regions. Moderate differences remain observable across profitability, capitalization, and liquidity indicators, reflecting structural differences in banking regulation, monetary conditions, and balance-sheet composition between the two banking systems. The higher average ROA observed for U.S. banks (1.00%) compared to European banks (0.62%) reflects differences in net interest margins, operating efficiency, and competitive dynamics between the two banking markets.

As shown in

Table 2. Correlation Matrix.

European Banks
Variable	ESG	ROA	ROE	Z-Score	Size	Capital Ratio	LTD	GDP	IR
ESG	1.00
ROA	0.35	1.00
ROE	0.38	0.75	1.00
Z-score	0.45	0.60	0.55	1.00
Size	0.30	0.40	0.35	0.25	1.00
Capital Ratio	0.40	0.50	0.45	0.60	0.50	1.00
LTD	−0.10	−0.20	−0.15	−0.30	0.05	−0.10	1.00
GDP Growth	0.15	0.30	0.25	0.20	0.10	0.15	−0.05	1.00
Interest Rate	−0.05	−0.10	−0.08	−0.12	0.00	−0.05	0.02	−0.10	1.00
U.S. Banks
Variable	ESG	ROA	ROE	Z-Score	Size	Capital Ratio	LTD	GDP	IR
ESG	1.00
ROA	0.30	1.00
ROE	0.32	0.75	1.00
Z-score	0.28	0.60	0.55	1.00
Size	0.18	0.25	0.28	0.20	1.00
Capital Ratio	0.22	0.40	0.45	0.50	0.30	1.00
LTD	−0.05	−0.10	−0.12	−0.15	−0.05	−0.10	1.00
GDP Growth	0.10	0.20	0.22	0.25	0.05	0.15	−0.05	1.00
Interest Rate	0.00	−0.05	−0.06	−0.08	−0.01	−0.03	0.02	−0.10	1.00

, the pairwise correlations between the main variables remain moderate, with no coefficients approaching levels that would indicate serious collinearity concerns. Consequently, the model’s coefficients can be interpreted with confidence, and the incremental contribution of each variable to bank performance and stability can be reliably assessed.

To address the issue of multicollinearity more conclusively than the simple correlation matrix allows, we now present the VIF diagnostics, which assess how much the variance of each coefficient is inflated due to correlations with the other predictors.

Table 3. Variance Inflation Factors (VIFs).

Variable	European Banks	U.S. Banks
ESG Score	1.85	1.72
Size (log assets)	1.92	1.85
Capital Ratio	1.78	1.69
Loan-to-Deposit Ratio	1.45	1.38
GDP Growth	1.32	1.28
Interest Rate	1.21	1.19
Mean VIF	1.59	1.52

reports results of VIFs Test.

The variance inflation factors (VIFs) for all variables are well below the conventional threshold of 5 (and substantially below 10), with mean VIFs of 1.59 for European banks and 1.52 for U.S. banks. These results confirm that multicollinearity is not a concern in our analysis and that the estimated coefficients can be interpreted reliably.

3.2. Econometric Methodology: PSTR Model

3.2.1. Linearity Test

To verify whether the relationship between ESG and bank performance can be adequately captured by a linear specification, we begin by conducting a series of linearity tests prior to estimating the PSTR model. Specifically, we employ both the Lagrange Multiplier (LM) test and its finite-sample adjusted version, the F-test, as proposed in Gonzalez et al. (2005). These tests assess the null hypothesis of linearity against the alternative of a smooth transition between regimes. Rejecting the linearity hypothesis indicates that the impact of ESG varies across different levels of sustainability engagement and therefore justifies the use of a nonlinear framework such as the PSTR model. This step is essential to ensure that the dynamics of ESG effects are not incorrectly constrained to be constant across banks and over time. To improve the reliability of statistical inference, heteroskedasticity-robust standard errors clustered at the bank level were employed throughout the empirical analysis. Clustering at the bank level accounts for potential serial correlation and heteroskedasticity within individual banks over time. Results are reported in

Table 4. Linearity Test.

Banks	European Banks		U.S. Banks
Test Type	Test Statistic	p-Value	Test Statistic	p-Value
LM test	30.12	0.000	26.54	0.000
F test	13.58	0.000	11.93	0.000

.

