ESG and Firm Performance: The Moderating Role of Firm-Level Financial Stability

Abstract

This study examines how environmental, social, and governance (ESG) influences firm performance and assesses whether firm-level financial stability moderates this relationship. A sample of 272 companies from six countries in ASEAN over the period of 2020–2024 is investigated using the Bayesian regression. The findings reveal a positive relationship between ESG and firm market performance, suggesting that stronger sustainability practices tend to enhance market values. Firm-level financial stability amplifies this association, indicating that financially stable firms are more capable of converting ESG initiatives into sustainable long-term values. Among the three ESG dimensions, the environmental pillar exhibits the clearest positive association with firm performance. However, the moderating effect of firm-level financial stability is more evident for social and governance pillars, implying that financially stable firms derive greater value from social and governance practices. The study underscores the importance of aligning ESG strategies with firm-level financial stability to foster sustainable performance.

1. Introduction

Environmental, social, and governance (ESG) practices have become an increasingly important component of corporate strategy and financial decision-making [1]. Firms are under growing pressure from investors, regulators, and other stakeholders to address environmental and social concerns and maintain effective governance [2]. This creates a question of whether ESG activities translate into improved firm performance.

This question has received extensive empirical attention, yet the evidence remains inconclusive. While a substantial body of research suggests that stronger ESG performance improves profitability [1,3], other studies report weak, insignificant [2], or negative effects [4]. One possible explanation for these divergent findings is that the economic benefits of ESG are not certain. ESG engagement often requires firms to commit financial resources that may generate benefits only over an extended period. As such, whether these benefits are economically significant enough to be reflected in performance and market valuation may depend on firms’ ability to sustain ESG investments and convert them into valuable organizational outcomes.

This study first examines the relationship between ESG and firm performance, proxied by Tobin’s Q. Tobin’s Q measures the market’s valuation of a firm’s assets relative to their replacement cost and therefore reflects investors’ expectations regarding the firm’s future growth opportunities, risk, and ability to generate economic returns [5]. This measure is particularly relevant as many of the potential benefits of ESG practices are forward-looking. For example, green environmental practices may not immediately appear in accounting earnings but may influence investors’ expectations about the firm’s future cash flows and valuations.

In this study, ESG should be understood not simply as a set of responsible activities but as a form of strategic investment that can generate firm-specific capabilities. Stakeholder theory suggests that ESG practices can strengthen relationships with stakeholders, thereby supporting the cooperation required for long-term value creation. From a transaction cost perspective, effective ESG practices may reduce frictions associated with monitoring, coordination, contracting, and opportunistic behavior. The resource-based view further suggests that sustained ESG engagement can enable firms to develop organizational capabilities that are difficult for competitors to replicate. These mechanisms can enhance firms’ future economic prospects and reduce uncertainty, contributing to improved performance.

However, the development of ESG-related capabilities requires sustained financial commitment [6]. The economic benefits of these investments are therefore unlikely to be realized immediately [7]. This study thus argues that firm-level financial stability represents an important condition in the ESG-firm performance relationship. Financially stable firms are better able to absorb the short-term costs associated with ESG investments and maintain their commitment to such activities. In contrast, firms facing financial distress may be forced to prioritize short-term liquidity and survival, limiting their ability to sustain ESG investments or fully develop the organizational capabilities associated with them. Organizational slack theory provides a basis for this argument.

It should be noted that the role of firm-level financial stability may differ across the three ESG dimensions [3]. Environmental, social, and governance practices involve different resource requirements and value-creation mechanisms. Consequently, the extent to which firm-level financial stability facilitates the realization of ESG-related benefits may vary across the environmental, social, and governance pillars. In addition to examining the aggregate ESG score, this study therefore separately investigates the environmental, social, and governance dimensions and their interactions with firm-level financial stability.

This study contributes to the ESG literature in three ways. First, it provides further evidence on whether ESG is associated with firm performance. Second, this study moves beyond the conventional examination of ESG-performance relationship. Its principal contribution is to investigate whether financial stability conditions firms’ ability to translate ESG into firm value, rather than treating financial health as merely another firm characteristic. Third, by separating the environmental, social, and governance dimensions, the study provides insights into whether firm-level financial stability has the same moderating effect on the value implications of different ESG activities. Fourth, the Bayesian framework allows the study to account for heterogeneity across firms, industries, countries and years while directly quantifying posterior uncertainty. These provide a more context-sensitive assessment of the relationship between ESG and firm performance than studies focusing exclusively on the average direct effects.

2. Literature Review and Hypotheses

2.1. ESG and Firm Performance

Environmental, social, and governance (ESG) has become an increasingly important strategic consideration for firms seeking to achieve sustainable competitive advantage [8] and respond to growing expectations from various stakeholders [2]. ESG has traditionally been viewed as a corporate responsibility; however, firms increasingly integrate ESG into strategic decision-making because of its potential to improve long-term business performance. Despite extensive empirical research, the relationship between ESG and firm performance remains inconclusive [9,10,11,12]. While many studies report positive effects [13,14,15], others find insignificant or even negative relationships [16,17], suggesting that ESG may not be automatically associated with superior financial performance.

Stakeholder theory serves as a key lens for explaining how ESG influences firm performance [13]. The theory argues that firms create value sustainably by managing relationships with stakeholders [18]. In other words, the long-term existence and success of organizations depend on the support of both shareholders and stakeholders, including employees, customers, suppliers, investors, regulators, and local communities [19]. ESG strengthens these relationships, as it shows that a firm recognizes and addresses the interests of multiple stakeholders through responsible corporate behaviors [20]. As such, firms are more likely to benefit from stronger employee commitment [21] and reduced conflict with regulators and local communities. These create conditions that enhance organizational performance.

However, ESG should not be considered a collection of sustainability activities, but a strategic organizational investment [22]. Similar to investments in innovation, digital transformation, or organizational capability, ESG requires substantial financial resources [6], managerial commitment [23], and organizational coordination before financial benefits can be realized. In other words, investments in cleaner technologies, employee development, governance systems, and compliance processes often increase operating expenditures in the short term with uncertain and delayed returns [24]. Consequently, the economic value of ESG should not be evaluated solely by whether firms adopt and disclose sustainability practices but by whether these practices improve firms’ ability to use resources more efficiently and generate sustainable profitability.

While stakeholder theory explains why ESG creates added values to organizations [13], it provides less insight into how stronger stakeholder relationships are converted into improved financial performance. Better stakeholder relationships do not necessarily increase profitability unless they improve firms’ operating efficiency or reduce the costs associated with conducting business. Transaction cost economics offers a complementary explanation by emphasizing that firms continuously incur costs when negotiating, monitoring, coordinating, and enforcing relationships with various stakeholders [25]. These transaction costs arise because economic exchanges are characterized by uncertainty, incomplete information, and the potential for opportunistic behavior [26].

Accordingly, ESG practices are believed to contribute to firm performance by reducing transaction costs. For example, firms with effective environmental management systems are less likely to incur penalties, litigation costs, or disruptions arising from environmental incidents [27]. Meanwhile, social initiatives strengthen relationships between firms and employees, customers, suppliers, and communities [28], which helps to reduce conflicts, labor turnover, contractual disputes, and coordination costs. Likewise, governance mechanisms improve board oversight, reduce agency conflicts, and strengthen internal monitoring, thereby reducing inefficiencies arising from opportunistic managerial behavior. Consequently, ESG contributes to improved profitability.

The resource-based view also offers a theoretical base for understanding the relationship between ESG and firm performance. The resource-based view conceptualizes firms as unique bundles of strategic resources [29], with ESG acting as an intangible reputational asset capable of creating competitive advantages [21]. Superior ESG performance reflects a firm’s organizational capability to effectively manage complex stakeholder relationships, thereby contributing to superior firm performance [21]. In addition, ESG-driven practices create valuable organizational capabilities by improving resource efficiency and operational effectiveness [7]. For example, green manufacturing practices reduce waste and optimize energy and material use [30], enabling firms to lower production costs and develop strategic capabilities that support sustained competitive advantages. Ultimately, ESG investment positively contributes to firm performance.

Based on these arguments, ESG is expected to improve firms’ performance. This leads to the formulation of the first hypothesis.

Hypothesis 1. 

ESG is positively associated with firm performance.

Treating ESG as a single construct may obscure the distinct impacts of its individual dimensions on firm performance. Environmental, social, and governance activities differ considerably in their strategic objectives, resource requirements, and channels of value creation. Examining these dimensions separately therefore provides a better understanding of how different ESG practices influence firms’ profitability.

