This study aims to examine the relationship between corporate environmental, social and governance (ESG) performance and capital-market value creation in China, with a focus on testing for U-shaped effects and exploring the underlying mechanisms that drive this relationship.
Panel data on Chinese A-share listed firms from 2007 to 2022 are used. The study uses fixed-effects panel regression models and a generalized propensity score approach, and also investigates financing constraints and green innovation mechanisms. Besides, conduct subgroup analyses by ownership type, industry pollution intensity and firm size to assess heterogeneity in the relationship between corporate ESG performance and capital-market value creation.
Corporate ESG performance and capital-market value creation exhibit a U-shaped relationship. Firms with very high or very low ESG scores outperform those with moderate ESG engagement in terms of market valuation. The analysis of mechanisms shows that ESG performance has an inverted U-shaped effect on financing constraints and a U-shaped effect on green innovation. Furthermore, the U-shaped effect is more pronounced for private firms, firms in less-polluting industries and smaller firms, whereas it is muted or nonexistent for state-owned enterprises, firms in high-pollution industries and larger firms.
To the best of the authors’ knowledge, this study is the first to provide comprehensive evidence of a U-shaped relationship between corporate ESG performance and capital-market value creation. It identifies financing constraints and green innovation as key channels through which ESG performance creates value. The findings offer novel insights into how context factors – such as ownership structure, industry environmental risk and firm size – modulate the returns on corporate sustainability investments. These insights have practical implications for managers seeking to maximize the value impact of ESG initiatives, investors evaluating corporate ESG efforts and policymakers aiming to support sustainable corporate practices in emerging markets.
1. Introduction
Climate change and sustainability challenges have become urgent global issues in recent years. Extreme weather events, rising sea levels and ecological degradation are directly threatening human and economic systems (Tita et al., 2025). Accordingly, governments and international organizations are pushing for a fundamental shift toward sustainable development (Prabhakar, 2025). Notably, the United Nations’ Sustainable Development Goals (SDGs) emphasize environmental protection. Achieving these goals requires collective efforts from government, society and particularly firms (Abudy, et al., 2023). As firms pursue profits, integrating environmental responsibility and social good into business strategy has emerged as an important indicator of corporate strength and future potential (Khanchel and Lassoued, 2022). With growing public and regulatory attention on climate risks and sustainability, corporate ESG performance has become a key consideration for investors in assessing long-term capital-market value creation (Tang and Yang, 2024).
China’s dual-carbon commitments, which require carbon emissions to peak by 2030 and to reach carbon neutrality by 2060, generate strong regulatory and market pressures for firms to internalize environmental externalities (Xu et al., 2025). At the same time, the rapid evolution of disclosure standards and the growing scrutiny from regulators and institutional investors have further increased the prominence of ESG considerations in China’s capital markets (Yu et al., 2018). Understanding the extent to which corporate ESG performance translates into market-based firm value is both theoretically significant and practically relevant, particularly in view of China’s economic scale and its pivotal role in global sustainability transitions (Yu and Xiao, 2022; Xu et al., 2024).
Prior research indicates that strong corporate ESG performance can contribute positively to capital-market value creation (Hamdouni, 2025). From the perspective of stakeholder theory, firms that fulfill social responsibilities and adopt sound governance practices gain support from key constituencies such as employees, customers, suppliers and local communities. This support can translate into competitive advantage and superior financial outcomes (Tantalo and Priem, 2016). In contrast, the shareholder value maximization view raises concerns about the cost-effectiveness of ESG investments. Some studies suggest that ESG activities may divert managerial resources or result in greenwashing practices that provide no substantial short-term benefits to firm performance and may even hinder growth prospects (Kim and Lyon, 2015; Qi et al., 2023). Consequently, empirical evidence on the relationship between corporate ESG performance and capital-market value creation remains mixed. While many studies document a positive association, others find no significant effect or even a negative effect, suggesting that the relationship is more complex than a simple linear trend (Barnett and Salomon, 2012). More recently, scholars have provided growing evidence that the impact of sustainability initiatives may follow a curvilinear pattern (Barnett and Salomon, 2012; Grassmann, 2021).
