Financial Constraints and the ESG–Firm Performance Nexus in the Automotive Industry: Evidence from a Global Panel Study

Abstract

This study examines the complex relationship between environmental, social, and governance (ESG) and financial performance in the automotive industry, with a particular focus on how financial constraints shape this relationship. Using a global data set for the period 2008 to 2023 and employing a range of panel data techniques, including those addressing endogeneity concerns, we find that higher ESG scores positively affect financial performance. Specifically, a one-point rise in ESG score corresponds to an estimated 1–1.7% increase in the market-to-book ratio, with the effect reaching approximately 1.6% for firms facing financial constraints. These findings highlight the economic significance of ESG engagement, particularly for resource-constrained companies. The novelty of this study is that it focuses on the automotive sector, an industry with limited ESG-specific research, and that it makes a theoretical contribution by linking ESG performance outcomes to financial constraints, an angle largely overlooked in prior research. The findings offer critical policy insights, emphasizing the strategic importance of ESG initiatives for value creation under varying financial conditions.

1. Introduction

Sustainable development is gaining increasing global attention. The growing awareness within societies, along with the Paris Agreement and the UN Sustainable Development Goals (SDGs), has significantly contributed to this heightened focus [1]. Firms are now being driven by international standards, government interventions, changing stakeholder expectations, media pressure, and emerging business trends to prioritize sustainability. ESG criteria, which evaluate a firm’s impact on the environment, society, and governance, are becoming increasingly important to customers, investors, and the public. ESG investments can enhance corporate reputation, enable cost savings, improve risk management, attract funding, and help firms achieve long-term sustainability goals [2]. However, these investments also impose additional costs on firms and have long payback periods [3,4,5]. Consequently, the ESG and financial performance association continues to be multifaceted and varies across industries.

ESG factors span a wide array of concerns, including climate change, labor policies, supply chain management, board diversity, and ethical behavior. These concerns are particularly critical for automotive firms due to the industry’s significant negative impact on the environmental and societal impacts. Historically, the automotive industry has not paid much attention to the ESG impact of its operations. Traditional assembly line manufacturing requires substantial energy, uses toxins, metals, plastics, and a large labor force [6]. Additionally, most vehicles (99% of those on the road) use fossil fuels, leading to increased carbon emissions. According to Greenpeace [7], the transportation sector accounts for 24% of direct carbon emissions, with passenger vehicles responsible for 45% of this share. Decarbonizing the transportation sector is crucial, as annual greenhouse gas emissions in 2050 are projected to be 90% higher than in 2020 if no action is taken. ESG Risk Atlas [8] ranks the Autos and Auto Parts sector 8th out of 34 industries, where a higher ranking indicates greater relative ESG risk exposure. Firms operating in this sector are under pressure to align with sustainability standards, which require significant changes in production processes and substantial investments in research and development. Despite some progress, Greenpeace [7] indicates that more accelerated actions are needed.

In addition to environmental concerns, automotive firms face significant challenges in the social and governance domains. Socially, the industry has a significant impact on employment, with about 8% of the manufacturing workforce in the US and Europe employed in the automotive industry [9,10]. The transition to electric vehicles will impact the workforce, requiring fewer assembly workers as electric vehicles can be manufactured in two-thirds of the time it takes to manufacture internal combustion engine vehicles. For example, Ford Motor Co. announced in August 2022 that it would lay off 3000 workers as part of a restructuring program [11]. If such layoffs are not managed transparently and fairly, they could negatively impact firms’ social pillar scores. Moreover, as the industry shifts toward more technology-focused jobs, workforce upskilling will be essential. Product safety also remains another important concern, as technological advancements do not always guarantee safety. For example, Tesla has recalled 2 million cars in the US since 2012 due to problems with self-driving features. In addition, the COVID-19 pandemic highlighted the significance of employee health, supply chain resilience, and community support, all of which have elevated the importance of the social pillar in ESG evaluations. On the governance front, some automakers have lobbied against stricter regulations and participated in emissions cartels to limit competition in diesel passenger vehicle emission control technologies [7]. These actions significantly increase governance risks. Indeed, Volkswagen and BMW were fined USD 1 billion for their involvement in emissions cartels [12]. Furthermore, as conventional business models evolve, firms will need to innovate continuously and incorporate diverse local and technical perspectives. A diverse management and board composition will become imperative. Reference [13] also highlights that the automotive industry’s complex multi-tiered supply chains involve numerous stakeholders, including suppliers, customers, employees, regulators, communities, industry associations, and investors. Integrating ESG practices at multiple levels within these supply chains is crucial for overall firm performance.

In summary, automotive firms face substantial ESG challenges. ESG investments are costly and take a long time to bear fruit, resulting in a limited number of firms in the industry producing ESG reports. The industry’s high sensitivity to ESG-related risks makes it an important player in achieving SDGs. Demonstrating ESG’s influence on financial performance can encourage firms to realign their sustainability initiatives and disclosure practices. Although there is an extensive literature on the ESG–firm performance nexus, empirical findings remain inconclusive and varied. Moreover, only a few studies have examined this relationship within the automotive industry [13,14,15]. As noted by [2,16,17], the correlation between ESG and performance metrics diverges by industry and specific ESG factors. Analyzing this issue within a single industry helps overcome the problem of heterogeneity between different industries and provides industry-specific implications.

From the limited number of studies in the automotive industry, Reference [13] contributes to understanding how stakeholder influences impact firm sustainability practices and outcomes. Drawing on a data set comprising more than 600 companies within the global automotive sector, they find that stakeholders significantly influence firms’ ESG scores. However, the study lacks a direct analysis of how ESG governance and stakeholder influences on ESG practices impact financial performance. Reference [18] investigates how both favorable and unfavorable corporate social responsibility practices (CSRP) influence the financial performance of global automotive companies, with market-based assets serving as a moderating variable. The results based on the Generalized Method of Moments (GMM) approach reveal that positive (negative) CSRP enhances (diminishes) financial performance, with market-based assets amplifying these relationships. Reference [15] explores the link between ESG and firm performance across Asian markets using panel data analysis, though without accounting for potential endogeneity. The findings indicate a positive association between ESG scores and return on assets (ROA). In addition, Reference [14] examines the effect of ESG factors on firm value using structural equation modeling (SEM) and finds mixed results. However, SEM, which is useful in analyzing cross-sectional data, does not capture the complexity of temporal dynamics as effectively as panel data techniques. Reference [19], using multiple regression analysis, finds no significant relationship between ESG and various performance indicators in India, further highlighting the mixed and varied results. Reference [20] argues that due to the inconsistent results regarding the relationship between ESG and firm performance, a contingency perspective is essential to uncover the context and conditions that determine this relationship. Therefore, we scrutinize this relationship in a global context using panel data methodology, addressing endogeneity concerns and analyzing the role of financial constraints, which has not been previously investigated. A critical question in this research area is whether spending on ESG activities depends on “slack resources” [21,22]. The “slack resources theory” assumes that financially successful firms have more resources to make investments in socially responsible activities [23]. Therefore, a company that has performed better financially in previous/current years will also have a higher social performance in current years. References [23,24] find that firms that have available slack resources invest more in ESG activities, achieving higher financial performance, and refer to this as a “virtuous circle”. References [20,25,26] provide evidence that financial slack has a moderating role in the ESG-performance relationship. Unlike previous studies, this study does not use single metrics that measure only one or two individual aspects of slack (e.g., profitability ratios, free cash flow ratio, current ratio, etc.). Instead, we use a comprehensive index to measure whether firms have more or less funds to invest in ESG activities. This index measures external financial constraints using internal financial metrics and market conditions. To our knowledge, Reference [27] conducts the only study on financial constraints in the ESG-performance relationship and finds a negative moderation on financial performance in China.

Against this backdrop, we use a panel data methodology for a global sample of 90 firms over the period 2008–2023 in this study. Overall, we reveal that the effect of ESG on financial performance is positive, with the social pillar proving consistently significant. Interestingly, when we group the firms by their financial constraints using the WW index, we show that the ESG scores of firms with fewer financial constraints are significantly higher than those of the others. However, the impact of ESG on performance is only significant for the more financially-constrained firms. Our results suggest that the market perceives financially-constrained firms as undervalued, making any positive ESG actions they undertake more noteworthy and impactful, signaling potential for significant improvement and return on investment. Our study offers crucial and timely implications for firms, investors, and regulatory bodies. In practice, our findings suggest that financially-constrained firms can leverage ESG investments to further improve their financial standing, which in turn can reduce their financial constraints. Theoretically, our study expands the understanding of the ESG-firm performance relationship by highlighting the critical role of financial constraints and providing a nuanced view that contrasts with existing theories.