For example, environmental practices often require substantial investment in cleaner technologies, resource-efficient production processes, and environmental management systems [31]. However, they may improve long-term profitability and resilience [32] by reducing energy consumption, minimizing waste, lowering regulatory compliance costs, and reducing the likelihood of environmental penalties [33]. Furthermore, effective environmental management reduces firms’ exposure to environmental uncertainty and operational disruptions, thereby making production activities more efficient. Consequently, environmental practices positively contribute to firm performance [10].

Hypothesis 1a. 

Environmental practices are positively associated with firm performance.

Unlike environmental initiatives that primarily improve production efficiency, social initiatives influence firm performance by strengthening cooperation among stakeholders. Investments in employee welfare, workplace safety, customer satisfaction, and community engagement [34] improve organizational commitments and reduce conflicts with both internal and external stakeholders. As performance depends on effective coordination among stakeholders, stronger social relationships can improve firms’ efficiency and ultimately increase firm performance [34].

Hypothesis 1b. 

Social practices are positively associated with firm performance.

Meanwhile, good governance concerning ownership structure, board independence, transparency, and disclosure quality [35] may reduce agency conflicts and improve strategic decision-making [36]. Compared with environmental and social initiatives, governance directly influences the efficiency with which firms allocate resources and monitor managerial behavior. Effective governance also facilitates the successful implementation of environmental and social initiatives by ensuring that these activities remain aligned with firms’ long-term strategic priorities. Consequently, good governance is associated with positive performance [37].

Hypothesis 1c. 

Governance practices are positively associated with firm performance.

2.2. The Moderating Role of Firm-Level Financial Stability

All the above theories suggest that ESG should improve firm performance but also imply that it may not automatically generate economic values [7]. ESG investments require sustained implementation before organizational capabilities can develop and transaction costs decline. Consequently, improvements in firm performance depend not only on ESG adoption and disclosure but also on firms’ ability to maintain long-term commitment to ESG initiatives. As ESG often requires significant resources [6], firms’ financial conditions may determine whether ESG investments can successfully build organizational capabilities that improve firm performance.

Organizational slack theory provides a useful framework for understanding this relationship. Organizational slack refers to excess resources beyond those required for current operations, which firms can allocate to discretionary activities such as ESG initiatives [38]. In fact, organizational slack provides firms with the flexibility to pursue strategic initiatives, such as ESG investments, that can strengthen competitive advantages [39] and firm performance [40]. In addition, slack resources serve as a buffer, enabling firms to sustain operations and maintain long-term strategic commitments under adverse conditions [39].

Building on the organizational slack theory, the effectiveness of ESG thus depends not only on firms’ willingness to invest but also on their ability to sustain these investments or their level of firm-level financial stability. In fact, a firm’s financial stability represents an important condition that enables firms to convert ESG investments into superior operating performance.

Accordingly, financially stable firms generally possess greater slack resources [38], allowing them to absorb the costs associated with ESG investments. These firms are therefore better positioned to capitalize on the value created by ESG. In contrast, financially burdened firms may face liquidity constraints that drive them to prioritize short-term financial performance over long-term sustainability investments, limiting their ability to fully realize the benefits of ESG practices. This study thus argues that firm-level financial stability strengthens the positive relationship between ESG and firm performance. This leads to the formulation of the second hypothesis.

Hypothesis 2. 

Firm-level financial stability positively moderates the relationship between ESG and firm performance.

The moderating role of firm-level financial stability may also extend to the associations between individual environmental, social, and governance dimensions and firm performance. This leads to the formulation of the three sub-hypotheses as follows.

Hypothesis 2a. 

Firm-level financial stability positively moderates the relationship between environmental practices and firm performance.

Hypothesis 2b. 

Firm-level financial stability positively moderates the relationship between social practices and firm performance.

Hypothesis 2c. 

Firm-level financial stability positively moderates the relationship between governance practices and firm performance.

3. Materials and Methods

3.1. Research Design

This study examines the relationship between ESG and firm performance, and whether the relationship is moderated by firm-level financial stability. Complementing the panel-regression approaches commonly used in prior ESG research, this study employs the Bayesian regression to provide a comprehensive view of the association between ESG and firm performance. The Bayesian approach is relevant as studies on ESG often involve complex uncertainty, heterogeneous firm characteristics, and unstable estimates from correlated ESG variables and interaction terms. The Bayesian framework incorporates prior information and produces full posterior distributions for model parameters [41], which allows reporting the probability that ESG has a positive impact on firm performance as well as the probability that firm-level financial stability may strengthen or weaken the effects.

In fact, the Bayesian framework provides several advantages over the traditional panel regression techniques. First, it allows direct probabilistic interpretations of model parameters, which is more informative than dichotomous significance testing [42]. Second, weakly informative priors can stabilize coefficient estimates and reduce instability caused by multicollinearity between ESG, firm-level financial stability, and their interaction terms. Finally, the Bayesian approach propagates uncertainty throughout the estimation process, producing credible intervals that describe a plausible range of parameter values rather than relying on asymptotic approximations [42].

This choice of Bayesian approach improves on earlier studies, which mainly focus on whether ESG coefficients are statistically significant. This method examines the likelihood, magnitude and uncertainty of ESG associations and considers whether these relations vary with firms’ financial conditions. The study thus contributes methodologically by providing a more flexible and probabilistic evaluation of the ESG-performance relationship, particularly in a context in which firm heterogeneity, firm-level financial stability and data limitations may compromise the reliability of traditional panel estimations.

3.2. Data Collection and Sampling

This study draws on firm-level data obtained from the Refinitiv Eikon database, one of the most comprehensive sources for standardized ESG and financial information. The initial dataset covers publicly listed companies across six ASEAN countries, including Singapore, Malaysia, Thailand, Indonesia, the Philippines, and Vietnam. To ensure comparability and consistency in ESG reporting, the sample is restricted to firms with complete ESG scores for five consecutive years within the period of 2020–2024. The initial dataset comprises 272 ASEAN firms observed over five years, providing a maximum of 1360 firm-year observations. Some observations contain missing values for the financial variables. The study does not impute missing values. Instead, it applies complete-case deletion by excluding any firm-year observation with a missing value for a variable included in the regression model. This procedure produces a final unbalanced estimation sample of 1124 firm-year observations from 272 firms. The sample is unbalanced across countries, reflecting differences in market size, disclosure practices, and ESG reporting in the region. Indonesia accounts for the largest proportion of firms in the dataset, followed by Thailand and Singapore, while the remaining countries contribute smaller shares. The uneven representation of the six ASEAN countries results from differences in the availability of complete ESG and financial data in Refinitiv rather than from intentional sampling choices.

The sample encompasses ESG disclosures and firm-level characteristics for Refinitiv-rated listed firms across six ASEAN countries, providing a robust empirical foundation for analyzing the ESG-performance relationship in emerging markets.

3.3. Variables and Measurements

This study employs a set of firm-level financial and ESG variables to investigate the relationship between sustainability practices and firm performance. To mitigate the influence of extreme outliers, log-transformations are applied for continuous variables where appropriate to reduce skewness.

3.3.1. Dependent Variables

Firm performance is a multidimensional construct that can be assessed using accounting-based, market-based, operational, and non-financial indicators [43]. Because our study investigates whether ESG engagement is recognized by capital-market participants and translated into firm value, we focus specifically on market-based firm performance, measured by Tobin’s Q. Tobin’s Q captures market-based valuation and reflects investors’ expectations about a firm’s future growth opportunities. Tobin’s Q is determined as the ratio between the market value of equity and debt over the book value of total assets. It captures the investors’ expectations about the future cash flows, growth opportunities, risk and intangible assets and other value-relevant resources. The forward-looking characteristic of this proxy is particularly relevant to ESG research as the benefits of ESG may gradually emerge. Tobin’s Q provides a more comprehensive measure of the value implications of ESG activities than traditional accounting metrics. In addition, Tobin’s Q is less susceptible to differences arising from accounting-policy choices and earnings management practices. In fact, Tobin’s Q is widely used in ESG research [44,45,46,47,48].

3.3.2. Independent Variables

ESG data were collected from the Refinitiv Eikon database, a widely adopted platform in empirical research. The ratings cover three dimensions, including environmental (E), social (S), and governance (G), with scores ranging from 0 to 100, based on publicly disclosed sustainability information. The combined ESG score aggregates these dimensions into a single metric, enabling comprehensive assessment of sustainability performance. Higher scores indicate stronger ESG practices, which are increasingly viewed as drivers of long-term value creation and risk mitigation. The Refinitiv measure relies substantially on corporate reporting; it reflects both the underlying ESG practices and the extent and quality of ESG disclosure. Firms with more developed reporting capabilities may therefore receive higher scores partly because their practices are more observable.