This study examines the relationship between corporate ESG performance and capital-market value creation using a sample of Chinese A-share listed firms, and it addresses potential functional-form bias that may underlie prior mixed findings. The analysis evaluates two internal mechanisms, namely financing constraints and green innovation. For financing constraints, we test an inverted U-shaped association that links ESG performance to access to finance and the cost of capital (Cheng et al., 2023; Gillan et al., 2010). For green innovation, we test a U-shaped association that reflects the integration of sustainability into innovation strategy and performance outcomes (Câmara et al., 2025; Houston and Shan, 2022). We also assess heterogeneity across ownership structures, industry pollution intensity and firm size to identify context-specific thresholds and payoffs. Methodologically, we use firm and year fixed effects, multiple definitions of Tobin’s Q and a generalized propensity score approach for continuous treatments to strengthen identification and robustness.
The remainder of the paper is organized as follows. Section 2 reviews the relevant literature. Section 3 develops the hypotheses. Section 4 describes the data and method. Section 5 presents the empirical results. Section 6 provides the discussion, including conclusions and implications for practice and policy.
2. Literature review
Stakeholder theory argues that ESG engagement enhances firm reputation, strengthens relationships with employees, customers, suppliers and investors, and mitigates risks associated with social and environmental externalities, which can translate into competitive advantage and superior financial outcomes (Tang and Yang, 2024; Tantalo and Priem, 2016). Empirical studies also indicate that firms with higher ESG performance benefit from reduced information asymmetry, improved financing conditions and greater resilience during adverse shocks (Yu and Xiao, 2022; Yu et al., 2018). In contrast, the shareholder primacy view emphasizes the potential costs and inefficiencies of ESG initiatives (Zhao et al., 2025). Investments in social or environmental projects may reflect overinvestment or agency problems that divert resources from value-maximizing opportunities, thereby weakening short-term profitability and market valuation (Friedman, 2007; Jensen, 2002; Khanchel and Lassoued, 2022).
Given these competing perspectives, recent work investigates whether the relationship between corporate ESG performance and capital-market value creation is nonlinear (Wanyan and Zhao, 2024). Some evidence points to an inverted U-shaped association, consistent with diminishing returns to responsibility beyond a threshold (Barnett and Salomon, 2012). Other studies document a U-shaped association in which firms with very low or very high environmental investments outperform those with moderate engagement (Grassmann, 2021). In parallel, research on mechanisms proposes that strong ESG performance can operate as a credible signal that improves access to capital and lowers its cost (Cheng et al., 2023; Gillan et al., 2010), while weaker or symbolic engagement may fail to convince investors and can coincide with tighter financing constraints (Apergis et al., 2022; Bai et al., 2022). A complementary line of work links robust ESG commitments to green innovation through eco-efficient technologies and product development, with meaningful gains materializing when sustainability is integrated into strategy and operations (Lee et al., 2022; Chen et al., 2022; Chouaibi et al., 2022).
3. Hypotheses development
Building on the above theoretical perspectives, we propose that the relationship between corporate ESG performance and capital-market value creation is nonlinear. Two key mechanisms underpin this relationship: financing constraints and green innovation. From the shareholder-value perspective, moderate ESG engagement imposes additional costs without generating sufficient benefits or credibility, which may temporarily depress market valuation. From the stakeholder or long-term value perspective, however, once ESG performance reaches a high level, the benefits of enhanced reputation, stakeholder trust and risk mitigation outweigh the costs, ultimately leading to superior capital-market value creation. This reasoning implies that corporate ESG performance may exhibit a U-shaped relationship with market-based firm value:
Corporate ESG performance has a U-shaped relationship with capital-market value creation.