Our contribution is fourfold. First, it provides a focused sector-specific analysis by investigating the ESG-financial performance nexus in the automotive industry, a sector with heightened exposure to environmental risk and ESG scrutiny, yet limited scholarly attention. Second, while prior literature has explored ESG endogeneity using alternative econometric tools, we apply a Two-Stage Least Squares Instrumental Variables (2SLS-IV) framework specifically tailored to this industry, which is neglected in [15,19] and captures the dynamic nature of firm-level ESG reporting and investment outcomes over time. Third, our study is one of the first to incorporate a comprehensive measure of financial constraints (the WW index) to explore how limited financial flexibility conditions the ESG-performance relationship. While Reference [27] touches on this topic, there is no prior research that sheds light specifically on the automotive industry. Finally, by disaggregating ESG into its environmental, social, and governance pillars, we offer actionable insights into which dimensions matter most under financial constraints for investors, policymakers, and firm managers.

This paper is structured as follows: Section 2 reviews the pertinent literature and describes the motivational background of the study. Section 3 presents the materials and methods. Section 4 involves the empirical findings. Section 5 provides a dedicated and detailed discussion. Section 6 concludes the paper.

2. Literature Review and Theoretical Motivation

2.1. Traditional View vs. Revisionist View

The linkage between ESG and firm performance has sparked extensive discussion and ongoing debate in the academic literature. The traditional view, rooted in neo-classical theory, argues that additional costs imposed by engaging in environmentally and socially responsible activities can adversely affect financial performance. These activities often require high initial investments, have long payback periods, and provide limited short-term benefits [3,4,5]. Additionally, stringent regulations can lead to increased operating costs, higher prices, and reduced competitiveness. Reference [28] famously argues that corporate social responsibility (CSR) initiatives can create agency problems, where managers engage in activities that do not align with shareholder interests, potentially diminishing shareholder value [29]. Supporting this perspective, References [30,31] provide empirical evidence of the negative impact of ESG on firm performance.

In contrast, the revisionist view, proposed by [32], contends that environmental regulations and socially-responsible actions can lead to innovation, improve operational efficiency, and ultimately enhance firm value. Regulatory pressures, they argue, can drive firms to innovate and create competitive advantages by reducing overall costs and improving production efficiency. Furthermore, legitimacy theory, stakeholder theory, and the resource-based view suggest that socially-responsible behavior can enhance long-term financial success. These theories posit that firms that address stakeholder concerns, gain legitimacy with the public, and use sustainability as a strategic asset will experience improved performance [5,33,34,35,36,37].

2.2. Slack Resources Theory and Virtuous Circle

While these theories describe the direction of CSR’s effect on corporate performance, Reference [38] posits that firm performance determines the ability to undertake costly social activities. Spending on CSR activities, especially discretionary ones, depends on “slack resources” [21,22]. The “slack resources theory” posits that firms with higher financial performance can be more socially responsible because they have more resources to invest in such activities [23]. Therefore, a firm that has performed well financially in previous years will also perform better socially in current years. Reference [22] shows that the past performance has a stronger correlation with CSR than does subsequent performance. Reference [23] also discovers a correlation between historical financial performance and corporate social performance, supporting the “slack resources theory”. They argue that corporate social performance positively impacts future financial performance, creating a “virtuous circle”. Reference [24] extends these findings, demonstrating a virtuous circle for return on equity, where financial performance two years ahead improves corporate social performance one year ahead, which in turn improves current financial performance. Recent studies by [20,25,26] provide evidence of the moderating role of financial slack in the ESG-performance relationship.

2.3. ESG and Financial Performance: A Brief Literature Overview

The literature on ESG performance begins with studies on “environmental responsibility/performance” and “CSR/performance”, later grouped under sustainability. Although empirical studies yield mixed results, reviews and meta-analyses generally find a positive impact of environmental responsibility on corporate performance [39,40,41,42,43]. Meta-analyses by [40,44,45,46] also reveal a positive relationship between social performance/sustainability performance. Reference [47], who reviews meta-analyses post-2015, supports these findings. References [45,48] show that the environmental concerns have the greatest influence on this positive relationship.

Although a positive association is generally found between ESG and performance in the meta-analyses [45,47], empirical studies present varying results: positive (see [49,50] among others), negative [25,30,31], or insignificant [20,26,51]. More specifically, Reference [49] in the US shows that capital and market distortions negatively affect firm value and sustainable growth, but that institutional quality positively moderates this relationship. References [20,26] find that financial slack and [51] find that competitive advantage positively moderates/mediates the ESG and performance relationship. While most studies test a linear relationship between ESG and performance, References [2,52] provide evidence of a non-linear relationship (inverted U-shape) between ESG and sustainable growth and firm value.

2.4. Industry-Specific Insights and the Gap in the Literature

As highlighted by [2,16,17], the ESG-firm performance relationship varies significantly across industries and is influenced by specific ESG factors. For example, Reference [16] demonstrates that the negative moderating effect of growth opportunities on the ESG-firm value relationship diminishes when a firm operates in a litigious industry. Similarly, Reference [17] emphasizes the importance of environmental scores for the mining and retail sectors in G20 countries. Reference [27], in turn, argues that ESG practices enhance the performance of non-heavy polluting industries in China. Industry-specific studies have shown particular research interest in sectors like energy and tourism [53,54,55,56,57,58,59].

To ensure a comprehensive understanding of this research gap, we conduct a systematic search in databases such as Scopus and WOS using combinations of keywords including “ESG”, “firm performance”, and “automotive industry”. We screen studies published between 2000 and 2024 and conclude that despite the substantial focus on various industries, only a handful of studies explore the ESG-firm performance relationship in the automotive industry, which is highly sensitive to environmental and social risks. The gap is especially pronounced when considering the interplay between “financial constraints” and the ESG-performance relationship, which has been overlooked in prior automotive-specific studies.

To further evidence the research gap, we critically review the existing studies. For instance, Reference [13] analyzes data from over 600 automotive firms globally and finds that stakeholders exert a significant influence on ESG scores. However, their study does not directly examine how ESG governance and stakeholder influence affect financial performance. Reference [15], focusing on the automotive industry in Asia between 2009 and 2020, uses fixed as well as random effects models to establish a positive relationship between ESG (and all its pillars) and ROA, but does not adequately control for endogeneity. Another notable study by [14] employs SEM to examine the relationship between ESG and firm value globally from 2015 to 2020, but the results are inconclusive. More importantly, the SEM methodology is less suited to the complexity of time series data compared to panel data analysis. Finally, none of these studies examine the role of financial constraints in the context of the ESG and performance relationship within the sector.

Our study addresses these gaps by employing a focused industry-level analysis while incorporating financial constraints into the empirical framework.

2.5. Hypothesis Development

Considering the theoretical frameworks and empirical studies discussed, we propose the following hypotheses:

H1. 

ESG performance has a positive impact on firm performance.

This hypothesis is grounded in the “revisionist view”, “stakeholder theory,” and the “resource-based view”, which posit that firms that invest in ESG activities enhance their competitive advantage by satisfying stakeholders, reducing costs through innovation, and improving their reputation, all of which can positively affect financial performance [5,32,33]. The environmental aspect of ESG, in particular, can lead to cost savings and operational efficiency, contributing to better financial outcomes [45,48].

H2. 

ESG performance has a negative impact on firm performance.

In contrast, this hypothesis relies on the “traditional view” and “agency theory.” These frameworks suggest that ESG initiatives may divert resources from core business activities, leading to higher costs, lower profits, and potential conflicts between managers and shareholders. Agency theory suggests that managers might engage in ESG initiatives to advance their own interests or personal reputation, which may not align with shareholder interests, ultimately harming firm value [28,29].

H3. 

ESG scores are higher for financially unconstrained firms.

This hypothesis stems from the “slack resources theory”, which suggests that firms with better financial health are more likely to allocate resources toward ESG initiatives. Financially-unconstrained firms, having greater access to resources, are better positioned to invest in ESG activities, leading to higher ESG scores [22,23]. The theory posits that well-performing firms can afford to act more socially responsible because they have the capacity to invest in such activities.

H4. 

The relationship between ESG performance and firm performance is more positive for financially unconstrained firms.

This hypothesis builds on the concept of the “virtuous circle” and the moderating role of financial slack. Firms with sufficient financial resources are better equipped to invest in ESG activities, which can lead to financial performance improvement. Research by [20,25,26] shows that financial slack positively moderates the ESG and firm performance relationship. However, Reference [27] shows that financial constraints weaken this relationship. Therefore, we expect financially unconstrained firms to exhibit a stronger positive link between ESG and financial performance.