Both the LM and F tests strongly reject the null hypothesis of linearity, confirming that the impact of ESG on ROA, ROE, and Z-score is nonlinear in European and U.S. banks. The null hypothesis is rejected across all specifications, indicating that the marginal effect of ESG in the high-ESG regime is significantly different from that observed in the low-ESG regime. These findings provide direct statistical support for the existence of regime-dependent ESG effects.

3.2.2. Selecting Number of Regimes

To determine the appropriate number of transition regimes in the PSTR specification, we conduct a sequential test following the procedure of Gonzalez et al. (2005). The test evaluates whether additional regimes significantly improve the model’s explanatory power.

Table 5. Number-of-Regimes Test.

Banks	European Banks		U.S. Banks
Test	Test Statistic	p-Value	Test Statistic	p-Value
${\mathit{H}}_{\mathbf{0}}\mathbf{:}\mathit{r}\mathbf{=}\mathbf{1}$ vs. ${\mathit{H}}_{\mathbf{1}}\mathbf{:}\mathit{r}\mathbf{=}\mathbf{2}$	3.65	0.125	3.92	0.135
${\mathit{H}}_{\mathbf{0}}\mathbf{:}\mathit{r}\mathbf{=}\mathbf{2}$ vs. ${\mathit{H}}_{\mathbf{1}}\mathbf{:}\mathit{r}\mathbf{=}\mathbf{3}$	1.72	0.210	1.88	0.225

reports the results for European and U.S. banks. For both samples, the null hypothesis of a single regime cannot be rejected in favor of two regimes, and similarly, the test for moving from two to three regimes also fails to reject the null. The p-values associated with each step exceed conventional significance thresholds, indicating that the introduction of additional transition functions does not provide statistically meaningful improvements. These results suggest that a single-regime PSTR model is sufficient to capture the nonlinear dynamics of the ESG–performance relationship in both European and U.S. banking sectors.

A single transition function (2 regimes: low vs. high ESG) is sufficient to model the nonlinear effects of ESG on European and U.S. banks. Introducing additional regimes does not significantly improve model fit, keeping the PSTR model parsimonious and interpretable.

3.2.3. PSTR Estimation Results

Table 6. Impact of the Aggregate ESG Score on Bank Performance and Stability in European Banks.

	Dependent Variable: ROA		Dependent Variable: Z-Score		Dependent Variable: ROE
Regimes	Low-ESG Regime β0	High-ESG Regime β0 + β1	Low-ESG Regime β0	High-ESG Regime β0 + β1	Low-ESG Regime β0	High-ESG Regime β0 + β1
ESG Score	0.0013 (0.0010)	0.0056 *** (0.0013)	0.092 ** (0.038)	0.231 *** (0.043)	0.0025 (0.0019)	0.0102 *** (0.0029)
Size (log assets)	0.0141 ** (0.0052)	0.0172 *** (0.0050)	0.035 *** (0.011)	0.033 *** (0.010)	0.028 *** (0.0063)	0.031 *** (0.0060)
Capital Ratio	0.0325 *** (0.0065)	0.0412 *** (0.0063)	0.145 *** (0.024)	0.189 *** (0.022)	0.062 *** (0.0148)	0.078 *** (0.0140)
Loan-to-Deposit Ratio	−0.0027 (0.0019)	−0.0016 (0.0017)	−0.008 ** (0.004)	−0.005 ** (0.003)	−0.003 (0.0030)	−0.002 (0.0028)
GDP Growth	0.0341 *** (0.0111)	0.0285 *** (0.0106)	0.045 ** (0.020)	0.039 ** (0.018)	0.042 *** (0.0116)	0.039 *** (0.0112)
Interest Rate	−0.0039 (0.0035)	−0.0027 (0.0034)	−0.005 * (0.003)	−0.035 (0.003)	−0.015 * (0.0075)	−0.009 (0.0061)
γ (Smoothness)	8.9		7.6		9.7
c (Threshold ESG)	54.8		53.1		55.6

The numbers in parentheses are robust standard errors clustered at the bank level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level.

reports the results for European banks, showing estimated coefficients and significance levels across different ESG regimes, while

Table 7. Impact of the Aggregate ESG Score on Bank Performance and Stability in U.S. banks.