Corporate ESG performance can influence a firm’s financing environment. At low levels of ESG engagement, firms incur minimal ESG-related costs and generally face stable financing conditions. As ESG performance rises to a moderate level, however, firms may encounter tighter financing constraints because increased ESG expenditures and uncertain payoffs reduce the willingness of investors and creditors to provide capital. When ESG performance reaches a high level, signaling a credible long-term commitment, firms tend to regain easier access to financing and may benefit from lower costs of capital. In other words, moderate ESG engagement may exacerbate financing frictions, whereas high ESG engagement can alleviate them. Based on this reasoning, we propose the second hypothesis:
Corporate ESG performance and financing constraints exhibit an inverted U-shaped relationship.
Little incentive to invest in new environmentally friendly technologies or products. At moderate levels, resources may be diverted to compliance activities or symbolic sustainability initiatives, which generate limited innovative output. In contrast, high ESG performance reflects the integration of sustainability into corporate strategy and culture, thereby encouraging substantial R&D investment and external partnerships that foster green patents and technological advances (Soomro et al., 2025). On this basis, we expect ESG performance and green innovation to follow a nonlinear pattern in which significant innovation gains emerge only at higher levels of ESG commitment. Accordingly, we propose the third hypothesis:
Corporate ESG performance and green innovation exhibit a U-shaped relationship.
4. Data and method
4.1 Sample selection and data sources
This study uses panel data on Chinese listed firms covering the period from 2007 to 2022, obtained from the CSMAR and WIND databases. The sample includes all non-financial firms listed on the Shanghai and Shenzhen Stock Exchanges. Firms in the financial industry are excluded because of their distinct reporting standards and ESG considerations. To reduce the influence of outliers, all continuous financial variables are winsorized at the 1% and 99% levels.
4.2 Variable description and measurement
We measure corporate ESG performance using the ESG rating index developed by Sino-Securities Index Information Service (Shanghai) Co., Ltd. The rating is reported on a numeric scale from 1 to 8, where 1 represents the lowest ESG performance and 8 represents the highest, corresponding roughly to letter grades from C to AAA. Each firm’s annual ESG score is used as the ESG performance metric. When multiple rating updates are available within a year, we calculate the annual average score. Higher values indicate stronger ESG performance. As a robustness check, we also construct an alternative metric defined as the annual median ESG score for each firm, and the results remain qualitatively similar.
Consistent with prior studies (Hall, 2018; de Oliveira and Basso, 2023), we use Tobin’s Q as the primary indicator of capital-market value creation. Tobin’s Q is defined as the market value of the firm divided by the replacement cost of its assets. The computation follows the definitions presented in Table 1.
Definitions of Tobin’s Q
| Indicator | Calculation method |
|---|---|
| Q1 | Market value A / Total assets |
| Q2 | Market value A / (Total assets–Net intangible assets–Net goodwill) |
| Q3 | Market value B / total assets |
| Q4 | Market value B / (total assets–net intangible assets–net goodwill) |
| Indicator | Calculation method |
|---|---|
| Q1 | Market value A / Total assets |
| Q2 | Market value A / (Total assets–Net intangible assets–Net goodwill) |
| Q3 | Market value B / total assets |
| Q4 | Market value B / (total assets–net intangible assets–net goodwill) |
A higher Tobin’s Q indicates that the market assigns greater value to a firm’s assets relative to their book cost, reflecting investor expectations of future growth and the value of intangible resources. We adopt Tobin’s Q as the dependent variable because it is a forward-looking, market-based indicator of firm performance. Recognizing that Tobin’s Q can be computed in multiple ways, and following Smirlock et al. (1984), we construct four versions of Q. Q1 serves as the baseline measure, while Q2, Q3 and Q4 incorporate alternative treatments of debt valuation and adjustments for intangible assets and goodwill.
In addition, we include a set of control variables that may influence firm value, guided by prior research on firm valuation and ESG impacts (Bolognesi and Burchi, 2023; He et al., 2024). These controls capture key financial, operational and governance characteristics, allowing us to isolate the effect of ESG performance. The definitions are summarized in Table 2.