3. Data and Methodology

3.1. Data

Our estimation sample includes all firms operating in the automotive industry that received ESG scores at least once during the sample period 2008–2023. We do not apply any further exclusion criteria; therefore, our data set represents the full set of automotive firms with available ESG data globally over this period. The sample begins in 2008, as this marks the year when most firms first received ESG scores. We choose to limit the sample to 2023 for two reasons. First is due to the availability of ESG reports up to the end of that year at the time of data collection. Second, our instrumental variable analysis incorporates the country-level SDG Index, sourced from the 2024 edition of the Sustainable Development Report, which only provides data up to the year 2023 (see Section 3.2). Importantly, the coverage of the post-COVID era allows us to capture the ESG dynamics shaped by the pandemic, even though some related disruptions are still unfolding.

In this vein, the final data set comprises 90 automotive firms across 23 countries, resulting in 1432 firm-year observations. We adopt an inclusive criterion to classify a company as part of the automotive industry based on The Refinitiv Business Classification (TRBC), which identifies firms under Auto and Truck Manufacturers (29 firms/460 observations), Auto, Truck, and Motorcycle Parts (48 firms/764 observations) and Tires and Rubber Products (13 firms/208 observations).

While the data set is limited to public firms due to the availability of reliable ESG and financial performance data, it includes a diverse range of companies varying in size (e.g., small-cap and large-cap firms) and geography. Specifically, the market capitalization of the firms in our sample ranges from $23.22 billion (minimum) to $299.62 billion (maximum), with a standard deviation of $26.25 billion, ensuring the inclusion of both smaller and larger players within the public automotive sector. Furthermore, the data set captures firms across multiple geographic regions. Japan has the largest number of observations (334). The US and Germany have 224 and 110 total observations, respectively. The sample distribution is given in

Table 1. Sample distribution across countries.

Country	Auto and Truck Manufacturers		Auto, Truck and Motorcycle Parts		Tires and Rubber Products		Total
	Number of Firms	Number of Obs.	Number of Firms	Number of Obs.	Number of Firms	Number of Obs.	Number of Firms	Number of Obs.
Australia			3	48			3	48
Brazil			3	48			3	48
Canada			5	80			5	80
China	3	48	1	16			4	64
France			3	48	1	16	4	64
Germany	4	62	3	48			7	110
Hong Kong	2	30					2	30
India			2	32	1	16	3	48
Italy			1	16			1	16
Japan	10	160	9	142	2	32	21	334
South Korea	1	16	2	32	1	16	4	64
Malaysia	1	16					1	16
Netherlands	1	16	1	16			2	32
Singapore	1	16					1	16
South Africa			1	16			1	16
Spain			1	16			1	16
Sweden			2	30			2	30
Switzerland					1	16	1	16
Taiwan	1	16			3	48	4	64
Thailand					1	16	1	16
Turkey	2	32			2	32	4	64
UK			1	16			1	16
US	3	48	10	16	1	16	14	224
Total	29	460	48	764	13	208	90	1432

Note: This table provides information regarding the distribution of observations across countries.

.

The exact number of observations relies on the variables, as observations are missing for some of the firms. In other words, our analysis is built on an unbalanced panel data set as the number of observations varies across firms and years due to missing data for some variables. No imputation methods were used. This approach allows us to include all available data points and thus maximize the use of our data set.

We investigate how the performance of automotive firms is influenced by ESG scores. Our variables of interest are explained in

Table 2. Variable definitions.

Variable	Annotation	Definition
MB	Market-to-book ratio	Market capitalization/total equity (year-end)
ESG	ESG score	Aggregate ESG score
ENV	ENV score	Environmental pillar score
SOC	SOC score	Social pillar score
GOV	GOV score	Governance pillar score
SIZE	Size	Log of total assets (year-end)
LEV	Leverage	Total debt outstanding/total assets (year-end)
DPR	Dividend payout ratio	Gross dividends-Common stock/income available to common excluding extraordinary items (year-end)
ROA	Return on assets	After-tax income/total assets (year-end)
OPER	Operating profitability	EBIT/total assets (year-end)
RD	Research and development expense	Expenses for research and development/total assets (year-end)
INTAN	Intangible assets	Intangible assets/total assets (year-end)

Note: This table explains the definitions of variables of interest in the study.

. We obtain all the data from the LSEG (previously, Thomson-Reuters Refinitiv Eikon) database, a leading global financial markets data provider. LSEG compiles its data from a variety of credible public sources, which include annual reports, sustainability reports, regulatory filings, and supplementary information from third-party data providers.

All variables are winsorized at the 1st and 99th percentiles to alleviate the potential impact of outliers.

Table 3. Descriptive statistics.

	Obs.	Mean	Median	St. Dev.	Max.	Min.
MB	997	1.791	1.367	1.343	12.378	0.088
ESG	997	47.448	47.935	21.222	88.503	5.228
ENV	997	48.278	48.697	28.915	96.258	0.000
SOC	997	45.612	42.701	25.280	93.570	1.436
GOV	995	49.385	48.075	20.591	91.848	7.259
SIZE	997	22.778	22.673	1.794	26.655	16.493
DPR	997	0.373	0.283	0.448	3.135	0.000
LEV	997	0.249	0.246	0.146	0.619	0.003
R&D	997	0.025	0.022	0.021	0.085	0.000
INTAN	997	0.039	0.019	0.045	0.194	0.000
ROA	997	5.401	4.913	5.431	22.342	−15.321
OPER	997	0.073	0.064	0.061	0.296	−0.231

Note: This table summarizes the basic statistics of the variables of interest within the observation sample used in this study.

contains the descriptive statistics of our observation sample.

The difference in the observation numbers compared to the estimation sample reflects the unbalanced data set structure. The mean value of MB is 1.791. Looking at the ESG and pillar scores, the ESG, ENV, SOC, and GOV mean values are 47.448, 48.278, 45.612, and 49.385, respectively. The data suggest that automotive firms receive the highest and lowest scores for their environmental activities, showing the greatest variation compared to the other ESG pillars.

We present the correlation coefficients in Panel A of

Table 4. Correlation matrix and VIF analysis.

Panel A
Variables	MB	ESG	ENV	SOC	GOV	SIZE	DPR	LEV	RD	INTAN	ROA	OPER
MB	1.000
ESG	−0.166 *	1.000
ENV	−0.241 *	0.875 *	1.000
SOC	−0.096 *	0.922 *	0.698 *	1.000
GOV	−0.031	0.583 *	0.281 *	0.437 *	1.000
SIZE	−0.359 *	0.578 *	0.628 *	0.478 *	0.213 *	1.000
DPR	−0.062	−0.090 *	−0.095 *	−0.100 *	0.017	−0.144 *	1.000
LEV	−0.086 *	0.292 *	0.223 *	0.273 *	0.226 *	0.287 *	−0.012	1.000
RD	−0.043	0.226 *	0.355 *	0.128 *	−0.012	0.152 *	0.042	−0.149 *	1.000
INTAN	−0.075	0.122 *	−0.010	0.130 *	0.268 *	−0.101 *	0.071	0.056	−0.123 *	1.000
ROA	0.461 *	−0.173 *	−0.183 *	−0.146 *	−0.067	−0.224 *	0.040	−0.351 *	−0.092 *	0.001	1.000
OPER	0.375 *	−0.130 *	−0.166 *	−0.078	−0.062	−0.188 *	0.004	−0.207 *	−0.018	0.053	0.650 *	1.000
Panel B
Mean VIF	1.290

Note: This table provides the pairwise correlations between the variables of interest within the observation sample used in this study. * denotes 1% significance level.

, which shows that all ESG scores are highly correlated with each other. Surprisingly, almost all correlation coefficients between financial performance and ESG indicators are significantly negatively correlated. Our heuristic analysis shows that these two sets of variables are also negatively related. However, these preliminary findings should be interpreted with caution as they may change due to the panel structure of our data set and the presence of other covariates in the empirical models. On the other hand, we perform a Variance Inflation Factor (VIF) analysis to ensure that multicollinearity is not a significant concern among variables. The results reported in Panel B of Table 4 demonstrate that the VIF values are all below the commonly accepted levels and are available upon request.

3.2. Methodology

We carry out all estimations within our empirical framework by Stata 17. Since the Hausman test justifies the use of fixed-effects over random-effects (p-values < 0.01 for all models, we begin our analysis with a fixed-effects panel regression to examine how ESG valuations affect firm performance. The fixed-effects model is particularly appropriate for this study as it helps to control for unobserved heterogeneity that is specific to each firm but remains constant over time. By using firm-specific fixed effects, we can account for factors such as management style, corporate culture or firm-level characteristics that might otherwise distort the results. Additionally, fixed-effects panel regression is beneficial in controlling for omitted variable bias arising from time-invariant characteristics. It allows us to focus on within-firm variation rather than between-firm variation. This is important as many firm-level characteristics (such as industry, location, or size) can remain relatively constant but still influence the relationship between ESG and financial performance. By applying this methodology, we ensure that the results reflect the impact of ESG changes within a company rather than differences between companies. Fixed-effects also allow us to control for year-specific effects to capture macroeconomic shocks, regulatory changes or industry-wide trends that could affect all companies in a given year.