Firm-level financial stability (FS) is measured using Altman’s Z-score [49], a widely used indicator of firms’ financial health. Higher values of the Z-score indicate greater financial stability and lower likelihood of financial distress. In this study, firm-level financial stability is employed as a moderating variable to investigate whether firms with stronger financial stability can use ESG to improve their performance.

3.3.3. Control Variables

To account for factors influencing both firm performance and ESG, three control variables were included: firm size, sales growth, and financial leverage [12,48,50]. Firm size, measured as the logarithm of total assets (SIZE), reflects resource availability and is generally associated with superior financial and social outcomes [10,51]. Leverage (LEV), calculated as the ratio of total liabilities to total assets, captures solvency and risk exposure. While higher leverage can reduce financing costs, it may also constrain cash flow and increase operational risk [52]. Prior research suggests that firms with greater leverage often emphasize ESG improvements as a risk mitigation strategy [16]. Sales growth (SG) is also included in the model to control for firms’ underlying opportunities and operating momentum, which can independently affect firm performance [33,44].

3.4. Model Specifications

The baseline model examines the direct effect of ESG performance on firm performance. It aims to validate the influence of ESG and firm-level financial stability on firm performance.

$${\mathrm{FP}}_{\mathrm{i},\mathrm{t}}={{\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{ESG}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{SIZE}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{LEV}}_{\mathrm{i}.\mathrm{t} }+{\mathsf{\beta }}_5{\mathrm{SG}}_{\mathrm{i}.\mathrm{t} }+{\mathrm{v}}_{\mathrm{t}}+{\mathrm{w}}_{\mathrm{k}}+{\mathrm{z}}_{\mathrm{c}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(1)

where

• v_t: Year-level random effect: ${\mathrm{v}}_{\mathrm{t}}\sim \mathcal{N}\left(0,{\mathsf{\sigma }}_{\mathrm{year}}^2\right)$
• w_k: Industry-level random effect (for industry k): ${\mathrm{w}}_{\mathrm{k}}\sim \mathcal{N}\left(0,{\mathsf{\sigma }}_{\mathrm{ind}}^2\right)$
• z_c: Country-level random effect (for country c): ${\mathrm{z}}_{\mathrm{c}}\sim \mathcal{N}\left(0,{\mathsf{\sigma }}_{\mathrm{country}}^2\right)$
• ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$: Residual error term: ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}\sim \mathcal{N}\left(0,{\mathsf{\sigma }}_{\mathrm{e}}^2\right)$

The second model explores the impact of individual components of ESG, namely environmental (E), social (S), and governance (G), and financial stability on firm performance.

$${\mathrm{FP}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{E}}_{\mathrm{i}.\mathrm{t}}+{{{\mathsf{\beta }}_2\mathrm{S}}_{\mathrm{i}.\mathrm{t}}+{{\mathsf{\beta }}_3\mathrm{G}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{SIZE}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{LEV}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_7{\mathrm{SG}}_{\mathrm{i}.\mathrm{t}}+{\mathrm{v}}_{\mathrm{t}}+{\mathrm{w}}_{\mathrm{k}}+{\mathrm{z}}_{\mathrm{c}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(2)

The following interaction model is used to investigate the moderating effect of firm-level financial stability on the relationship between ESG and financial performance.

$${\mathrm{FP}}_{\mathrm{i},\mathrm{t}}={{\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{ESG}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{{{\mathsf{\beta }}_3\mathrm{ESG}}_{\mathrm{i}.\mathrm{t}}\ast \mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_4{\mathrm{SIZE}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{LEV}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{SG}}_{\mathrm{i}.\mathrm{t}}+{\mathrm{v}}_{\mathrm{t}}+{\mathrm{w}}_{\mathrm{k}}+{\mathrm{z}}_{\mathrm{c}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(3)

Model 4 jointly includes the environmental, social, and governance pillars and their respective interactions with firm-level financial stability. Each pillar coefficient and interaction is therefore interpreted conditional on the other ESG pillars and their interactions.

$$\begin{array}{c}{\mathrm{FP}}_{\mathrm{i},\mathrm{t}}={{\mathsf{\beta }}_0+{\mathsf{\beta }}_1\mathrm{E}}_{\mathrm{i}.\mathrm{t}}+{{{\mathsf{\beta }}_2\mathrm{S}}_{\mathrm{i}.\mathrm{t}}+{{\mathsf{\beta }}_3\mathrm{G}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{{\mathsf{\beta }}_5\mathrm{E}}_{\mathrm{i}.\mathrm{t}}\ast {\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{{\mathsf{\beta }}_6\mathrm{S}}_{\mathrm{i}.\mathrm{t}}\ast {\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{{\mathsf{\beta }}_7\mathrm{G}}_{\mathrm{i}.\mathrm{t}}\ast {\mathrm{FS}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_8{\mathrm{SIZE}}_{\mathrm{i}.\mathrm{t}}\\ +{\mathsf{\beta }}_9{\mathrm{LEV}}_{\mathrm{i}.\mathrm{t}}+{\mathsf{\beta }}_{10}{\mathrm{SG}}_{\mathrm{i}.\mathrm{t}}+{\mathrm{v}}_{\mathrm{t}}+{\mathrm{w}}_{\mathrm{k}}+{\mathrm{z}}_{\mathrm{c}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}\end{array}$$

(4)

The composite ESG score, the environmental, social, and governance pillar scores, and the financial-stability measure are standardized as z-scores before estimation. Specifically, the variables are transformed by subtracting their sample means and dividing by their standard deviations, resulting in variables with a mean of zero and a standard deviation of one. This transformation ensures that independent variables are measured on a comparable scale, thereby improving numerical stability and facilitating the interpretation of regression coefficients. In particular, standardized coefficients can be interpreted in terms of the change in the dependent variable associated with a one standard deviation change in the predictor. This approach is especially useful in models that include interaction terms, as it helps reduce multicollinearity and enhances the comparability of effect sizes across variables. The models also include firm-level control variables commonly associated with market valuation and year random intercepts to account for shocks shared across firms in a given year. The model also accounts for industry effects to control for systematic differences across sectors. This helps mitigate potential biases arising from industry-specific characteristics, such as regulatory conditions, market structure, operating risks, and ESG expectations. By controlling for industry heterogeneity, the analysis provides a more reliable estimation of the relationship between ESG and firm performance. Country-level effects are included to account for persistent cross-country differences in the financial market development, institutional quality, regulatory requirements, and ESG reporting requirements. This is consistent with evidence that green finance development and its institutional determinants vary substantially across geographic regions [53].

3.5. Model Estimation, Priors and Diagnostic Protocol

We estimated our hierarchical model using the Bayesian Markov Chain Monte Carlo (MCMC) sampling via brms package in R.

For priors diagnostic, the paper uses four different models: (i) informative normal prior is based on a strong prior direction of the coefficients; (ii) base line employs weakly informative Normal (0, 1) priors for all regression coefficients; (iii) Student t-prior and (iv) weakly informative prior use the distributions of normal and student t.

The Bayesian multilevel models were estimated using Markov Chain Monte Carlo (MCMC) sampling using the No-U-Turn Sampler (NUTS). Four independent chains were run, with 5000 iterations per chain. The first 2000 iterations of each chain were discarded as warm-ups, and every fifth post-warm-up draw was retained. This procedure produced 600 retained posterior draws per chain and 2400 posterior draws in total.

Convergence was evaluated using the Gelman-Rubin potential scale reduction factor ($\widehat{R}$) and values of $\widehat{R}<1.01$ across all parameters were taken as evidence of chain convergence. Effective sample sizes were examined to assess the precision of posterior summaries, and the sampling output was checked for divergent transitions and other warnings.

Posterior predictive checks were used to assess whether the fitted models adequately reproduced the principal characteristics of the observed firm-performance distributions.

Expected out-of-sample predictive performance is compared using Pareto-smoothed importance-sampling leave-one-out cross-validation (PSIS-LOO). Higher expected log predictive density (ELPD-LOO), or equivalently lower LOO information criterion (LOOIC), indicates better expected predictive performance. Differences in ELPD-LOO are interpreted relative to their standard errors rather than from rank order alone. Pareto-k diagnostics are examined to assess the reliability of the importance-sampling approximation.

3.6. Endogeneity and Temporal-Ordering Robustness Check

Potential endogeneity may arise because firms with stronger market performance could have greater resources and incentives to invest in ESG activities. Moreover, unobserved factors may affect both ESG and Tobin’s Q. We therefore conduct two complementary robustness checks.

First, we replace contemporaneous ESG with its one-period-lagged value to examine whether ESG performance precedes subsequent firm market performance. The lagged specification improves temporal ordering by reducing contemporaneous feedback between ESG and Tobin’s Q.