Definitions of control variables
| Variable | Abbreviation | Definition |
|---|---|---|
| Leverage | Tl | Total liabilities divided by total assets |
| Cash flow | Cflow | Operating cash flow divided by total assets |
| Firm size | lnSale | Natural logarithm of annual sales revenue |
| Firm age | Age | Natural logarithm of the number of years since founding or listing |
| Board size | Bdnum | Number of directors on the board |
| Ownership concentration | Stock | Percentage of shares held by the top 10 shareholders |
| Variable | Abbreviation | Definition |
|---|---|---|
| Leverage | Tl | Total liabilities divided by total assets |
| Cash flow | Cflow | Operating cash flow divided by total assets |
| Firm size | lnSale | Natural logarithm of annual sales revenue |
| Firm age | Age | Natural logarithm of the number of years since founding or listing |
| Board size | Bdnum | Number of directors on the board |
| Ownership concentration | Stock | Percentage of shares held by the top 10 shareholders |
Table 3 reports the descriptive statistics for the main variables. The average ESG score, measured on a scale from 1 to 8, is 4.148 with a standard deviation of 1.026, indicating substantial variation across firms and over time. The minimum and maximum values are 1 and 8, respectively, covering the full rating spectrum. The squared ESG variable has a mean of 18.250. The mean of Tobin’s Q (Q1) is 2.062 with a standard deviation of 1.355, suggesting considerable heterogeneity in market valuations, with a maximum value of 8.952. Alternative definitions of Tobin’s Q (Q2–Q4) yield mean values of approximately 2.255, 2.602 and 2.843, respectively, with similarly wide distributions.
Descriptive statistics of main variables
| Variables | N | Mean | SD | Min. | Max. |
|---|---|---|---|---|---|
| ESG1 | 38269 | 4.148 | 1.026 | 1 | 8 |
| ESG12 | 38269 | 18.250 | 8.229 | 1 | 64 |
| Age | 38269 | 2.891 | 0.346 | 1.792 | 3.526 |
| Bdnum | 22029 | 8.215 | 1.443 | 5 | 12 |
| Stock | 34557 | 0.589 | 0.154 | 0.234 | 0.907 |
| Tl | 38269 | 0.421 | 0.210 | 0.0505 | 0.929 |
| Cflow | 38265 | 0.047 | 0.071 | −0.170 | 0.249 |
| lnSale | 38252 | 21.470 | 1.465 | 18.370 | 25.660 |
| Q1 | 38269 | 2.062 | 1.355 | 0.850 | 8.952 |
| Q2 | 38269 | 2.255 | 1.528 | 0.887 | 10.070 |
| Q3 | 38269 | 2.602 | 1.927 | 0.837 | 11.790 |
| Q4 | 38269 | 2.843 | 2.151 | 0.876 | 13.300 |
| FC | 28705 | −1.015 | 0.076 | −1.239 | −0.836 |
| lnGenpat | 33591 | 0.353 | 0.758 | 0 | 3.584 |
| Dummy_soe | 37154 | 1.817 | 1.085 | 1 | 8 |
| Dummy_pollu | 38269 | 0.178 | 0.382 | 0 | 1 |
| Dummy_size | 38269 | 0.500 | 0.500 | 0 | 1 |
| Variables | N | Mean | Min. | Max. | |
|---|---|---|---|---|---|
| 38269 | 4.148 | 1.026 | 1 | 8 | |
| 38269 | 18.250 | 8.229 | 1 | 64 | |
| Age | 38269 | 2.891 | 0.346 | 1.792 | 3.526 |
| Bdnum | 22029 | 8.215 | 1.443 | 5 | 12 |
| Stock | 34557 | 0.589 | 0.154 | 0.234 | 0.907 |
| Tl | 38269 | 0.421 | 0.210 | 0.0505 | 0.929 |
| Cflow | 38265 | 0.047 | 0.071 | −0.170 | 0.249 |
| lnSale | 38252 | 21.470 | 1.465 | 18.370 | 25.660 |
| Q1 | 38269 | 2.062 | 1.355 | 0.850 | 8.952 |
| Q2 | 38269 | 2.255 | 1.528 | 0.887 | 10.070 |
| Q3 | 38269 | 2.602 | 1.927 | 0.837 | 11.790 |
| Q4 | 38269 | 2.843 | 2.151 | 0.876 | 13.300 |
| 28705 | −1.015 | 0.076 | −1.239 | −0.836 | |
| lnGenpat | 33591 | 0.353 | 0.758 | 0 | 3.584 |
| Dummy_soe | 37154 | 1.817 | 1.085 | 1 | 8 |
| Dummy_pollu | 38269 | 0.178 | 0.382 | 0 | 1 |
| Dummy_size | 38269 | 0.500 | 0.500 | 0 | 1 |
The financing constraint index (FC) has an average value of −1.015, ranging from −1.239 to −0.836, where higher values represent more severe constraints. Green innovation is proxied by the number of invention patents related to environmental or energy technologies, measured as the natural logarithm of one plus the count of such patents (lnGenpat). The mean of lnGenpat is 0.353, equivalent to about 1.42 patents per firm, with a maximum of 3.584 (about 36 patents).