We estimate the following model to examine how ESG scores influence corporate performance:

$${PERF}_{i,t}=\alpha + {\beta }_1{ESG}_{i,t}+ {\beta }_2{CONTROL}_{i,t}+{\beta }_3\sum {(year effects)}_t+ {\epsilon }_{i,t}$$

(1)

In our baseline Equation (1), ${PERF}_{i,t}$ denotes the MB for firm i in year t. MB represents the firm value and is defined as the ratio of the market capitalization for the fiscal period to the total equity for the same period.

${ESG}_{i,t}$ is the independent variable to measure the ESG performance of firm i in year t, taking into account the aggregate ESG scores as well as the individual pillar scores. Since ESG scores amalgamate components that have already been shown in the literature to influence financial performance individually, we opt to use ESG scores generated by the LSEG database as a proxy for sustainability to test our hypotheses. These scores are calculated using verifiable, publicly available data. The overall ESG score is based on three pillars, namely environmental, social, and governance. However, LSEG evaluates companies over more than 400 metrics, which are then categorized under ten categories. Resource use, emissions, and innovation are the categories of the environmental pillar. Workforce, human rights, community, and product responsibility are the categories of the social pillar. The categories of the governance pillar are management, shareholders, and CSR strategy. The scores of these categories are combined to form the pillar scores. Each category has a different weighting in the different sectors. The pillar score is the relative sum of the category weights. While the environmental and social pillar scores show the relative performance of firms compared to other firms in the same sector, the governance pillar score shows the relative performance compared to firms in the same country of origin. LSEG’s methodology broadly aligns with the key ESG disclosure framework of the Global Reporting Initiative (GRI) in terms of covering key ESG areas and providing comprehensive sustainability data.

${CONTROL}_{i,t}$ are the firm-specific control variables representing size, leverage, profitability, and capacity. We include these variables because of their importance as identified in the literature exploring the relationship between ESG and firm performance.

For robustness, we extend our empirical analysis by considering the problem of endogeneity [60] using the 2SLS-IV method. This method is especially effective in addressing endogeneity, which arises when explanatory variables are correlated with the error term, resulting in biased and inconsistent estimates. The 2SLS technique tackles this issue by employing IVs that are correlated with the endogenous regressors but uncorrelated with the model’s error term, thereby yielding more reliable parameter estimates.

Identifying a valid IV that correlates with the endogenous regressor, i.e., ESG scores, while remaining unrelated to the dependent variable, i.e., firm performance, poses a significant methodological challenge in the context of 2SLS estimation. A valid IV must satisfy two key criteria: the relevance condition, which requires that the IV be significantly correlated with ESG scores after accounting for model controls, and the exclusion restriction, which demands that the IV influence firm performance only through its effect on ESG scores, conditional on the full set of controls. Following the approach outlined by [61,62,63], this study employs the industry-level ESG average as the IV. This is computed as the arithmetic mean of ESG (or pillar-specific) scores across all firms in a given year, excluding the focal firm (i.e., firm i). The justification for using this IV stems from the observation that firms tend to align their ESG disclosures with industry peers due to shared operational structures and investment environments. Nevertheless, the industry average ESG score is unlikely to have a direct effect on a specific firm’s performance. Moreover, although firm performance could affect its own ESG disclosures, it is implausible for such firm-level outcomes to systematically influence the broader industry average. These considerations support the use of the ESG industry average as a credible IV: it exhibits a strong correlation with firm-level ESG scores while plausibly satisfying the exclusion restriction with respect to firm performance. The validity of the IV used in this study is evaluated through the utilization of two statistical tests: the Kleibergen-Paap rk LM statistic for the under-identification test and the Hansen statistic test for the over-identification test. These tests provide robust assessments of the validity and integrity of the IV in our analysis. Following the 2SLS-IV approach, we conduct instrumental variable regressions in two stages. In the first stage, we predict the ESG performance of the firms using the instrument and other control variables. In the second stage, we utilize predicted ESG performance from the first stage as our main variable of interest to estimate its influence on firm performance. We use the xtivreg2 routine written by [64] to employ the 2SLS-IV estimation method.

Although prior research makes use of ESG/pillar industry means as instruments (see, among others, [61,62,63,65,66,67], industry-wide ESG means could be influenced by intra-industry dynamics [68], which could result in these instruments not being fully exogenous. Inspiring from [69], who demonstrates a positive relationship between a country’s SDG Index score and firm’s ESG performance, we therefore pursue country SDG Index scores as a second alternative instrument to further test the robustness of the results. The SDG Index measures a country’s achievement progress towards the 17 SDGs set by the UN [70] and quantifies the percentage of goals achieved, with a maximum score of 100 indicating full achievement of all SDGs. We collect these scores from the 2024 Sustainable Development Report [70]. It is plausible to anticipate that firms operating in countries that have made significant progress in achieving SDGs might possess improved ESG performance [71]. However, to gauge the relevance of the instrument, we strictly follow [72] and instrument the average SDG Index score of the neighbouring countries. For instance, we use the average of Germany, Switzerland, Italy, and Spain’s SDG Index scores as an instrument for the SDG Index for France. In cases where a country does not share a land border with any other nation, we instead compute the average SDG Index scores of the geographically or economically closest countries. Previous research indicates that this instrument meets the relevance criterion [72,73]. Although the exclusion condition cannot be directly tested, there is no apparent reason why firm performance should be linked to SDG advancements in a neighboring country, except through its influence on the SDG progress of the country where the firm is located.

Next, we follow the slack resources theory and consider financial constraints to answer the question of whether a firm with available financial resources can improve its ESG scores. We proxy the financial constraints by the WW index [74], which is given as follows:

WW index = −0.091 × CF − 0.062 × DIV + 0.021 × LTD − 0.044 × TA + 0.102 × ISG − 0.035 × SG

(2)

In Equation (2), CF is cash flows divided by total assets; DIV is a dummy capturing whether the firm pays a cash dividend; LTD is long-term debt divided by total assets; LNTA is the log of total assets; ISG is industry-level sales growth, and SG is firm-level sales growth. Following [75,76,77], we extrapolate from the sample using the coefficient estimates reported by [74] instead of re-estimating the structural model using their samples. A firm with greater financial restrictions is indicated by a higher index. In our case, firms are categorized as financially unconstrained if the value of the WW index in a given year is below the median value.

4. Results

4.1. Baseline Results

Table 5. Baseline panel regression results.

Variables	Model 1	Model 2	Model 3	Model 4
ESG	0.009 *** (0.003)
ENV		0.003 (0.002)
SOC			0.005 * (0.003)
GOV				0.005 ** (0.002)
SIZE	−0.531 *** (0.141)	−0.488 *** (0.141)	−0.499 *** (0.140)	−0.447 ** (0.137)
DPR	−0.082 (0.061)	−0.080 (0.061)	−0.078 (0.061)	−0.092 (0.061)
LEV	1.664 *** (0.404)	1.677 *** (0.406)	1.743 *** (0.405)	1.634 *** (0.406)
RD	−5.638 * (3.098)	−5.082 (3.101)	−5.150 * (3.095)	−5.323 * (3.092)
INTAN	4.161 *** (1.283)	3.907 *** (1.283)	4.117 *** (1.290)	3.890 *** (1.279)
ROA	0.056 *** (0.009)	0.055 *** (0.009)	0.057 *** (0.009)	0.055 *** (0.009)
OPER	1.573 * (0.882)	1.607 * (0.887)	1.498 * (0.884)	1.545 * (0.883)
Cons.	12.176 *** (3.098)	11.400 *** (3.121)	11.582 *** (3.100)	10.419 *** (3.038)
Year FE	Yes	Yes	Yes	Yes
R2	0.246	0.245	0.250	0.249
Obs.	922	922	922	920

Note: This table provides the results of our baseline panel regression models. Standard deviations of parameters of estimates are in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

provides the baseline regression results. Due to the unbalanced nature of our data, we check stationarity of the variables using Fisher-type unit root tests. Unreported results show that all variables are stationary. Given the specification test results, to deal with the presence of heteroscedasticity (Stata command: xttest3), autocorrelation (Stata command: xtserial), and correlation among the panels (Stata command: xtcdf) in the model, we use the Driscoll-Kraay [78] estimator (Stata command: xtscc). Qualitatively, the outcomes are comparable to baseline panel regressions. Data are available on request.