Second, to address potential endogeneity concerns arising from reverse causality and omitted variable bias, we estimate a Bayesian joint instrumental-variable model using a leave-one-out industry-year mean ESG score as an instrument for firm-level ESG performance.

For firm i operating in industry j during year t, the instrument is constructed as:

$$\mathit{IV}\_{\mathit{ESG}}_{i,j,t}= {\displaystyle \frac{({\sum }_{k\ne i}{\mathit{ESG}}_{k,j,t})-{\mathit{ESG}}_{i,j,t}}{N_{j,t}-1}}$$

where the summation represents the aggregate ESG score of all firms in industry j and year t exclude the ESG_i,j,t score of firm i and N_j,t denotes the total number of observed firms in industry j during year t.

The validity of the instrument depends on two identification requirements: relevance and the exclusion restriction. Firstly, industry peers within the same sector and year face identical macro-regulatory pressures, industry-wide environmental standards, and sustainability trends. Consequently, the average ESG engagement of industry peers exerts a strong influence on firm i’s strategic ESG behavior. Secondly, by explicitly subtracting firm i’s own ESG score from the industry aggregate and reducing the denominator by one (N_j,t − 1), we eliminate mechanical self-correlation. The aggregate ESG practices of other firms in the industry affect firm i‘s market performance (Tobin’s Q) exclusively through its effect on firm i’s own ESG, rather than directly impacting firm i’s individual firm value.

4. Results

4.1. Descriptive Statistics

The descriptive analysis in Table 1 reveals that ESG scores among the sampled firms are relatively balanced, with the combined ESG score averaging 53.53, and the three pillars, environmental (49.82), social (57.7), and governance (51.83), showing a moderate level of ESG engagement. Among the three pillars, firms appear to have better social practices, with the highest average S score of 57.7, compared with the environmental and governance pillars. The wide range of ESG scores and the three pillars suggest significant heterogeneity in ESG practices among firms.

Table 1. Descriptive statistics of variables.

	n	Mean	sd	Median	Trimmed	Mad	Min	Max
ESG	1124	53.53	17.45	55.53	54.03	18.1	1.89	92.03
E	1124	49.82	22.97	50.54	50.49	26.88	0	96.85
S	1124	57.7	20.2	58.91	58.6	21.65	0.33	97.24
G	1124	51.83	21.9	52.47	52.23	26.44	1.48	96.36
FS	1124	3.53	3.57	2.2	2.85	1.63	0.14	37.01
Z″-score	1124	6.05	6.32	3.91	4.9	3.04	−0.84	66.07
Tobin’s Q	1124	1.58	1.64	1.1	1.25	0.47	0.25	23.22
SIZE	1124	7.76	1.45	7.82	7.78	1.63	3.83	11.03
LEV	1124	0.2	0.15	0.18	0.19	0.19	0	0.64
SG	1124	0.07	0.29	0.04	0.05	0.17	−0.79	2.47

Note. The statistics are based on the common complete-case analytical sample of 1124 firm-year observations from 272 ASEAN firms over 2020–2024. Firm size is measured as the natural logarithm of total assets. ESG, its pillar scores, financial stability, and Z″-score are presented in their original units for descriptive purposes but are standardized before regression estimation. Tobin’s Q, leverage, and sales growth are presented in their original analytical forms.

Meanwhile, the mean value of FS is higher than the median, suggesting right-skewness. Trimmed means remain significantly higher than the median, which shows that firm-level financial stability varies among firms in the sample. Tobin’s Q, the proxy measure of firm performance averages 1.58, but the median is only 1.1. The relatively low median suggests that a typical firm is valued slightly above the book value. There exist firms with high perceived value in the dataset, evidenced by a very high maximum Tobin’s Q.

Regarding control variables, the mean and standard deviation of total assets (SIZE) show considerable dispersion. The mean and median of leverage indicate a moderate reliance on debt financing of the average firm in the sample. The wide range of sales growth (SG) shows that the sample includes both rapidly growing firms and firms experiencing sales declines.

Table 2 presents the correlation matrix for the variables used in the analysis. The environmental, social, and governance dimensions display strong correlations, reaffirming that the three pillars form a cohesive sustainability construct. By contrast, their associations with financial performance (Tobin’s Q) are low, suggesting that ESG practices do not have a direct influence on firms’ market-based outcomes in this sample. The results indicate that while ESG scores demonstrate high internal consistency, their immediate financial implications appear limited when compared with other fundamental firm characteristics such as size and leverage.

Table 2. Correlation matrix of variables.

	ESG	E	S	G	FS	Z″-Score	Tobin’s Q	SIZE	LEV	SG
ESG	1	0.84	0.88	0.66	0	0	0.13	0.32	0.08	0.02
E	0.84	1	0.73	0.27	−0.06	−0.07	0.06	0.37	0.13	0.02
S	0.88	0.73	1	0.36	0	−0.01	0.11	0.29	0.09	0
G	0.66	0.27	0.36	1	0.05	0.05	0.12	0.09	−0.02	0.03
FS	0	−0.06	0	0.05	1	0.97	0.72	−0.36	−0.5	0.06
Z″-score	0	−0.07	−0.01	0.05	0.97	1	0.67	−0.35	−0.5	0.05
Tobin’s Q	0.13	0.06	0.11	0.12	0.72	0.67	1	−0.25	−0.21	0.04
SIZE	0.32	0.37	0.29	0.09	−0.36	−0.35	−0.25	1	0.43	0.07
LEV	0.08	0.13	0.09	−0.02	−0.5	−0.5	−0.21	0.43	1	0.01
SG	0.02	0.02	0	0.03	0.06	0.05	0.04	0.07	0.01	1

Firm-level financial stability exhibits moderate to strong positive associations with firm performance, with a correlation coefficient of 0.72, raising a potential concern regarding multicollinearity. However, the VIF values for all variables included in the model as shown in Table 3 are below 2, indicating that multicollinearity is unlikely to cause substantial inflation of the estimated coefficient variances. These results imply that variations in firm-level financial stability are systematically aligned with changes in market-based performance. Such correlations provide preliminary support for the inclusion of firm-level financial stability in the regression model. Nevertheless, it should be noted that correlation analysis alone does not establish causal inference, and further empirical testing is needed to validate the relationship.

Table 3. VIF test.

Term	VIF	VIF 95% CI	adj. VIF	Tolerance	Tolerance 95% CI
ESG	1.25	[1.18, 1.35]	1.12	0.8	[0.74, 0.85]
SIZE	1.4	[1.31, 1.52]	1.18	0.71	[0.66, 0.76]
LEV	1.3	[1.22, 1.40]	1.14	0.77	[0.71, 0.82]
FS	1.35	[1.27, 1.46]	1.16	0.74	[0.68, 0.79]
SG	1.01	[1.00, 2.10]	1.01	0.99	[0.48, 1.00]

4.2. Results of Bayesian Hierarchical Models

We estimate a Bayesian hierarchical panel regression using MCMC sampling. Weakly informative priors are applied to regression coefficients. Specifically, we specify weakly informative normal priors for all regression coefficients, β ~ Normal (0, 1), to regularize extreme estimates while allowing sufficient flexibility.

The model is estimated using four Markov chains with 5000 iterations each, including 2000 warm-up iterations. Sampling is performed using the Hamiltonian Monte Carlo implemented in R (R 4.6.1). The Bayesian diagnostic results confirm that the estimates are reliable. $\widehat{R}$ statistic equals 1, demonstrating convergence among the Markov chains. Both bulk and tail effective sample sizes are sufficiently large, indicating stable posterior estimates.

The Bayesian regression results of the base model indicate that ESG is positively associated with firm performance (Table 4). Specifically, a one-standard-deviation increase in ESG is associated with an expected 0.29-unit increase in Tobin’s Q, holding all other variables constant. The 95% credible interval [0.22, 0.36] excludes zero, supporting Hypothesis 1. This result is consistent with the stakeholder theory and other empirical studies [44,48,54]. The Bayesian estimate indicates a stronger positive association between ESG and firm performance than that suggested by the bivariate correlation. Unlike the correlation coefficient, which captures an unconditional association between the two variables, the regression coefficient represents the expected change in firm performance associated with a one-standard-deviation change in ESG, conditional on the covariates and the model’s multilevel effects. This stronger conditional association suggests that observed firm characteristics and unobserved group-level heterogeneity may suppress part of the positive ESG-performance relationship in the unadjusted analysis.

Table 4. Results for the baseline model (Model 1).