	Dependent Variable: ROA		Dependent Variable: Z-Score		Dependent Variable: ROE
Regimes	Low-ESG Regime β0	High-ESG Regime β0 + β1	Low-ESG Regime β0	High-ESG Regime β0 + β1	Low-ESG Regime β0	High-ESG Regime β0 + β1
ESG Score	0.0015 (0.0009)	0.0062 *** (0.0015)	0.090 ** (0.038)	0.225 *** (0.046)	0.0027 (0.0011)	0.0101 *** (0.0028)
Size (log assets)	0.0142 * (0.0053)	0.0168 *** (0.0050)	0.034 *** (0.011)	0.032 *** (0.009)	0.027 *** (0.0026)	0.030 *** (0.0049)
Capital Ratio	0.0318 * (0.0066)	0.0405 * (0.0064)	0.148 * (0.025)	0.185 * (0.023)	0.060 * (0.0145)	0.076 * (0.0138)
Loan-to-Deposit Ratio	−0.0028 (0.0019)	−0.0017 (0.0018)	−0.008 ** (0.004)	−0.005 * (0.003)	−0.003 (0.0031)	−0.002 (0.0029)
GDP Growth	0.0345 *** (0.0112)	0.0288 *** (0.0108)	0.047 ** (0.021)	0.039 ** (0.019)	0.042 *** (0.0116)	0.039 *** (0.0113)
Interest Rate	−0.0040 (0.0036)	−0.0028 (0.0033)	−0.005 * (0.003)	−0.003 (0.003)	−0.009 * (0.0057)	−0.006 (0.0053)
γ (Smoothness)	8.6		7.8		9.5
c (Threshold ESG)	55.1		53.5		55.4

The numbers in parentheses are robust standard errors clustered at the bank level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level.

provides the corresponding results for U.S. banks. Comparing the results across Europe and the U.S. offers a transatlantic perspective on how ESG initiatives translate into profitability and risk mitigation under different regulatory and market environments. These tables collectively highlight the thresholds at which ESG practices begin to generate tangible benefits for banks in each region. To account for potential heteroskedasticity and within-bank serial correlation, all standard errors are computed using heteroskedasticity-robust estimators clustered at the bank level. Our results should be interpreted as evidence of a robust nonlinear association between ESG performance and bank outcomes, with an endogenously estimated threshold. Future research using quasi-experimental designs or instrumental variables would be valuable to complement our analysis.

The PSTR estimation results provide strong evidence of nonlinear effects of ESG performance on bank profitability and stability, confirming that the benefits of ESG are regime-dependent rather than uniform across all levels of engagement. Consistent with theoretical expectations (Goss & Roberts, 2011; Cheng et al., 2014; Cornett et al., 2016), control variables perform as anticipated: larger banks exhibit higher ROA, reflecting their capacity to leverage ESG initiatives effectively, while higher capital ratios positively influence profitability. Macroeconomic conditions, proxied by GDP growth, also contribute positively to bank performance, whereas interest rates exert a weak, statistically insignificant negative effect. The economic significance of crossing the ESG threshold is substantial. For European banks, moving from the low-ESG to high-ESG regime is associated with an increase in ROA of approximately 0.43 percentage points (from 0.0013 to 0.0056). For a bank with average total assets of €50 billion, this implies an additional annual profit of approximately €215 million. The increase in the Z-score of 0.139 points (from 0.092 to 0.231) represents a significant enhancement in financial resilience, reducing the estimated probability of insolvency by approximately 28% based on the Z-score’s inverse relationship with insolvency probability.

The results suggest that ESG engagement does not generate immediate financial benefits at low levels of sustainability commitment. Instead, ESG effects become economically and statistically significant only after banks achieve a sufficient level of ESG maturity. This finding supports the existence of threshold effects and highlights the importance of long-term ESG integration strategies.

The stronger profitability effects observed for U.S. banks may reflect stronger market-based incentives and institutional investor pressure, whereas European banks appear to benefit relatively more in terms of financial stability due to stricter ESG regulatory frameworks and more harmonized sustainability disclosure requirements.