A quick inspection of the data reveals a positive association between ESG scores and firm performance. In numerical terms, a one-point rise in ESG scores leads to a 1 percent enhancement in MB. We also find a moderate effect of the SOC and GOV pillar scores on MB. While the coefficient of the ENV pillar is statistically insignificant, the economic significance is noteworthy. Specifically, we find that if the ENV score increases by one standard deviation (i.e., 28.915, see Table 3), MB will increase by about 8.68% (28.915 × 0.003). This point estimate implies that increasing the ENV score by one standard deviation would increase MB by 4.84% relative to the mean MB of 1.791 (Note that the impact of a one standard deviation rise in the ENV variable on the MB variable is computed as: 0.003 (coefficient of ENV reported in Table 5) × 28.915 (standard deviation of ENV as reported in Table 3)/1.791 (mean of MB as reported in Table 3) = 4.84%). These findings provide substantial support for H1. Consequently, our results necessitate the explicit rejection of H2. Note that, in addition to our sector-level analysis, we also run our regressions by sub-sector (i.e., Auto and Truck Manufacturers; Auto, Truck and Motorcycle Parts and Tires; and Rubber Products, as reported in Table 1). While the direction of the ESG-performance relationship remains largely consistent, the statistical significance is questioned, particularly for the “Auto, Truck and Motorcycle Parts” segment. This appears to be due to a substantial reduction in the number of observations when the sample is partitioned, which limits the statistical power of the analysis and increases the risk of drawing unreliable inferences. Unreported results are available on request. Moreover, we also use Tobin q as another performance measure, however, we find that the results are mostly insignificant)

4.2. SLS-IV Results

Table 6. Instrumental panel regression results.

Variables	Model 1a		Model 1b		Model 2		Model 3		Model 4
	First-Stage	Second-Stage	First-Stage	Second-Stage	First-Stage	Second-Stage	First-Stage	Second-Stage	First-Stage	Second-Stage
ESG		0.011 ** (0.005)		0.017 ** (0.008)
ENV						0.006 (0.004)
SOC								0.007 * (0.004)
GOV										0.004 (0.003)
ESG/ENV/SOC/GOV_IND	0.946 *** (0.072)				1.017 *** (0.109)		0.888 *** (0.070)		0.487 *** (0.148)
SDG INDEX SCORE			124.136 *** (10.978)
SIZE	13.343 *** (1.408)	−0.552 *** (0.147)	15.874 *** (1.384)	−0.065 (0.231)	20.573 *** (1.833)	−0.524 *** (0.149)	14.847 *** (1.789)	−0.530 *** (0.146)	5.921 ** (2.391)	−0.446 *** (0.137)
DPR	−0.462 (0.677)	−0.082 (0.061)	0.132 (0.695)	−0.174 *** (0.065)	−1.205 (0.926)	−0.079 (0.061)	−1.110 (0.864)	−0.076 (0.061)	1.752 (1.151)	−0.090 (0.062)
LEV	1.641 (4.377)	1.650 *** (0.405)	1.917 (4.523)	1.167 *** (0.420)	4.171 (5.942)	1.644 *** (0.409)	−9.046 (5.565)	1.757 ** (0.406)	9.960 (7.529)	1.654 *** (0.407)
RD	87.776 ** (34.045)	−5.858 * (3.130)	107.865 *** (34.882)	−7.909 ** (3.361)	78.099 * (46.561)	−5.352 * (3.125)	84.092 * (43.464)	−5.378 * (3.112)	104.191 * (57.857)	−5.187 * (3.101)
INTAN	−39.139 *** (14.250)	4.244 *** (1.295)	−31.777 ** (14.629)	2.703 * (1.413)	−24.379 (19.484)	3.973 *** (1.287)	−68.430 *** (18.187)	4.280 *** (1.309)	−14.441 (24.240)	3.875 *** (1.279)
ROA	−0.093 (0.096)	0.056 *** (0.009)	−0.187 * (0.097)	0.051 *** (0.009)	0.040 (0.131)	0.054 *** (0.009)	−0.249 ** (0.123)	0.057 *** (0.009)	−0.043 (0.160)	0.050 *** (0.009)
OPER	6.758 (9.526)	1.583 * (0.883)	12.541 (9.732)	2.450 *** (0.922)	−2.870 (12.986)	1.664 * (0.891)	14.274 (12.162)	1.478 * (0.885)	12.647 (16.288)	1.541 * (0.883)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Centered R2	0.472	0.132	0.446	0.108	0.401	0.100	0.427	0.161	0.062	0.106
Obs.	920	922	920	922	920	922	920	922	918	920
F-statistics	92.00 ***	14.250 ***	82.85 ***	13.73 ***	68.90 ***	14.04 ***	76.59 ***	13.930 ***	6.82 ***	14.320 ***
Cragg-Donald Wald	176.41 ***		129.11 ***		87.68 ***		161.69 ***		10.98 ***
Anderson LM	145.55 ***		111.77 ***		79.32 ***		135.38 ***		10.84 ***
LM test (p-value)		0.000		0.000		0.000		0.000		0.032
Hansen J-test (p-value)		0.749		0.994		0.754		0.199		0.992

Note: This table provides the results of our instrumental panel regression models. Standard deviations of parameters of estimates are in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

presents the results of our 2SLS-IV panel regression, with the first-stage results displayed in the left-hand columns of each model, where ESG and its respective pillar scores (ENV, SOC, GOV) serve as the dependent variables. Our findings confirm that the industry-level ESG, environmental, social, and governance averages (IVs) are statistically significant predictors of firm-level ESG scores, demonstrating their strong explanatory power. This suggests that firms tend to align their ESG performance with industry peers, which strengthens the instrumental variables’ relevance condition.

Moreover, the first-stage F-statistics are above conventional thresholds, confirming that the IVs are not weak instruments. The Anderson canonical correlation LM statistic indicates that the IVs are not underidentified, reinforcing their appropriateness in our estimation strategy.

Consistent with previous estimates, the second stage results indicate a positive association between ESG and firm performance, thereby supporting H1 after accounting for endogeneity. Among the individual pillars, only the coefficient of SOC is positive and significant, implying that social factors such as workforce policies, community engagement, labor practices, and product safety have emerged as the most immediate drivers of financial performance in the industry. One possible explanation is that the post-pandemic period (2020–2023) intensified stakeholder scrutiny over employee well-being, supply chain resilience, and social responsibility, prompting firms to invest more heavily in these areas.

The significance of the LM statistics alongside the insignificance of the Hansen test suggests that the instruments effectively capture ESG scores without overidentification concerns. As a result, companies that can effectively address social issues, such as ensuring worker well-being, fostering diversity and inclusion, upskilling the workforce, and maintaining strong community relations, may be better positioned to enhance their financial performance.

4.3. Financial Constraints

Panel A of

Table 7. WW index grouping and tests of differences in means.

Panel A—WW Index Grouping
Groups	Obs.	Mean	Median	St. Dev.	Max.	Min.
All firms	997	−1.046	−1.054	0.134	0.014	−1.265
Financially-unconstrained firms	498	−1.120	−1.109	0.053	−1.054	−1.265
Financially-constrained firms	499	−0.972	−1.008	0.149	0.014	−1.055
Panel B—Tests of differences in means
Variables	Financially-unconstrained firms (Obs. = 410)		Financially-constrained firms (Obs. = 587 a)		t statistic
ESG	56.087		41.414		15.438 ***
ENV	61.956		38.724		18.351 ***
SOC	53.658		39.992		11.883 ***
GOV	52.033		47.528		4.328 ***
MB	1.579		2.106		−7.069 ***

Note: This table provides the data pertaining to each group of WW index classification and the results of the test of differences in means of ESG scores and performance of financially-constrained and financially-unconstrained firms. *** denotes 1% significance level. ^a GOV variable has 585 observations for financially-constrained firms.

shows summary statistics for the WW index. Notably, the WW index is a negative value, with more negative values indicating less constrained firms, and less negative (or closer to zero) values indicating more constrained firms [79]. As previously noted, we group firms in terms of their financial constraints based on the median in a given year. Specifically, firms are classified as financially-constrained or -unconstrained each year, based on whether their annual WW index falls above or below the sample median for that year (Note that, over the sample period, the share of Auto and Truck Manufacturers, Auto, Truck, and Motorcycle Parts and Tires and Rubber Products in firm-year observations in the financially-constrained (financially-unconstrained) group is 54%, 36%, and 10% (21%, 64%, and 15%), respectively). We obtain the results in Panel B of Table 7 after splitting the sample, which suggests that the ESG scores of firms with fewer financial constraints are significantly higher than those of firms with greater financial constraints [21,22]. It seems that financially unconstrained firms have more resources available, which allows them to invest significantly in ESG initiatives. Therefore, H3 is not rejected.