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.06	0.47	2.06	3.94	1.00	1769	2008
ESG	0.29	0.04	0.22	0.36	1.00	2390	2414
SIZE	−0.11	0.03	−0.17	−0.06	1.00	2256	2266
LEV	0.30	0.02	0.27	0.34	1.00	2593	2436
FS	1.51	0.04	1.42	1.59	1.00	2371	2017
SG	0.00	0.11	−0.22	0.23	1.00	2452	2036
Random-intercept standard deviations
Industry	0.76	0.36	0.36	1.65	1.00	2177	1901
Year	0.21	0.15	0.06	0.58	1.00	1949	1647
Country	0.45	0.24	0.16	1.07	1.00	1818	2072
Residual	1.05	0.02	1.00	1.09	1.00	2126	2332

Note. n = 1124 firm-year observations. Tobin’s Q is expressed in its original units. ESG and firm-level financial stability are standardized as z-scores. Firm size is the natural logarithm of total assets. The model includes country, industry, and year random intercepts. Posterior means, posterior standard deviations, 95% credible intervals, $\widehat{R }$ and effective sample sizes are reported.

Financial stability is also positively associated with Tobin’s Q. Accordingly, a one-standard-deviation increase in firm-level financial stability is associated with an improvement in Tobin’s Q. The posterior mean coefficient of 1.51 indicates a strong and credible positive association of financial health and firm performance. The finding suggests that stronger firm-level financial stability is highly valued by the market.

The control variables, including size and leverage, show meaningful association with performance. The negative posterior mean coefficient of size implies that larger firms tend to have lower Tobin’s Q. This finding is similar to those of [44,47,55]. Larger firms might encounter operational inefficiencies despite their availability of resources, whereas smaller firms can benefit from agility but have inadequate capacity to invest in long-term ESG commitments [44]. Leverage has a positive posterior mean coefficient, indicating that leverage is positively associated with firm value. Sales growth does not exhibit a meaningful posterior relationship with Tobin’s Q. The result suggests the effect of sales growth is not supported in this sample.

Model 2 looks at the effect of individual pillars on market-based performance. Results in Table 5 show that among the three pillars of ESG, the environmental pillar exhibits the clearest positive association with Tobin’s Q, supporting Hypothesis 1a. Thus, a one-standard-deviation increase in environmental performance is associated with an expected 0.22 unit increase in Tobin’s Q. The social pillar has a positive posterior mean of 0.09, but its credible interval includes zero [−0.01, 0.19]. The evidence is therefore insufficient to establish a statistically credible independent association between social performance and Tobin’s Q. The governance coefficient is also positive but comparatively small. Its lower credible bound is reported as 0.00, making the evidence borderline and sensitive to rounding. Hypotheses 1b and 1c are not well supported. The Bayesian estimates for the individual ESG pillars differ from their corresponding zero-order correlations after accounting for the covariates and multilevel structure. Specifically, the environmental pillar exhibits statistically credible associations with firm performance, whereas the association for the social and governance pillars remains uncertain. This divergence is not unexpected, as zero-order correlations capture unconditional bivariate relationships, while the Bayesian estimates represent conditional associations. Moreover, posterior credibility depends not only on the magnitude of the estimated association but also on posterior uncertainty, potential confounding, the covariance among the predictors, and the independent variation contributed by each ESG pillar. Model 2 also shows a similar impact of firm-level financial stability on performance.

Table 5. Results for Model 2.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.30	0.49	2.33	4.28	1.00	1977	1769
E	0.22	0.05	0.12	0.32	1.00	2332	2285
S	0.09	0.05	−0.01	0.19	1.00	2359	2372
G	0.07	0.04	0.00	0.14	1.00	2367	2360
SIZE	−0.13	0.03	−0.19	−0.07	1.00	2308	2149
LEV	0.30	0.02	0.26	0.34	1.00	2309	2193
FS	1.61	0.05	1.52	1.70	1.00	2420	2413
SG	−0.00	0.11	−0.23	0.22	1.00	2337	2349
Random-intercept standard deviations
Industry	0.75	0.34	0.36	1.57	1.00	2197	2332
Year	0.23	0.16	0.07	0.63	1.00	2144	2317
Country	0.51	0.27	0.20	1.22	1.00	2059	2415
Residual	1.05	0.02	1.00	1.09	1.00	2387	2374

Table 6 describes the moderating effect of firm-level financial stability on the association of ESG performance and firms’ market value (Model 3). In detail, ESG shows a positive association with firms’ financial performance, and the relationship is strengthened by firm-level financial stability. The posterior mean coefficient of ESG is positive, with a 95% credible interval between 0.17 and 0.3. Since the credible interval is above zero, the results provide strong evidence of a positive relationship where a one-standard-deviation increase in ESG is associated with an expected increase of approximately 0.23 in Tobin’s Q, holding other variables constant.

Table 6. Results for Model 3.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.02	0.38	2.28	3.78	1.00	2056	2069
ESG	0.23	0.03	0.17	0.30	1.00	2631	2475
SIZE	−0.10	0.03	−0.15	−0.05	1.00	2316	1945
LEV	0.30	0.02	0.27	0.34	1.00	2232	2329
FS	1.49	0.04	1.41	1.57	1.00	2489	2460
SG	−0.01	0.11	−0.22	0.21	1.00	2333	2060
ESG × FS	0.35	0.03	0.30	0.40	1.00	2347	2195
Random-intercept standard deviations
Industry	0.73	0.33	0.35	1.67	1.00	2137	2253
Year	0.21	0.15	0.06	0.60	1.00	1959	2132
Country	0.27	0.18	0.07	0.73	1.00	1775	2118
Residual	0.99	0.02	0.95	1.03	1.00	2482	2364

The coefficient of firm-level financial stability is strongly positive, with a posterior mean of 1.49 and a 95% credible interval from 1.41 to 1.57. The finding clearly shows that firms with stronger financial health are better valued by the market. It confirms the fact that firm-level financial stability is an important determinant of firms’ market performance.

More importantly, the interaction term between ESG and FS is also positive, with a posterior mean of 0.35 and a 95% credible interval ranging from 0.3 to 0.4. This result represents a conditional association, not definitive evidence of a causal moderating mechanism. The findings support Hypothesis 2, which suggests that financial health positively moderates the relationship between ESG and firm performance. The positive interaction coefficient indicates that the ESG-Tobin’s Q association becomes stronger as firm-level financial stability increases. Firm-level financial stability may provide complementary capacity that enables firms to finance, sustain, and coordinate ESG initiatives and convert the resulting stakeholder benefits into market value. This reasoning is consistent with firm-level evidence that the performance value of a strategic activity depends on complementary organizational resources and capabilities [56].

Specifically, ESG has a stronger positive association with Tobin’s Q for firms with higher financial stability. In contrast, the benefits of ESG are weaker for firms with weaker financial conditions. Consistent with the organizational slack theory, these findings imply that investors tend to value ESG investments when firms have sufficient financial strength to sustain ESG commitments and convert them into long-term value.

Results of Model 4 provide evidence on how the moderating effect of firm-level financial stability differs across the three pillars of ESG (Table 7). The environmental pillar retains a statistically credible positive association with Tobin’s Q, with a posterior mean coefficient of 0.14 and the 95% credible interval from 0.05 to 0.23. The main associations for the social ($\mathsf{\beta }$ = 0.09, 95% CrI [−0.00, 0.18]) and governance pillars ($\mathsf{\beta }$ = 0.06, 95% CrI [−0.01, 0.12]) remain uncertain.

Table 7. Results for Model 4.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	2.98	0.43	2.07	3.81	1.00	2284	2062
E	0.14	0.05	0.05	0.23	1.00	2491	2092
S	0.09	0.05	−0.00	0.18	1.00	2278	2412
G	0.06	0.03	−0.01	0.12	1.00	2250	2398
SIZE	−0.12	0.03	−0.17	−0.07	1.00	2451	2164
LEV	0.29	0.02	0.26	0.33	1.00	1997	2022
FS	1.51	0.05	1.42	1.60	1.00	2395	2411
SG	−0.01	0.11	−0.22	0.21	1.00	2533	2299
E × FS	−0.07	0.05	−0.17	0.03	1.00	2349	2291
S × FS	0.34	0.06	0.22	0.46	1.00	2313	2340
G × FS	0.24	0.04	0.17	0.31	1.00	2455	2143
Random-intercept standard deviations
Industry	0.71	0.31	0.33	1.48	1.00	2207	2068
Year	0.23	0.16	0.07	0.68	1.00	1810	1698
Country	0.34	0.20	0.11	0.87	1.00	2055	1911
Residual	0.97	0.02	0.93	1.01	1.00	2290	2372

The interaction term between firm-level financial stability and the environmental component is negative and insignificant. Thus, Hypothesis 2a is not supported by posterior evidence. Therefore, the result does not establish that financial stability either strengthens or weakens the environmental practices-performance association.