For profitability, ESG scores show no significant impact in the low-ESG regime (β₀ = 0.0013 for European banks; β₀ = 0.0014 for U.S. banks), but their effect becomes positive and highly significant in the high-ESG regime (β₀ + β₁ = 0.0056 for Europe; 0.0062 for the U.S.), highlighting a threshold effect consistent with prior studies emphasizing nonlinear ESG–performance relationships (Buallay, 2019; Bitar et al., 2018). The estimated thresholds for profitability (53.1–54.8 for European banks; 54.2–55.3 for U.S. banks) indicate that only 35–42% of banks in our sample currently operate above the critical ESG threshold, depending on the performance measure and region. This implies that the majority of banks (58–65%) remain in the low-ESG regime, where the marginal effects of sustainability engagement on financial outcomes are statistically insignificant. This suggests that incremental ESG improvements below the threshold yield limited financial gains, while crossing the threshold generates substantial and statistically significant improvements in profitability and stability. The smoothness parameters (γ ≈ 8–10) indicate relatively rapid transitions between regimes. Based on average annual ESG score improvements of approximately 1.5–2.0 points per year, banks below the threshold would require approximately 2.5–3.5 years to cross the critical threshold. This suggests that ESG investments represent a medium-term strategic commitment rather than a short-term tactical decision. Therefore, the identified nonlinearity should be interpreted as a sharp but continuous transition rather than a perfectly gradual adjustment process. Despite this relatively high transition speed, the PSTR framework remains useful because it estimates the threshold endogenously and allows regime-specific marginal effects to emerge directly from the data.

Similarly, Z-score results demonstrate that ESG enhances bank stability, with stronger effects in the high-ESG regime. The slightly lower thresholds for stability (c = 53.1 for European banks; c = 53.5 for U.S. banks) suggest that improvements in solvency and resilience begin at moderately lower ESG levels, yet still require substantive commitment. Notably, the proportion of banks above the threshold is consistently higher for profitability measures (ROA/ROE) than for stability (Z-score), suggesting that financial stability benefits require slightly higher ESG maturity levels, a finding consistent with the higher threshold estimates observed for the Z-score (55.6 for Europe; 56.1 for the U.S.). These findings align with evidence that strong governance and environmental practices reduce risk-taking, improve credit quality, and enhance resilience to adverse shocks (Nollet et al., 2016; Capelle-Blancard & Petit, 2019; D’Orazio & Popoyan, 2019). The sharp regime transition observed for stability (γ ≈ 7.6 for Europe; γ ≈ 7.8 for the U.S.) further confirms the nonlinear nature of ESG’s contribution to risk mitigation.

Overall, the results underscore that ESG effects in banking are highly contingent on the level of engagement. Substantive ESG adoption is necessary to unlock profitability and stability gains, with U.S. banks experiencing slightly higher returns in profitability, possibly reflecting stronger market pressures and voluntary disclosure practices, while European banks exhibit marginally stronger stabilizing effects, likely due to more stringent regulatory guidance (European Commission, 2023; Busch & Friede, 2018). These findings provide empirical support for the view that ESG is not only a tool for corporate responsibility but also a strategic lever for financial performance and risk management when implemented with sufficient depth and maturity (Weber, 2017; Friede et al., 2015).

The cross-regional differences observed in our analysis can be understood through the lens of institutional theory. The European Union’s comprehensive regulatory framework for sustainable finance—including the SFDR, EU Taxonomy, and CSRD—creates a harmonized environment that reduces information asymmetry and enhances the credibility of ESG commitments. This regulatory certainty may explain why European banks experience larger stability gains: the regulatory framework provides clear guidelines for risk management and disclosure, enabling banks to more effectively translate ESG practices into reduced risk exposure.

In contrast, the U.S. institutional environment is characterized by greater market-based incentives and investor pressure. The absence of harmonized national ESG regulations means that ESG engagement is primarily driven by investor demands, voluntary disclosure standards (SASB, TCFD), and market competition. This market-driven approach may place greater emphasis on financial performance and shareholder value, potentially explaining why U.S. banks achieve slightly higher profitability gains from ESG maturity. The political contestation surrounding ESG in the U.S. may also create stronger selective pressures, where banks that commit to ESG must demonstrate clear financial benefits to justify their engagement.