Interestingly, however, our results also indicate that the MB, denoting financial performance, of financially-constrained firms is by far higher. In fact,

Table 8. Baseline panel regression results (MB and Financial Constraints).

Variables	Financially-Constrained Firms				Financially-Unconstrained Firms
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
ESG	0.011 ** (0.005)				0.006 (0.004)
ENV		0.005 (0.004)				0.003 (0.003)
SOC			0.007 * (0.004)				0.002 (0.003)
GOV				0.005 (0.003)				0.004 (0.002)
SIZE	−1.161 *** (0.214)	−1.136 *** (0.217)	−1.137 *** (0.214)	−1.058 *** (0.211)	0.172 (0.181)	0.219 (0.176)	0.215 (0.184)	0.236 (0.174)
DPR	−0.199 ** (0.099)	−0.191 * (0.100)	−0.198 ** (0.099)	−0.212 ** (0.100)	0.144 ** (0.066)	0.146 ** (0.066)	0.152 ** (0.066)	0.140 ** (0.066)
LEV	2.464 *** (0.582)	2.554 *** (0.583)	2.521 *** (0.582)	2.429 *** (0.589)	0.675 (0.531)	0.576 (0.544)	0.730 (0.540)	0.654 (0.531)
RD	−9.551 * (4.865)	−9.034 * (4.879)	−9.477 * (4.876)	−9.250 * (4.884)	−2.197 (3.445)	−1.817 (3.446)	−1.358 (3.402)	−2.112 (3.423)
INTAN	4.408 ** (1.892)	4.206 ** (1.898)	4.501 ** (1.901)	4.207 ** (1.898)	2.568 (1.635)	2.329 (1.626)	2.364 (1.640)	2.419 (1.623)
ROA	0.054 *** (0.014)	0.055 *** (0.014)	0.055 *** (0.014)	0.052 *** (0.014)	0.069 *** (0.012)	0.068 *** (0.013)	0.071 *** (0.013)	0.070 *** (0.012)
OPER	3.178 ** (1.428)	3.083 ** (1.434)	3.118 ** (1.431)	3.263 ** (1.438)	1.411 (1.102)	1.611 (1.115)	1.391 (1.111)	1.353 (1.102)
Cons.	25.221 *** (4.540)	24.831 *** (4.611)	24.791 *** (4.538)	23.138 *** (4.496)	−3.938 (4.198)	−4.917 (4.119)	−4.813 (4.266)	−5.336 (4.066)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
R2	0.215	0.214	0.216	0.211	0.044	0.025	0.044	0.038
Obs.	519	519	519	517	403	403	403	403

Note: This table provides the results of our baseline panel regression models for financially-constrained and financially-unconstrained firms. Standard deviations of parameters of estimates are in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

shows that financially-constrained firms seem to have a more positive and stronger relationship between ESG performance and firm performance. These results lead us to reject H4. It is worth noting that our unreported descriptive statistics reveal a notable difference in the distribution of the MB ratio between financially-constrained and unconstrained firms. Specifically, the standard deviation of MB for unconstrained firms is considerably lower than that for constrained firms (0.988 vs. 1.567), indicating less dispersion and variation in firm valuations within this group. This structural difference implies that the explanatory power of ESG and other predictors is naturally limited for unconstrained firms, as there is less cross-sectional variation to explain. The results of our instrumental panel regression analysis, presented in

Table 9. Instrumental panel regression results (MB and Financial Constraints).

Variables	Financially-Constrained Firms				Financially-Unconstrained Firms
	Model 1	Model 2	Model 3	Model 4	Model 1	Model 2	Model 3	Model 4
ESG	0.016 * (0.009)				0.007 (0.005)
ENV		0.009 (0.007)				0.008 * (0.004)
SOC			0.013 * (0.007)				0.000 (0.004)
GOV				0.003 (0.005)				0.003 (0.002)
SIZE	−1.204 *** (0.224)	−1.202 *** (0.233)	−1.195 *** (0.222)	−1.061 *** (0.211)	0.164 (0.186)	0.176 (0.179)	0.241 (0.189)	0.239 (0.174)
DPR	−0.199 ** (0.099)	−0.184 * (0.100)	−0.197 ** (0.100)	−0.207 ** (0.100)	0.143 ** (0.066)	0.137 ** (0.067)	0.152 ** (0.066)	0.143 ** (0.067)
LEV	2.426 *** (0.585)	2.559 *** (0.584)	2.499 *** (0.584)	2.479 *** (0.597)	0.674 (0.531)	0.407 (0.554)	0.692 (0.544)	0.663 (0.531)
RD	−9.840 ** (4.891)	−9.162 * (4.892)	−9.940 ** (4.913)	−9.105 * (4.894)	−2.300 (3.489)	−2.739 (3.495)	−1.258 (3.407)	−1.853 (3.431)
INTAN	4.507 ** (1.900)	4.227 ** (1.902)	4.754 ** (1.923)	4.199 ** (1.899)	2.603 (1.645)	2.463 (1.633)	2.268 (1.648)	2.367 (1.624)
ROA	0.054 *** (0.014)	0.056 *** (0.014)	0.055 *** (0.014)	0.053 *** (0.014)	0.069 *** (0.012)	0.064 *** (0.013)	0.070 *** (0.013)	0.070 *** (0.012)
OPER	3.199 ** (1.430)	3.040 ** (1.438)	3.109 ** (1.435)	3.203 ** (1.443)	1.406 (1.102)	1.858 * (1.128)	1.442 (1.114)	1.383 (1.103)
Cons.	26.022 *** (4.698)	26.169 *** (4.910)	25.893 *** (4.686)	23.252 *** (4.504)	−3.773 (4.289)	−4.015 (4.164)	−5.353 (4.366)	−5.381 (4.068)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Centered R2	0.237	0.245	0.233	0.056	0.159	0.183	0.183	−0.931
Obs.	519	519	519	517	403	403	403	403
F-statistics	37.380 ***	36.640 ***	38.210 ***	31.650 ***	16.530 ***	28.430 ***	26.780 ***	3.440 ***
LM (p-value)	0.000	0.000	0.000	0.000	0.000	0.000	0.001	0.999
Hansen J (p-value)	0.322	0.226	0.291	0.355	0.087	0.003	0.013	1.000

Note: This table provides the results of our instrumental panel regression models for financially-constrained and financially-unconstrained firms. Standard deviations of parameters of estimates are in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

, also confirm these findings.

Table 9 further indicates that the environmental pillar has a significant and positive impact on financial performance for financially unconstrained firms. A justifiable explanation is that investors may prioritize environmental risks in the automotive industry due to their direct and immediate financial and regulatory implications. Companies capable of actively managing and reducing environmental risks without liquidity constraints may be perceived as more resilient and better positioned to sustain long-term performance.

4.4. Additional Analysis

To further address endogeneity concerns, we supplement the 2SLS-IV estimation with the two-step system generalized method of moments (2SYS-GMM) approach, which leverages lagged values of the explanatory variables as instruments to control for potential correlations between the regressors and the error term. This dynamic panel estimation technique is particularly well-suited for data sets featured by a large number of cross-sectional units (N = 90) and a relatively short temporal dimension (T = 14) [80]. However, it is worth noting that while the original Arellano-Bond estimator, known as “difference GMM”, may face certain limitations. For instance, its performance in finite samples could be problematic when the panel comprises a limited number of cross-sectional observations, particularly if the analyzed variables are highly persistent, or close to random walk processes [81,82]. Consequently, lagged levels frequently serve as weak instruments for equations in first differences. In this study, we use the xtabond2 routine written by [83] to employ the 2SYS-GMM estimation method. The routine employs lagged values of the dependent variable and explanatory variables as instruments, combining these techniques to handle endogeneity and achieve consistent, efficient estimates [83]. Specifically, it enables the use of the “system GMM” estimators by including moment conditions for the equations in levels, using appropriately lagged first-differences of the variables as instruments [84]. Such an approach not only accounts for the dynamics of the relationship over time but also mitigates endogeneity, corrects for heterogeneity, and provides robust standard errors [85]. The use of lagged variables as IVs is a widely adopted strategy to mitigate simultaneity and reverse causality concerns. Since lagged values originate from prior periods, they are less susceptible to contemporary firm- or industry-level shocks, strengthening their suitability as IVs [86]. Additionally, lagged variables often serve as strong predictors of current outcomes, fulfilling the relevance condition necessary for IV validity. To evaluate the appropriateness of the instruments and detect potential issues of serial correlation, we conduct the Hansen test for overidentifying restrictions alongside the Arellano-Bond tests for first- and second-order autocorrelation. Our results in Panel A of

Table 10. Additional regression results.