By contrast, the social interaction is positive and statistically credible ($\mathsf{\beta }$ = 0.34, 95% CrI [0.22, 0.46]), indicating that the association between social practices and Tobin’s Q becomes more positive as financial stability increases. The governance interaction is also positive and credible ($\mathsf{\beta }$ = 0.24, 95% CrI [0.17, 0.31]). Thus, Hypotheses 2b and 2c are supported. Financially stable firms therefore appear to be better able to translate their social and governance practices into market value.

4.3. Robustness Check

We assessed the robustness of the Bayesian estimates in three ways. First, we conducted a prior sensitivity analysis by re-estimating the model under alternative weakly informative priors [57]. Second, we evaluated model fit using posterior predictive checks, comparing replicated and observed data distributions. Third, we examined MCMC convergence diagnostics, the potential scale reduction factor ($\widehat{R}$ < 1.01), and effective sample sizes.

First, we conducted a prior sensitivity analysis by re-estimating the model under several alternative prior specifications. Following [57], the objective was to examine whether the posterior distributions were sensitive to reasonable changes in prior assumptions. The posterior estimates for the ESG parameter exhibit remarkable stability across all three prior specifications. Under the baseline of normal distribution, the posterior mean for ESG performance is $\mathsf{\beta }$ = 0.271 with a 95% equal-tailed credible interval (CrI) of 0.203 and 0.337. Incorporating informative domain knowledge, the normal distribution for ESG (0.3, 0.1) slightly increases the posterior estimate to $\mathsf{\beta }$ = 0.275 with 95% CrI [0.211, 0.343], reflecting a minor shrinkage toward historical empirical benchmarks. Importantly, specified under a heavy-tailed Student-t (1, 0, 2.5) prior, the posterior mean remains nearly identical to the baseline ($\mathsf{\beta }$ = 0.271, 95% CrI [0.202, 0.337]). The similar posterior estimate under the heavy-tailed Student-t coefficient prior indicates that the ESG estimate is not materially sensitive to the choice between Gaussian and heavy-tailed coefficient priors. However, the mixture distribution of Gaussian and heavy-tailed in weakly informative prior diagnostics seems to be the worst.

Second, cross-validation assesses the predictive stability of the estimated models, with results presented in Table 8. PSIS-LOO comparison indicates broadly similar expected out-of-sample predictive performance across the prior specifications. The ELPD differences are small relative to their standard errors, providing no decisive predictive advantage for one specification. Stability of the ESG coefficient is assessed separately by comparing its posterior mean and 95% credible interval across the prior specifications.

Table 8. Prior sensitivity analysis.

Prior Specification	ELPD Difference	SE Difference	ELPD-LOO	SE (ELPD-LOO)	pLOO	SE (pLOO)	LOOIC	SE (LOOIC)
Informative normal prior	0.0	0.0	−1684.9	108.0	50.0	20.7	3369.9	216.0
Base line	−2.7	2.1	−1687.7	109.9	47.3	18.1	3375.4	219.7
Student-t prior	−2.8	2.8	−1687.8	110.3	50.0	20.9	3375.5	220.7
Weakly informative prior	−3.4	2.7	−1688.4	110.3	49.7	19.7	3376.8	220.7

Third, all parameters achieved excellent MCMC convergence. The potential scale reduction factor ($\widehat{R}$) was equal to 1.00 for all parameters. Bulk and tail effective sample sizes exceeded 2000, indicating that the posterior estimates were based on a large number of effectively independent draws. No divergent transitions were observed, and trace plots showed satisfactory mixing across chains. Collectively, these diagnostics confirm the reliability of the posterior estimates.

To ensure the robustness of our findings, we re-estimate our models using a traditional panel data regression approach. This analysis serves as an alternative to the Bayesian regression method to assess the robustness of the results. The results obtained from panel regression reported in Table 9 are consistent with those of Bayesian estimates in both direction and statistical significance of the key coefficient, including the interaction terms between ESG and firm-level financial stability. Specifically, ESG remains positively associated with Tobin’s Q and the firm-level financial stability continues to exert a significant positive moderating effect. The findings indicate that our conclusions are not sensitive to the choice of estimation method.

Table 9. Results of robustness check of the 4 models using panel regression.

	Model 1	Model 2	Model 3	Model 4
(Intercept)	3.21129 ***	3.33113 ***	3.00864 ***	3.15547 ***
	(0.36687)	(0.38063)	(0.33259)	(0.33909)
ESG	0.28826 ***		0.22460 ***
	(0.03637)		(0.03439)
SIZE	−0.11384 ***	−0.12553 ***	−0.09860 ***	−0.11565 ***
	(0.02907)	(0.02960)	(0.02715)	(0.02745)
LEV	0.30496 ***	0.30408 ***	0.29892 ***	0.29430 ***
	(0.01979)	(0.01977)	(0.01860)	(0.01841)
FS	1.61632 ***	1.61533 ***	1.56748 ***	1.51151 ***
	(0.04550)	(0.04540)	(0.04302)	(0.04415)
SG	−0.00735	−0.00609	−0.00899	−0.00702
	(0.11390)	(0.11377)	(0.10730)	(0.10593)
E		0.21562 ***		0.14184 **
		(0.05002)		(0.04679)
S		0.08394		0.08881 +
		(0.05153)		(0.04798)
G		0.06848 +		0.05943 +
		(0.03534)		(0.03291)
ESG × FS			0.36619 ***
			(0.02983)
E × FS				−0.06993
				(0.05062)
S × FS				0.34089 ***
				(0.05899)
G × FS				0.24287 ***
				(0.03735)
SD (Intercept ctry)	0.32256	0.36864	0.18632	0.24333
SD (Intercept ind)	0.55581	0.56210	0.53371	0.52013
SD (Intercept year)	0.12989	0.14027	0.13273	0.14298
SD (Observations)	1.04666	1.04358	0.98510	0.97067
Num.Obs.	1124	1124	1124	1124
R2 Marg.	0.552	0.547	0.606	0.611
R2 Cond.	0.678	0.684	0.707	0.716
AIC	3372.6	3378.9	3239.8	3231.5
BIC	3422.9	3439.2	3295.1	3306.9
ICC	0.3	0.3	0.3	0.3
RMSE	1.04	1.03	0.98	0.96
+ p < 0.1, * p < 0.05, ** p < 0.01, *** p < 0.001

Note. The model uses the same 1124 observations, variable transformations, covariates, and country, industry, and year grouping structure as the corresponding Bayesian model. Values in parentheses are standard errors. Tobin’s Q remains in its original units, while ESG and firm-level financial stability are standardized.

The panel regression yields positive coefficient estimates across all three individual ESG pillars. Specifically, the environmental pillar exhibits a statistically significant positive association with firm performance, fully aligning with the Bayesian framework. While the social and governance coefficients maintain consistent positive signs across both panel and Bayesian specifications, they lack robust statistical support across both estimation methods. The moderating role of firm-level financial stability in the environmental pillar-performance relationship is also sensitive to the estimation method. The conventional panel regression produces an insignificant negative interaction coefficient, whereas the corresponding Bayesian posterior distribution contains substantial probability mass on both sides of zero, with its 95% credible interval including zero. The consistent, positive interaction coefficients across both estimation approaches provide robust evidence that firm-level financial stability positively moderates the relationships between both the social and governance dimensions and firm performance.

Finally, to address the endogeneity, we use temporal order checking and Bayesian joint instrumental-variable model for ESG and Tobin’s Q.

Table 10 shows that the one-period-lagged ESG coefficient is not statistically credible, indicating insufficient evidence that prior-year ESG predicts current Tobin’s Q. While the direction of the lagged ESG estimate is consistent with the contemporaneous baseline, directional alignment alone does not confirm the primary result. Introducing the lag reduces the effective time dimension and statistical power, especially when ESG scores often change gradually over time. Consequently, while the lagged specification provides a partial check on temporal ordering, it yields inconclusive evidence regarding a persistent ESG effect or the elimination of reverse causality.

Table 10. Results of temporal order checking.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	2.45	0.45	1.54	3.32	1.00	1785	1243
L.ESG	0.04	0.03	−0.03	0.10	1.00	2386	2414
SIZE	−0.02	0.03	−0.07	0.04	1.00	2482	2254
LEV	0.31	0.02	0.27	0.35	1.00	2165	2238
FS	1.55	0.04	1.47	1.64	1.00	2226	2280
SG	−0.02	0.11	−0.25	0.20	1.00	2268	1339
Random-intercept standard deviations
Industry	0.73	0.33	0.34	1.65	1.00	1738	1168
Year	0.11	0.10	0.01	0.38	1.00	2050	2100
Country	0.34	0.18	0.12	0.81	1.00	2038	2052
Residual	1.08	0.02	1.03	1.12	1.00	2397	2332

Note. The dependent variable is Tobin’s Q in year t, and the principal explanatory variable is ESG in year t − 1 (L.ESG). The reduction in observations relative to the baseline model results from constructing the one-period lag and applying complete-case deletion.