These findings confirm that the ESG–performance relationship is characterized by significant nonlinearities and regime-switching dynamics, with banks needing to achieve a minimum level of ESG maturity before financial benefits materialize. Having established the existence and statistical significance of nonlinear threshold effects in the ESG–performance relationship, we now turn to a detailed examination of their economic magnitude and practical implications.

Table 8. Estimated Transition Parameters.

Dependent Variable	Threshold (c)	Smoothness (γ)	Transition Speed	% of Banks Above Threshold
European Banks
ROA	53.1	8.5	2.8 years	42.3%
ROE	54.8	9.2	2.6 years	38.7%
Z-score	55.6	10.1	2.4 years	35.1%
U.S. Banks
ROA	54.2	7.8	3.1 years	39.5%
ROE	55.3	8.5	2.8 years	36.2%
Z-score	56.1	9.4	2.5 years	32.8%

Note: Transition speed is calculated as the time (in years) required for a bank with average ESG improvement (1.5–2.0 points per year) to move from g = 0.1 to g = 0.9.

reports the estimated transition parameters, including the threshold locations, smoothness coefficients, implied transition speeds (in years), and the percentage of banks in the high-ESG regime, thereby providing a comprehensive quantitative basis for interpreting the economic relevance of our findings.

The estimated smoothness parameters (γ) ranging from 7.8 to 10.1 indicate a relatively rapid transition between regimes. To illustrate the practical meaning of these values, consider a bank with an ESG score of 50 (below the threshold). Using our estimated parameters for the ROA specification (γ = 8.5, c = 53.1), the transition probability is:

• At ESG = 50: g = 1/[1 + exp(−8.5 × (50 − 53.1))] = 1/[1 + exp(26.35)] ≈ 0.00;
• At ESG = 53.1 (threshold): g = 0.50;
• At ESG = 56: g = 1/[1 + exp(−8.5 × (56 − 53.1))] = 1/[1 + exp(−24.65)] ≈ 0.99.

This demonstrates that the transition from the low-ESG regime to the high-ESG regime occurs over a relatively narrow range of approximately 3 ESG points (from 53 to 56). Given that banks typically improve their ESG scores by 1.5–2.0 points per year (based on our sample’s observed annual changes), this implies that banks near the threshold can cross it within approximately 1.5 to 2 years. This finding has important managerial implications: ESG investments should be viewed as a strategic commitment with a relatively short payback period once the threshold is approached. To formally test whether the estimated coefficients differ significantly between European and U.S. banks, we conducted a series of Wald tests for coefficient equality across regions. The test statistics and p-values are reported in

Table 9. Cross-Regional Coefficient Equality Tests (Wald Tests): Results for Aggregate ESG.

Variable Tested	Regime	Wald Statistic	p-Value	Significant Difference?
ESG → ROA	Low-ESG	0.42	0.517	No
ESG → ROA	High-ESG	4.18	0.041	Yes (USA > Europe)
ESG → Z-score	Low-ESG	0.58	0.446	No
ESG → Z-score	High-ESG	2.10	0.147	No
ESG → ROE	Low-ESG	0.39	0.532	No
ESG → ROE	High-ESG	3.95	0.047	Yes (USA > Europe)

Notes: The null hypothesis is equality of coefficients between European and U.S. banks. Tests are conducted separately for each variable and each regime using a Wald test. Significant results (p < 0.10) are shown in bold. “Marginal” indicates significance at the 10% level.

.

The Wald test results for the aggregate ESG score reveal that in the low-ESG regime, the differences between European and U.S. banks are not statistically significant for all performance indicators (ROA: Wald = 0.42, p = 0.517; ROE: Wald = 0.39, p = 0.532; Z-score: Wald = 0.58, p = 0.446). This indicates that at low levels of ESG engagement, the marginal effects are comparable across both banking systems, which is consistent with the absence of significant ESG impacts in this regime. In the high-ESG regime, however, formal tests confirm significant differences for profitability measures. For ROA, the Wald statistic is 4.18 (p = 0.041), and for ROE, it is 3.95 (p = 0.047), both rejecting the null hypothesis of equality at the 5% significance level. These results statistically validate our earlier visual observation that U.S. banks experience slightly higher profitability gains from ESG once they surpass the critical threshold. In contrast, for the Z-score (financial stability), the Wald test yields a non-significant result (Wald = 2.10, p = 0.147), suggesting that the stability-enhancing effects of ESG are similar across both regions.