4.3 Model construction
To formally test the hypotheses, we estimate panel regression models that relate capital-market value creation and the proposed mechanisms to corporate ESG performance. For H1, the baseline specification for firm value is expressed in equation (1):
where:
Qit = Tobin’s Q of firm i in year;
t = capital-market value creation;
ESGit = ESG score; ESGit2 is its squared term;
Xit = vector of control variables;
= firm fixed effects;
= year fixed effects; and
= error term.
5. Empirical results
5.1 Regression results
The baseline relationship between corporate ESG performance and capital-market value creation is examined using the primary regression specification, and the results are reported in Table 4.
The results of baseline regression
| Variables | (1) Q1 | (2) Q1 | (3) F.Q1 |
|---|---|---|---|
| ESG1 | −0.578*** (0.057) | −0.576*** (0.063) | −0.565*** (0.061) |
| ESG12 | 0.054*** (0.007) | 0.064*** (0.008) | 0.063*** (0.008) |
| Tl | 0.205** (0.086) | 0.260*** (0.087) | |
| Cflow | 3.107*** (0.220) | 3.125*** (0.210) | |
| lnSale | −0.325*** (0.017) | −0.358*** (0.016) | |
| Age | 0.219*** (0.033) | 0.141*** (0.033) | |
| Bdnum | −0.028*** (0.006) | −0.025*** (0.006) | |
| Stock | −0.788*** (0.076) | −0.231*** (0.073) | |
| _cons | 3.479*** (0.120) | 10.136*** (0.405) | 10.650*** (0.357) |
| Industry | Yes | Yes | Yes |
| Year | Yes | Yes | Yes |
| N | 38269 | 22010 | 21115 |
| R2 | 0.192 | 0.262 | 0.268 |
| Variables | (1) Q1 | (2) Q1 | (3) F.Q1 |
|---|---|---|---|
| −0.578*** (0.057) | −0.576 | −0.565 | |
| 0.054 | 0.064 | 0.063 | |
| Tl | 0.205 | 0.260 | |
| Cflow | 3.107 | 3.125 | |
| lnSale | −0.325 | −0.358 | |
| Age | 0.219 | 0.141 | |
| Bdnum | −0.028 | −0.025 | |
| Stock | −0.788 | −0.231 | |
| _cons | 3.479 | 10.136 | 10.650 |
| Industry | Yes | Yes | Yes |
| Year | Yes | Yes | Yes |
| N | 38269 | 22010 | 21115 |
| R2 | 0.192 | 0.262 | 0.268 |
Robust standard errors clustered at the industry–year level are reported in parentheses below the regression coefficients. ***, ** and * indicate statistical significance at the 1, 5 and 10% levels, respectively. The same notation applies to the following tables
The baseline regression results reported in Table 4 provide strong evidence of a U-shaped relationship between corporate ESG performance and capital-market value creation. In Column (1), the coefficient on the linear ESG term is negative, whereas the squared term is positive and statistically significant. This finding indicates that capital-market value creation decreases at lower levels of ESG performance but begins to increse once ESG surpasses a critical threshold.