Variables	Panel A	Panel B				Panel C				Panel D
	2SYS-GMM	2SLS IV (Japan Excl.)	2SLS IV (US Excl.)	2SLS IV (Both Excl.)	Baseline (Both Excl.)	2SLS IV (Country FE)	2SLS IV (Country-by-Year FE)	Baseline (Country FE)	Baseline (Country-by-Year FE)	Baseline (One-Firm Country Excl.)
L. MB	0.559 *** (0.004)
ESG	0.002 *** (0.001)	0.011 * (0.006)	0.009 * (0.005)	0.009 (0.005)	0.011 *** (0.004)	0.008 * (0.004)	0.010 * (0.006)	0.009 *** (0.004)	0.009 ** (0.004)	0.011 *** (0.003)
SIZE	−0.159 *** (0.001)	−0.681 *** (0.180)	−0.489 *** (0.130)	−0.637 *** (0.161)	−0.652 *** (0.159)	−0.248 *** (0.060)	−0.189 (0.198)	−0.588 *** (0.147)	−0.197 (0.230)	−0.476 *** (0.152)
DPR	−0.165 *** (0.034)	−0.141 * (0.085)	−0.078 (0.057)	−0.128 (0.083)	−0.129 (0.083)	−0.047 (0.061)	−0.026 (0.062)	−0.199 ** (0.087)	−0.068 (0.087)	−0.108 * (0.063)
LEV	0.358 *** (0.096)	1.729 *** (0.525)	1.800 *** (0.359)	2.012 *** (0.471)	1.982 *** (0.469)	1.343 *** (0.361)	0.743 (0.479)	1.475 *** (0.478)	0.874 (0.594)	1.306 *** (0.421)
RD	−0.643 (0.633)	−9.163 ** (4.610)	−7.020 ** (2.819)	−11.532 *** (4.316)	−11.885 *** (4.278)	−0.067 (2.726)	−5.777 * (3.260)	−16.152 (4.449)	−5.317 (4.768)	−6.602 ** (3.103)
INTAN	0.333 (0.294)	4.986 *** (1.564)	3.071 ** (1.203)	3.678 ** (1.459)	3.791 *** (1.447)	2.489 ** (1. 096)	3.988 ** (1. 607)	3.676 ** (1.534)	3.874 ** (1.896)	4.166 *** (1.416)
ROA	0.031 *** (0.002)	0.065 *** (0.012)	0.059 *** (0.008)	0.074 *** (0.012)	0.073 *** (0.012)	0.062 *** (0.009)	0.043 *** (0.010)	0.068 *** (0.011)	0.058 *** (0.014)	0.053 *** (0.009)
OPER	0.303 (0.198)	1.182 (1.098)	1.608 ** (0.803)	1.008 (1.023)	1.067 (1.019)	2.264 *** (0.826)	2.396 ** (0.939)	1.402 (1.025)	2.007 * (1.083)	1.885 ** (0.912)
Cons.	4.055 *** (0.183)	14.941 *** (3.904)	11.251 *** (2.845)	13.982 *** (3.497)	14.277 *** (3.465)	4.928 *** (1.329)	4.552 (4.330)	13.949 *** (3.197)	4.477 (4.999)	11.083 *** (3.359)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	No	Yes	No	Yes
Obs.	920	621	814	513	513	922	922	922	922	858
Gr./Ins.	90/258
F-statistics	243.54 ***	11.830 ***	15.210 ***	12.690 ***	12.770 ***	11.510 ***	15.190 ***	16.690 ***	3.800 ***	12.430 ***
AR(1) (p-value)	0.002	-	-	-	-	-	-	-	-	-
AR(2) (p-value)	0.479	-	-	-	-	-	-	-	-	-
Hansen test (p-value)	1.000	-	-	-	-	-	-	-	-	-
Centered R2	-	0.176	0.162	0.205	0.255	0.132	0.114	0.217	0.210	0.279
LM (p-value)	-	0.000	0.000	0.000	-	0.000	0.000	-	-	-
Hansen J (p-value)	-	0.187	0.999	0.878	-	0.749	0.997	-	-	-

Note: This table provides the results of our additional regression models. Panel A reports 2SYS-GMM results. Panel B displays 2SLS IV/baseline regression results when Japan and/or the US are excluded from the sample. Panel C depicts the results based on regressions with country/year fixed effects. Panel D reports results where countries with only one firm are excluded. Standard deviations of parameters of estimates are in parentheses. *, ** and *** indicate significance at the 10%, 5% and 1% levels, respectively.

below are qualitatively comparable to those of the 2SLS-IV regressions.

As mentioned earlier, the highest concentration of firms is found in Japan and the US, with 334 and 224, respectively. These two countries account for 38.97% (43.99%) of the estimation (observation) sample. This may raise questions about the validity of our results due to the heterogeneity of the sample size. To mitigate these concerns, we re-run our models by excluding firms from Japan and the US, both individually and together. Our findings in Panel B of Table 10 remain qualitatively unchanged.

Finally, as Table 1 illustrates, our data set includes repeated observations for some firms over time, particularly in countries with a limited number of listed automotive firms (e.g., the UK, Singapore, and Spain). In such cases, the total number of observations may reflect longitudinal data for a single firm, which raises concerns about the confounding of firm-level and country-level effects, and the generalizability of cross-country comparisons. To address this issue, we adopt two levels of control. First, we include country fixed-effects to account for time-invariant structural differences across countries, such as differences in capital markets, institutional quality, or accounting regimes. Second, we implement country-by-year fixed effects to control for time-varying country-specific shocks. The results of this specification are presented in Panel C of Table 10. Furthermore, we have further dropped small country outliers, i.e., countries with only 1 firm, following [87] (See footnote 4 in [87]).

In summary, the economic significance of our findings underscores the central role that ESG investments play in driving performance in the automotive industry. Our results show that MB improves by around 1 to 1.7% for every point increase in the overall ESG score. This underlines that ESG performance has a positive impact on market valuation. This impact is particularly valuable for automotive companies looking to increase their market valuation by addressing sustainability issues, especially social risks. In particular, companies with financial constraints show a stronger positive relationship between ESG efforts and market performance. For example, a one-point increase in ESG score for these companies can lead to an improvement in MB of approximately 1.6% (see Table 9), suggesting that even smaller, targeted ESG investments can generate significant returns. This is in contrast to financially unconstrained companies, where higher ESG scores are observed, but with relatively lower immediate market value impacts due to the significant upfront costs associated with ESG investments. These results suggest that companies with tighter financial constraints may benefit more from ESG investments in the short term, possibly due to the signaling effect of responsible corporate behavior that attracts investors seeking undervalued opportunities.

5. Discussion

Our study confirms a significant positive relationship between overall ESG scores and firm performance, which is consistent with revisionist and resource-based views on ESG strategy [15,27,53,55,57,58]. While [15] documents a positive ESG–ROA link in the Asian automotive sector, our global study extends this perspective by using the MB ratio as the performance metric. The MB ratio captures investor sentiment and forward-looking valuation, suggesting that investors reward firms for their long-term ESG commitment rather than just their retrospective profitability. Besides, although Reference [27] shows that ESG enhances the performance of non-heavy polluting industries in China, the positive relationship in a specific heavy polluting industry underscores the importance of more focused, industry-specific studies.

Notably, we find that the social pillar is the strongest driver of financial performance in the automotive industry. While the governance pillar also shows positive associations in some cases, the environmental pillar does not exhibit a statistically significant effect. This pattern diverges from prior findings, such as [45,48], which emphasized the environmental pillar as the dominant driver of firm performance. Our results suggest that the COVID-19 pandemic may have acted as a structural shock, prompting a re-prioritization of ESG concerns across the industry. Specifically, the pandemic intensified focus on social factors, including employee health and safety, labor practices, supply chain resilience, and community engagement, areas that became vital for operational continuity and reputational strength during and after the crisis. The salience of the SOC pillar likely increased as firms navigated disrupted labor markets, heightened stakeholder scrutiny, and emerging expectations for corporate social responsibility. For example, automotive OEMs (Original Equipment Manufacturers) were compelled to implement stringent workplace safety protocols and ensure fair treatment of suppliers under stress, reflecting a shift from symbolic CSR to more substantive social actions [88]. Community engagement and workforce transformation, such as upskilling and support for digital transitions, also became central to ESG strategies. These shifts may have elevated investor appreciation for firms demonstrating tangible social responsibility, translating into improved market valuation. Our findings corroborate [15], who also highlights the prominent role of social ESG factors in firm valuation, and contribute to a growing understanding of how crisis periods can reshape the hierarchy of ESG priorities within industries.