As an additional endogeneity robustness analysis, we estimated a Bayesian joint instrumental-variable model using the leave-one-out industry-year mean ESG score as an instrument for firm ESG performance. The focal firm was excluded from the peer average to prevent mechanical self-correlation. The results of the ESG equation and firm-performance equation are reported in Table 11. Although the joint model provides evidence regarding the sensitivity of the ESG coefficient to an instrumental-variable specification, its interpretation remains conditional on the relevance and exclusion assumptions discussed in the methodology section.

Table 11. Results of Bayesian joint instrument-variable model.

Parameter	Posterior Mean	Posterior SD	95% Credible Interval	Rhat	Bulk ESS	Tail ESS
Panel A. ESG equation:
Intercept	11.45	1.71	[8.18, 14.86]	1.00	6932	6377
IV- ESG	−0.25	0.03	[−0.30, −0.20]	1.00	9938	6240
FS	0.12	0.04	[0.05, 0.19]	1.00	12,449	6559
SIZE	0.31	0.02	[0.27, 0.35]	1.00	13,724	5920
LEV	0.00	0.02	[−0.03, 0.04]	1.00	11,403	6075
SG	−0.13	0.09	[−0.32, 0.04]	1.00	13,724	6082
Panel B. Structural equation: dependent variable = Tobin’s Q
Intercept	3.20	0.60	[2.00, 4.40]	1.00	3737	5129
ESG	0.29	0.13	[0.02, 0.56]	1.00	3198	3723
FS	1.62	0.05	[1.52, 1.72]	1.00	6471	6156
SIZE	−0.11	0.05	[−0.22, −0.01]	1.00	3744	4651
LEV	0.30	0.02	[0.27, 0.34]	1.00	11,006	6462
SG	−0.01	0.12	[−0.23, 0.22]	1.00	12,318	5582
Panel C. Residual correlation
ESG equation ↔ Tobin’s Q equation	0.00	0.11	[−0.21, 0.22]	1.00	3268	3717
Panel D. Residual standard deviations
ESG equation	0.82	0.02	[0.78, 0.85]	1.00	13,775	5401
Tobin’s Q equation	1.05	0.02	[1.01, 1.11]	1.00	8524	5465

Panel A shows a negative and statistically credible coefficient for the leave-one-out peer ESG instrument. Because its posterior distribution is concentrated away from zero, the instrument satisfies the relevance condition by providing meaningful predictive power for the focal firm’s ESG score, conditional on covariates. While the negative coefficient suggests an inverse relationship with peer performance, potentially reflecting strategic differentiation or competitive positioning, instrument relevance relies on predictive association rather than a positive sign.

As shown in Panel B, after accounting for the joint instrumental-variable structure, the ESG coefficient in the Tobin’s Q equation remains positive with a 95% credible interval of 0.02 and 0.56. The posterior mean is similar to that obtained from the principal contemporaneous model, although its credible interval is substantially wider. The greater posterior uncertainty is expected because the instrumental-variable specification identifies the ESG coefficient using the variation associated with the peer-based instrument rather than all observed variation in ESG.

The estimated residual correlation between the ESG and Tobin’s Q equations is approximately zero. Thus, the model does not provide clear evidence of residual dependence between the unobserved determinants of ESG and Tobin’s Q after conditioning on the proposed instrument, controls, and multilevel structure. The joint-model results therefore strengthen the robustness of the positive contemporaneous ESG-performance association, conditional on the instrument assumptions, but do not conclusively establish causality.

To assess whether the moderation results are driven by mechanical overlap between firm-level financial stability and Tobin’s Q, we replace the original Altman Z-score with the emerging-market Z″-score [58]. The Z″-score replaces the market-value ratio with the book-value-of-equity-to-total-liabilities ratio and does not contain a market-based component. Results from Model 1 using the Z″-score yield consistent findings, reinforcing the robust positive association between the composite ESG score and Tobin’s Q (Table 12).

Table 12. Results of model 1 robustness check with Z″-score as the proxy for firm-level financial stability.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.37	0.47	2.47	4.32	1.00	2176	2143
ESG	0.31	0.04	0.23	0.39	1.00	2655	2294
SIZE	−0.14	0.03	−0.21	−0.08	1.00	2481	2143
LEV	0.28	0.02	0.23	0.32	1.00	2273	2207
Z″-score	1.32	0.05	1.23	1.41	1.00	2177	2226
SG	0.09	0.12	−0.15	0.32	1.00	2303	2337
Random-intercept standard deviations
Industry	0.73	0.32	0.34	1.57	1.00	1911	1997
Year	0.23	0.16	0.06	0.69	1.00	1920	1923
Country	0.39	0.24	0.10	0.98	1.00	1864	1801
Residual	1.16	0.03	1.11	1.21	1.00	2369	2068

Results in Table 13 show that the environmental estimate remains almost unchanged and its credible interval continues to exclude zero. The social estimate also changes very little and remains uncertain. The governance coefficient retains the same magnitude, but its Z″-score credible interval includes zero. The evidence for a direct governance association should therefore be described as weak or inconclusive.

Table 13. Results of model 2 robustness check with Z″-score as the proxy for firm-level financial stability.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.30	0.48	2.35	4.25	1.00	2002	2240
E	0.23	0.06	0.12	0.34	1.00	2192	2330
S	0.10	0.06	−0.01	0.21	1.00	2230	2316
G	0.07	0.04	−0.01	0.15	1.00	2555	2373
SIZE	−0.16	0.03	−0.22	−0.09	1.00	2241	2255
LEV	0.28	0.02	0.23	0.32	1.00	2434	2230
Z″-score	1.32	0.05	1.23	1.42	1.00	2313	2163
SG	0.09	0.13	−0.16	0.33	1.00	2365	2390
Random-intercept standard deviations
Industry	0.76	0.33	0.35	1.58	1.00	2204	2223
Year	0.25	0.20	0.07	0.74	1.00	2057	2150
Country	0.46	0.27	0.15	1.15	1.00	1960	2112
Residual	1.16	0.02	1.11	1.20	1.00	2523	2289

The Z″-score itself has a positive and statistically credible association with Tobin’s. The control-variable results also remain broadly consistent. Model 2 is therefore robust primarily with respect to the positive environmental association and the absence of clear social and governance associations.

Re-estimating Model 3 with Z″-score produces a positive and statistically credible ESG–financial-stability interaction (β = 0.32), 95% CrI [0.25, 0.38]) (Table 14). This result is consistent with the principal moderation specification and indicates that the interaction is robust to using an accounting-based financial-stability measure that excludes a market-value component.

Table 14. Results of model 3 robustness check with Z″-score as the proxy for firm-level financial stability.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.26	0.43	2.38	4.11	1.00	2285	2241
ESG	0.28	0.04	0.21	0.36	1.00	2416	2334
SIZE	−0.14	0.03	−0.20	−0.08	1.00	2370	2295
LEV	0.28	0.02	0.24	0.32	1.00	2456	2294
Z″-score	1.31	0.05	1.22	1.40	1.00	2314	2328
SG	0.08	0.12	−0.15	0.32	1.00	2334	2191
ESG × Z″-score score	0.32	0.03	0.25	0.38	1.00	2549	2263
Random-intercept standard deviations
Industry	0.70	0.32	0.33	1.54	1.00	1868	2161
Year	0.23	0.16	0.06	0.67	1.00	1911	1947
Country	0.23	0.16	0.03	0.61	1.00	1579	1710
Residual	1.12	0.02	1.07	1.16	1.00	2387	2332

The social interaction is virtually identical across the two financial-stability measures. The governance interaction is also highly stable in its direction, magnitude, and posterior uncertainty. These results provide strong robustness evidence that firm-level financial stability strengthens the social–performance and governance–performance associations.

The environmental interaction remains negative under both measures, but neither specification provides clear evidence of a moderating effect. In Table 15, the upper credible bound is reported as 0.00; because this may reflect rounding, the environmental interaction should be described as borderline rather than conclusively negative.

Table 15. Results of model 4 robustness check with Z″-score as the proxy for firm-level financial stability.