3.2.4. Robustness Checks

To further validate the robustness of our findings, we complement the baseline analysis conducted using the aggregate ESG score by re-estimating all models with the three underlying pillars examined separately. This additional step allows us to assess whether the environmental (E), social (S), and governance (G) dimensions exert heterogeneous effects on bank performance and stability. By isolating each component, we can better identify which specific sustainability drivers are most influential and determine whether the aggregate ESG score masks component-level dynamics. This disaggregated approach therefore strengthens the reliability of our results and provides a more nuanced understanding of how sustainability practices shape banking outcomes. The test statistics are shown in

Table 10. Disaggregated Effects of Environmental, Social, and Governance Components on Bank Performance and Stability in European Banks.

Variable	Regime	ROA	Z-Score	ROE
Environmental (E)	Low-ESG	0.0011 (0.0011)	0.086 ** (0.036)	0.0023 (0.0018)
	High-ESG	0.0061 * (0.0015)	0.218 * (0.042)	0.0098 * (0.0027)
Social (S)	Low-ESG	0.0015 (0.0012)	0.079 * (0.039)	0.0030 (0.0019)
	High-ESG	0.0048 (0.0016)	0.204 * (0.044)	0.0084 (0.0028)
Governance (G)	Low-ESG	0.0020 (0.0010)	0.096 ** (0.033)	0.0035 (0.0016)
	High-ESG	0.0073 * (0.0014)	0.235 * (0.041)	0.0108 * (0.0026)
Size (log assets)		0.0138 ** (0.0053)	0.033 *** (0.011)	0.027 *** (0.0062)
Capital Ratio		0.0312 * (0.0067)	0.142 * (0.023)	0.061 * (0.0142)
Loan-to-Deposit		−0.0028 (0.0019)	−0.007 ** (0.004)	−0.003 (0.0032)
GDP Growth		0.0345 *** (0.0110)	0.046 ** (0.020)	0.042 *** (0.0115)
Interest Rate		−0.0040 (0.0036)	−0.005 * (0.003)	−0.008 * (0.0055)
γ (smoothness)		8.5	7.7	9.6
c (threshold ESG)		54.5	53.0	55.8

The numbers in parentheses are robust standard errors clustered at the bank level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level.

and

Table 11. Disaggregated Effects of Environmental, Social, and Governance Components on Bank Performance and Stability in U.S. banks.

Variable	Regime	ROA	Z-Score	ROE
Environmental (E)	Low-ESG	0.0012 (0.0012)	0.089 ** (0.037)	0.0024 (0.0019)
	High-ESG	0.0063 * (0.0016)	0.222 * (0.043)	0.0101 * (0.0028)
Social (S)	Low-ESG	0.0014 (0.0013)	0.081 * (0.038)	0.0031 (0.0020)
	High-ESG	0.0050 (0.0017)	0.206 * (0.045)	0.0085 (0.0029)
Governance (G)	Low-ESG	0.0017 (0.0011)	0.094 ** (0.034)	0.0037 (0.0017)
	High-ESG	0.0071 * (0.0015)	0.231 * (0.042)	0.0110 * (0.0027)
Size (log assets)		0.0145 ** (0.0054)	0.034 *** (0.011)	0.027 *** (0.0063)
Capital Ratio		0.0305 * (0.0068)	0.145 * (0.024)	0.059 * (0.0145)
Loan-to-Deposit		−0.0029 (0.0020)	−0.008 ** (0.004)	−0.003 (0.0031)
GDP Growth		0.0342 *** (0.0113)	0.048 ** (0.021)	0.043 *** (0.0117)
Interest Rate		−0.0039 (0.0037)	−0.005 * (0.003)	−0.008 * (0.0056)
γ (smoothness)		8.8	7.9	9.4
c (threshold ESG)		55.3	53.8	55.6

The numbers in parentheses are robust standard errors clustered at the bank level. *, **, and *** indicate statistical significance at the 10%, 5%, and 1% level.

.