Our instrumental variable analyses confirm the robustness of these results, even after accounting for endogeneity. The continued significance of the social pillar under these alternative specifications strengthens the argument that social initiatives are not peripheral, but central, to market-based performance outcomes in this sector. These results also contribute a counterpoint to [14], who report mixed findings and emphasize the importance of industry-specific dynamics in ESG research.

The role of financial constraints further deepens the complexity of the ESG-performance nexus. Contrary to our expectations and previous studies suggesting that financial slack enhances ESG effectiveness [20,23,24,25,26,27], we find that financially-constrained firms exhibit a stronger and more positive ESG-performance relationship. Investors appear to reward ESG efforts more generously in firms with fewer financial resources, possibly interpreting such actions as credible signals of commitment and efficiency. These firms may prioritize targeted ESG initiatives, particularly in the social sphere, that generate immediate reputational and operational returns. For instance, even limited investments in worker safety or community outreach can foster goodwill and translate into tangible financial benefits. The statistically significant SOC coefficient in our subsample analysis supports this interpretation. In contrast, financially unconstrained firms may be in earlier stages of large-scale ESG investments, where upfront costs temporarily suppress short-term financial performance. While these firms report higher ESG scores, their MB ratios do not reflect proportional gains, likely due to the delayed realization of returns from capital-intensive initiatives such as environmental innovations or clean technology transitions. This aligns with our interpretation that ESG benefits, particularly for large and financially flexible firms, may materialize in the longer term. Taken together, these findings point to a strategic opportunity for financially-constrained firms. By adopting selective, high-impact ESG strategies, especially those targeting social outcomes, they can improve operational stability, strengthen stakeholder relations, and increase firm value.

6. Conclusions

6.1. Main Findings

This study investigates the relationship between ESG and firm performance in the automotive industry. We use MB as the performance metric to reflect investor sentiment and market expectations for future growth and profitability. Our findings indicate that firms with higher ESG scores generally outperform those with lower scores. This suggests that investors appreciate the long-term advantages of ESG investments in the automotive industry. Among the ESG pillars, the social pillar consistently, and the governance pillar to some extent, demonstrate a significantly positive relationship with firm performance, while the environmental pillar does not exhibit statistical significance. This finding highlights that investors increasingly prioritize social factors in their valuation of automotive firms, likely because these elements, such as labor practices, the well-being of employees, community engagement, and supply chain responsibility, have become more visible and financially relevant in recent years. The COVID-19 pandemic, heightened social awareness, and regulatory shifts have amplified the importance of corporate social responsibility, especially in industries with complex global operations like automotive manufacturing. In contrast, while environmental issues remain important, their financial impact appears less immediate or less discernible in this context. Our findings emphasize the need for firms in the automotive sector to embed robust social strategies into their business models, not only to satisfy stakeholders but also to improve firm valuation and long-term sustainability.

Additionally, we reveal that financially unconstrained companies have more success in improving their ESG scores, while this is not necessarily reflected in their market values. This finding contrasts with the virtuous circle observed by [24] and the moderation effects reported by [27] and other studies that use slack resources as a moderating variable [20,25,26]. In contrast, we demonstrate that investors attribute a higher market value to financially-constrained companies, even though these firms tend to have lower ESG scores. This could indicate that financially-constrained firms use ESG factors strategically to improve performance and mitigate financial constraints. It also suggests that investors favor firms that invest in sustainability issues, allowing financially-constrained firms to exploit this opportunity to enhance their market value. Regarding the individual ESG pillars, our analysis indicates that the environmental pillar is the only one with a significant and positive impact on firm performance for financially unconstrained firms. This could be because environmental risks are perceived by investors as more immediate and financially consequential, leading them to value firms that actively address these challenges without liquidity concerns.

Our study makes the following key contributions. First, it highlights the unique characteristics of the automotive industry that may require customized ESG strategies, providing a broader understanding of the link between ESG and firm performance, which is novel given the limited existing research. Second, unlike prior studies focused on specific regions or countries (e.g., References [15,19], our analysis is more comprehensive and accounts for endogeneity. Additionally, we build upon [14] by incorporating temporal dynamics through panel data and the 2SLS-IV methodology. Finally, we explore the differentiating effect of financial constraints on the ESG-performance relationship, a previously underexplored area in the automotive industry. Compared to previous studies (e.g., References [14,15,19,20], our research highlights the central role of financial constraints as a key moderator influencing the impact of ESG investments on firm performance. By analyzing the individual ESG dimensions, our study reveals which areas are more important under financial constraints. In particular, the finding that the environmental pillar is only significant for companies without financial constraints underlines that the impact of ESG dimensions on performance is not uniform and varies depending on a company’s financial structure.

6.2. Practical Implications

Our findings offer several implications for automotive firms, investors, and regulatory bodies. Rather than focusing solely on the costs of these investments, firms should consider ways to enhance their ESG performance, which aligns with the SDGs. Special attention should be given to the social aspect of ESG scores, as it consistently shows a positive impact on performance. Firms should also monitor and report on their ESG performance against relevant metrics and benchmarks to demonstrate progress, identify improvement areas, and build transparency and accountability with stakeholders. To improve ESG performance, automotive firms should develop a clear strategy for managing ESG risks and opportunities, incorporating relevant policies and practices. This includes identifying and prioritizing key ESG issues, setting ambitious targets, and developing reliable metrics to measure progress. Engaging with stakeholders, including customers, suppliers, employees, and communities, helps firms understand expectations and concerns about ESG issues, identify opportunities for improvement, and build trust. Financially unconstrained firms should continue to focus on ESG investments but manage investor expectations around the timeframe for financial returns. They should communicate the long-term benefits of their ESG initiatives. Firms with greater financial constraints should leverage their financial performance to attract investors who value both financial returns and sustainability efforts. Integrating ESG factors into decision-making processes, such as investment decisions, risk management, and product development, ensures that ESG considerations are fully integrated into company operations. Sustainable management practices, enhancing fuel efficiency, reducing emissions, conserving waste and water, using renewable energy, ensuring supply chain sustainability, and increasing transparency are steps firms can take. Last but not least, collaboration with other companies and stakeholders to address common ESG challenges can drive innovation, share best practices, and build industry-wide momentum for sustainable business practices.

Investors are key stakeholders in sustainable development, and our study shows that they prioritize ESG considerations in their investment decisions. Financially-constrained firms can benefit more from ESG investments as the market perceives positive ESG actions as a signal of significant improvement potential. Although ESG scores can indicate well-managed companies, investors should conduct thorough due diligence. Regulators should protect investors from companies that may attempt to disguise financial fragility through ESG reporting. Measures such as third-party verification of ESG reports, penalties for misreporting, and promoting integrated reporting can help ensure transparency and accountability.

6.3. Limitations and Future Research

One limitation of our study is that it does not take country-specific factors, cultural dimensions, and regulations into account on the ESG-firm performance link. This is mainly because the sample size is relatively small. Further research can explore whether ESG performance is used to mitigate financial constraints or increase innovation, specifically in the automotive industry. Furthermore, while ESG scores provide a useful snapshot of a firm’s sustainability performance, they do not necessarily capture all aspects of a company’s social and environmental impact. Alternative measures of ESG performance, such as specific sustainability metrics, would improve the comparability and validity of the results. In addition, the automotive sector, particularly in the post-pandemic period characterized by significant geopolitical and macroeconomic developments, including the Russo-Ukrainian war, energy market disruptions, the intensification of global climate policy debates and the growing political uncertainty surrounding events such as the US presidential election, is continuously undergoing significant changes that could reshape the relationship between ESG and financial performance. For example, the recent plant closures at Volkswagen and the wider crisis in the European automotive industry have highlighted the sector’s vulnerability to ongoing instabilities. Therefore, while our findings reflect important dynamics within the sample period, they may not fully capture the impact of the most recent developments. Future research could incorporate updated data to examine the additional implications of these developments on ESG strategies and financial outcomes. It is also worth noting that although we provide a sector-level analysis, research could explore ESG impacts within narrower automotive sub-segments (e.g., electric vehicle producers or parts manufacturers), where ESG pressures and stakeholder expectations may vary significantly. Additionally, future studies could investigate the channels through which ESG influences firm performance, such as differentiating firms based on vehicle type (e.g., electric vehicle manufacturers) or country-level sustainability policies and support for green finance. Finally, further exploration of the mechanisms that determine the role of financial constraints on the association between ESG and corporate performance could provide deeper insights into the financial implications of ESG strategies.