	Posterior Mean	Posterior SD	95% CrI Lower	95% CrI Upper	$\widehat{\mathit{R}}$	Bulk ESS	Tail ESS
Intercept	3.24	0.43	2.37	4.03	1.00	2157	2352
E	0.17	0.06	0.07	0.28	1.00	2198	2294
S	0.11	0.05	0.01	0.22	1.00	2174	1903
G	0.08	0.04	0.01	0.15	1.00	2375	2173
SIZE	−0.15	0.03	−0.21	−0.09	1.00	2362	2331
LEV	0.27	0.02	0.23	0.32	1.00	2500	2456
Z″-score	1.26	0.04	1.18	1.35	1.00	2523	2291
SG	0.08	0.12	−0.15	0.32	1.00	2485	2219
E × Z″-score	−0.10	0.05	−0.20	0.00	1.00	2349	2114
S × Z″-score	0.34	0.06	0.23	0.46	1.00	2369	2073
G × Z″-score	0.23	0.04	0.15	0.31	1.00	2438	2368
Random-intercept standard deviations
Industry	0.69	0.30	0.33	1.44	1.00	2068	2164
Year	0.25	0.17	0.08	0.71	1.00	2030	2294
Country	0.30	0.19	0.06	0.80	1.00	2036	2112
Residual	1.10	0.02	1.06	1.15	1.00	2567	2462

The main social and governance coefficients become statistically credible in the Z″-score model. However, these coefficients represent the pillar associations when Z″-score is at its mean and do not alter the principal moderation conclusion.

5. Discussion

The results of the Bayesian regressions show a positive contemporaneous association of ESG and firm performance across all the models. This finding is consistent with many recent empirical studies, emphasizing the role of ESG activities in enhancing firm value [15,45,51,59]. The findings are consistent with the stakeholder theory, which mandates that stakeholders highly value the firms’ engagement in ESG activities. ESG is not just a compliance requirement but can function as a strategic driver that helps improve market confidence. Though the ESG combined score positively associates with firm performance, the effect of individual components varies. Our results show that among the three pillars, environment and governance components are the factors associated with higher firm performance, and the environment pillar shows clearer evidence than the governance one. The study of Candio (2024) [15] makes similar conclusions on the stronger explanatory power of environment and governance factors on the performance of European firms but demonstrates that the governance pillar has a dominating effect. Meanwhile, the empirical work of Maji and Lohia (2024) [51] in an Indian sample observes only the significant positive relationship of the social and governance factors. The different impacts of the environmental, social, and governance dimensions on firm performance may reflect the heterogeneity in the socioeconomic conditions across the regions covered by the sample.

The moderating role of firm-level financial stability provides a more nuanced contribution. The interaction between ESG and firm-level financial stability is positive, indicating the role of firm-level financial stability in strengthening the effect of ESG on performance. In other words, ESG engagement may be associated with higher market value, potentially through stronger stakeholder relationships, reduced risk, and enhanced organizational legitimacy. This moderating effect also explains why mixed findings on the association between ESG and performance are reported in prior studies [10,16,17,60]. ESG engagement may not generate homogeneous benefits across firms. In fact, its effects are likely dependent on firm-specific financial conditions. Financially stable companies possess superior slack resources to absorb ESG investment costs [38]; thus, they are better positioned to convert ESG initiatives into improved performance and higher market valuation. Conversely, financially distressed firms may face liquidity constraints that restrict their capacity to derive value from ESG activities. Under conditions of financial distress, investors may view ESG expenditures as costly.

Results from the analysis of individual ESG dimensions further support this interpretation. The social pillar exhibits clear positive interaction with firm-level financial stability, indicating that the social factor is highly valued when firms are financially stable. Social initiatives are appreciated by investors when firms have the financial strength to sustain them. This finding is consistent with prior evidence that socially responsible firms are financially viable during periods of uncertainty or shock [61,62]. The governance dimension also shows positive interaction with firm-level financial stability. It suggests that governance performance is value-relevant to investors for firms with stronger firm-level financial stability. Effective governance can mitigate agency costs, enhance transparency, and foster investor confidence. However, for financially distressed firms, governance quality alone may not be sufficient to reduce investor concerns regarding liquidity constraints, solvency pressures, or default risk. This finding is consistent with prior empirical evidence, suggesting that governance and transparency enhance the credibility of ESG activities and reinforce their valuation effects [45,46].

Although the environmental pillar demonstrates evidence of a positive association, its moderating effect with firm-level financial stability is not statistically credible. This suggests that investors value environmental practices regardless of firms’ financial condition. Environmental initiatives may be perceived as value-relevant because they enhance regulatory compliance and resource efficiency. Unlike social or governance activities, the association of the environmental factor may be less dependent on firm-level financial stability because the benefits of environmental activities may be realized in the long term; hence, it is less sensitive to firms’ current financial stability. As a result, the moderating role of firm-level financial stability is not evident. In other words, financial health does not amplify the positive relationship between environmental performance and firm performance.

The findings should be interpreted in light of the unusual economic conditions of 2020–2024, encompassing both the COVID-19 disruption and subsequent recovery. Accordingly, firms faced significant change in market valuation, risk, financing conditions, and stakeholder expectations that may have influenced ESG engagement and financial performance simultaneously. While year-, country-, and industry-level random intercepts account for common temporal, institutional, and sectoral heterogeneity, they may not fully capture pandemic-related shocks that varied across firms, industries, and countries. Evidence suggests that green finance supported environmental and economic resilience during the recovery period, although its effectiveness depended on policy and financial conditions [63]. Therefore, the results should be generalized cautiously to more stable economic environments.

6. Conclusions, Contributions, and Limitations

6.1. Conclusions

The relationship between ESG and firm performance is explored in this study with a particular focus on the moderating role of firm-level financial stability. The contemporaneous findings uncover a positive association between ESG and Tobin’s Q, suggesting that firms with stronger engagement in ESG practices tend to exhibit higher market value. This association is consistent with the argument that ESG engagement may create value through improving stakeholder trust, mitigating risk, strengthening organizational legitimacy and strategic focus. In particular, the study results show a positive role of firm-level financial stability in moderating the interplay between ESG and performance. Firms with strong financial viability are in a better position to convert ESG activities into market value as they have resources, credibility and capacity to sustain ESG initiatives. In contrast, firms with financial distress may face funding constraints and survival pressures that limit the extent to which ESG practices can be implemented to enhance value to investors.

Deeper analysis of the influence of individual ESG pillars on firm performance reveals that the value implications are not uniform. Each dimension affects performance via distinct mechanisms. Environmental and governance components are positively related to firm performance, with the environment factor having a more pronounced association. However, when the moderating effect of firm-level financial stability is considered, the associations of individual ESG pillars differ significantly. Specifically, the governance and social dimensions are positively associated with firm value among firms with greater financial stability whereas the corresponding evidence for the environmental pillar remains inconclusive.

The results of this study indicate that the positive contemporaneous ESG-performance association is retained under the joint instrumental-variable specification but is not confirmed by the lagged analysis. The findings therefore remain conditional on the instrument’s exclusion restriction and should not be interpreted as definitive evidence of causality or the absence of reverse causality. Future research using longer panels, plausibly exogenous regulatory changes, or credible external instruments is needed to clarify the temporal and causal nature of the relationship.

6.2. Contributions

This study contributes to ESG literature by demonstrating that ESG and its pillars can enhance firm performance and that firm-level financial stability is an important moderator in the ESG-firm performance relationship. The results provide practical implications for managers, investors, and policymakers. For managers, ESG activities should be integrated with financial capacity. Specifically, investments in ESG activities may only benefit firms with strong financial stability. Meanwhile, investors should consider both ESG and firm-level financial stability in making investment decisions. For policymakers, the findings suggest that disclosure regulation is not sufficient. Improved institutional support, financial access, and reliable reporting practices are critical to enable financially constrained firms to convert ESG commitments into real value.

6.3. Limitations and Future Research

This study has several limitations. First, firm-level financial stability is measured using Altman’s Z-score. Although the Z-score provides a well-established measure of firms’ financial condition, it is primarily based on accounting information and may not capture all dimensions of firm-level financial stability, such as access to external financing or liquidity flexibility. Future research could therefore examine whether the moderating effect of firm-level financial stability remains consistent when alternative measures of firm-level financial stability are used.

Second, the empirical analysis is based on firms from six ASEAN countries. As such, the findings may not be generalizable to firms outside the ASEAN context. Differences in regulatory systems and institutional environments may influence both ESG practices and their valuation by investors. Future research could extend the analysis to other markets to examine whether the role of firm-level financial stability in the ESG-firm value relationship remains across different settings.

Third, firm performance is proxied by Tobin’s Q, which may be influenced by investor sentiment and other factors unrelated to firms’ ESG activities. Future research could use both market- and accounting-based indicators to examine if the findings are consistent across different measures of firm performance.

Fourth, the sample period coincides with the COVID-19 disruption and post-pandemic recovery. Although the multilevel model accounts for common year effects and persistent country and industry heterogeneity, it may not fully capture country-specific or industry-specific shocks that changed over time. Future research should examine a longer period and explicitly model country-by-year or industry-by-year variation.