The results show that, for all three ESG pillars, the effects on bank profitability (ROA, ROE) and stability (Z-score) are distinctly nonlinear, with significant positive impacts emerging only once each pillar surpasses a threshold of roughly 53.0–55.8 points. Among the pillars, Governance exerts the strongest influence, followed by Environmental and then Social factors. This hierarchy reflects governance’s pivotal role in shaping risk management practices, internal control systems, and strategic oversight within banks, which directly translates into improved financial outcomes. The Environmental pillar also shows substantial effects, particularly for stability (Z-score), likely due to reduced exposure to climate-related risks and stranded assets. Social factors, while still significant, exhibit slightly smaller coefficients, possibly because their financial benefits materialise over longer horizons or are more context-dependent.

The control variables behave in line with theoretical expectations. Larger institutions (proxied by Size) and those with stronger capital positions (Capital Ratio) tend to exhibit better performance and stability across all pillars and regimes. The positive and significant coefficient of Capital Ratio confirms that well-capitalised banks are more resilient to adverse shocks and enjoy higher profitability. In contrast, elevated loan-to-deposit ratios are associated with lower stability, reflecting greater liquidity risk and potential funding fragility. Macroeconomic conditions, captured by GDP growth, contribute positively to bank performance, while interest rates exert a weak and generally insignificant negative effect.

Moreover, the estimated transition smoothness parameters (γ) range between approximately 7.7 and 9.6 across pillars and regions, suggesting relatively sharp regime shifts around the identified ESG thresholds. This finding reinforces the existence of clear ESG maturity thresholds required for banks to unlock measurable financial benefits. The sharpness of the transition implies that incremental improvements in ESG scores below the threshold yield limited financial gains; only once banks cross the critical threshold do the benefits materialise more substantially.

Comparing the two regions, European banks exhibit slightly lower thresholds for stability (c ≈ 52.8–53.0) compared to U.S. banks (c ≈ 53.5–53.8), suggesting that the stability-enhancing effects of ESG emerge at moderately lower levels of engagement in Europe. This difference may be attributed to the more stringent and harmonised regulatory framework in the European Union, which encourages even moderately committed banks to adopt risk-mitigating ESG practices. Conversely, U.S. banks show marginally higher profitability thresholds (c ≈ 55.1–55.6) but also larger coefficient gains in the high-ESG regime, consistent with stronger market-based incentives and institutional investor pressure.

Finally, the Governance pillar consistently displays the highest threshold values (c ≈ 55.2–55.8) and the steepest transition parameters (γ ≈ 9.4–10.1), indicating that governance improvements need to reach a relatively high maturity level before their full financial benefits are realised. Once this level is attained, however, the payoff is substantial, reinforcing the view that governance quality is a foundational enabler of sustainable bank performance.

To formally test whether the estimated coefficients differ significantly between European and U.S. banks, we conducted a series of Wald tests for coefficient equality across regions. The test statistics and p-values are reported in

Table 12. Cross-Regional Coefficient Equality Tests (Wald Tests): Results for Disaggregated Pillars (E, S, and G).

Variable Tested	Regime	Wald Statistic	p-Value	Significant Difference?
Panel A: Environmental (E)
E → ROA	Low-ESG	0.51	0.475	No
E → ROA	High-ESG	3.87	0.049	Yes (USA > Europe)
E → Z-score	Low-ESG	0.62	0.431	No
E → Z-score	High-ESG	1.95	0.163	No
E → ROE	Low-ESG	0.44	0.507	No
E → ROE	High-ESG	3.62	0.057	Marginal
Panel B: Social (S)
S → ROA	Low-ESG	0.33	0.566	No
S → ROA	High-ESG	2.91	0.088	Marginal
S → Z-score	Low-ESG	0.47	0.493	No
S → Z-score	High-ESG	1.68	0.195	No
S → ROE	Low-ESG	0.38	0.538	No
S → ROE	High-ESG	2.75	0.097	Marginal
Panel C: Governance (G)
G → ROA	Low-ESG	0.55	0.458	No
G → ROA	High-ESG	5.02	0.025	Yes (USA > Europe)
G → Z-score	Low-ESG	0.71	0.399	No
G → Z-score	High-ESG	2.34	0.126	No
G → ROE	Low-ESG	0.48	0.488	No
G → ROE	High-ESG	4.81	0.028	Yes (USA > Europe)

Notes: The null hypothesis is equality of coefficients between European and U.S. banks. Tests are conducted separately for each variable and each regime using a Wald test. Significant results (p < 0.10) are shown in bold. “Marginal” indicates significance at the 10% level.

.

The governance component exhibits the strongest and most robust cross-regional differences. In the high-ESG regime, the Wald tests reject the null hypothesis for both ROA (Wald = 5.02, p = 0.025) and ROE (Wald = 4.81, p = 0.028), confirming that governance improvements translate into significantly higher profitability gains for U.S. banks compared to their European counterparts once governance quality reaches a sufficient maturity level. This finding likely reflects the stronger market discipline and shareholder activism prevalent in the U.S. banking environment.

For the environmental pillar, the Wald test reveals a marginally significant difference for ROA in the high-ESG regime (Wald = 3.87, p = 0.049), while the difference for ROE is borderline significant (Wald = 3.62, p = 0.057). These results suggest that U.S. banks may also benefit slightly more from environmental policies, possibly due to greater investor scrutiny of climate-related risks in the U.S. market. However, for the Z-score, the differences remain insignificant (p = 0.163). Interestingly, the social pillar displays the weakest cross-regional differences. The Wald tests for the high-ESG regime yield p-values of 0.088 for ROA, 0.097 for ROE, and 0.195 for the Z-score, indicating that the social dimension of ESG affects European and U.S. banks in a broadly similar manner. This homogeneity may reflect the universal nature of social factors (such as community relations and employee welfare), which are less dependent on regional regulatory specificities.

While our disaggregated analysis reveals important heterogeneity across the environmental, social, and governance pillars, several considerations warrant acknowledgment. First, the aggregate ESG score remains a useful and parsimonious metric for regulatory benchmarking and investor decision-making, as it reflects overall sustainability maturity and is widely employed in practice (Berg et al., 2022). However, our findings suggest that aggregation conceals differential effects: governance improvements yield the most immediate financial benefits, while environmental practices primarily affect long-term risk exposure. Second, the three pillars are not independent; strong governance often facilitates better environmental and social performance, and social practices can reinforce governance mechanisms (Eccles et al., 2014). Future research could fruitfully explore the interactions and potential complementarities among these pillars, as well as their differential effects across bank types (e.g., systemically important vs. regional banks) and over different time horizons. Third, while we treat the three pillars symmetrically in terms of threshold dynamics, future work could investigate whether each pillar exhibits distinct threshold levels and transition speeds, which would offer more granular guidance for bank managers and policymakers.

4. Conclusions and Policy Recommendations

This study investigates the impact of ESG performance on the profitability and financial stability of European and U.S. banks using a PSTR model. The empirical evidence reveals a clear nonlinear relationship, whereby ESG engagement enhances bank performance only after surpassing a critical ESG threshold. Banks with mature ESG practices enjoy significantly higher returns on assets and equity, as well as greater resilience to financial distress, compared to those with lower ESG scores. These results strongly suggest that ESG investments in European and U.S. banks do not produce uniform financial benefits. Instead, ESG’s positive effects on profitability and stability emerge only after banks reach a critical ESG engagement threshold. This nonlinear dynamic has important implications: For bank management, it stresses the need for comprehensive and consistent ESG strategies rather than piecemeal efforts to realize tangible financial gains. For investors, it signals that ESG disclosures and scores should be evaluated carefully, recognizing that marginal ESG improvements below the threshold may not translate into higher returns or lower risks. For regulators and policymakers, the findings highlight the value of supporting banks in overcoming barriers to ESG maturity, such as through incentives, guidance, and harmonized reporting standards.

Finally, while our PSTR framework endogenously estimates the transition threshold and allows for smooth regime changes, we acknowledge that the proportion of banks above or below the threshold, as well as the speed of transition between regimes, is estimated based on the sample period 2010–2022. The distribution may shift over time as regulatory frameworks evolve and ESG practices become more widespread. Future research could extend this analysis by examining how the threshold distribution changes over time, particularly in response to major regulatory reforms such as the full implementation of the EU Taxonomy and CSRD in Europe, or to shifts in U.S. climate policy.