Column (2), which incorporates the full set of control variables together with firm and year fixed effects, yields consistent results: the linear ESG coefficient remains significantly negative, the quadratic term remains significantly positive and the implied turning point is approximately 4.5 on the one-to-eight ESG scale. Firms with ESG scores below this threshold are associated with lower market valuations, while those above the threshold experience higher valuations as ESG improves.
Column (3) introduces a one-period lag of capital-market value creation as the dependent variable to mitigate concerns of reverse causality and to enhance the exogeneity of the controls. The results remain robust, as the coefficients on both the linear and quadratic ESG terms retain their sign, magnitude and statistical significance. This stability suggests that the observed U-shaped relationship is not driven by endogeneity concerns but reflects a persistent association between ESG performance and capital-market value creation.
A formal U-shape test further confirms that the slope is negative at low ESG levels and positive at high levels, with the turning point within the observed range and the curvature highly significant (Fatemi et al., 2024). Taken together, these results provide strong support for H1, which predicts a U-shaped relationship between corporate ESG performance and capital-market value creation.
5.2 Robustness test
We conducted a series of robustness checks to ensure that the results are not sensitive to alternative specifications. The main findings are reported in Tables 5 and 6.
Robustness test I
| Variables | (1) Q2 | (2) Q3 | (3) Q4 | (4) Q1 | (5) Q2 | (6) Q3 | (7) Q4 |
|---|---|---|---|---|---|---|---|
| ESG1 | −0.640*** (0.072) | −0.676*** (0.083) | −0.749*** (0.093) | ||||
| ESG12 | 0.067*** (0.009) | 0.083*** (0.010) | 0.087*** (0.012) | ||||
| Tl | 0.173* (0.095) | −0.186 (0.121) | −0.245* (0.135) | 0.220** (0.087) | 0.191** (0.096) | −0.175 (0.121) | −0.231* (0.135) |
| Cflow | 3.373*** (0.245) | 4.511*** (0.297) | 4.895*** (0.325) | 3.100*** (0.220) | 3.364*** (0.245) | 4.505*** (0.298) | 4.888*** (0.325) |
| lnSale | −0.337*** (0.019) | −0.567*** (0.026) | −0.593*** (0.026) | −0.327*** (0.017) | −0.339*** (0.019) | −0.569*** (0.026) | −0.595*** (0.030) |
| Age | 0.239*** (0.035) | −0.076 (0.054) | −0.062 (0.056) | 0.223*** (0.033) | 0.243*** (0.035) | −0.073 (0.054) | −0.058 (0.056) |
| Bdnum | −0.032*** (0.007) | −0.027*** (0.009) | −0.030*** (0.010) | −0.028*** (0.006) | −0.031*** (0.007) | −0.027*** (0.009) | −0.030** (0.010) |
| Stock | −0.922*** (0.085) | 1.816*** (0.114) | 1.878*** (0.132) | −0.793*** (0.076) | −0.930*** (0.086) | 1.816*** (0.1140) | 1.876*** (0.1310) |
| ESG2 | −0.515*** (0.057) | −0.577*** (0.064) | −0.589*** (0.074) | −0.663*** (0.083) | |||
| ESG22 | 0.057*** (0.007) | 0.061*** (0.008) | 0.073*** (0.009) | 0.077*** (0.010) | |||
| _cons | 10.858*** (0.459) | 15.429*** (0.546) | 16.450*** (0.611) | 10.024*** (0.394) | 10.743*** (0.448) | 15.270*** (0.533) | 16.291*** (0.597) |
| Industry | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| N | 22010 | 22010 | 22010 | 22010 | 22010 | 22010 | 22010 |
| R2 | 0.275 | 0.344 | 0.354 | 0.261 | 0.274 | 0.344 | 0.353 |
| Variables | (1) Q2 | (2) Q3 | (3) Q4 | (4) Q1 | (5) Q2 | (6) Q3 | (7) Q4 |
|---|---|---|---|---|---|---|---|
| −0.640*** (0.072) | −0.676*** (0.083) | −0.749*** (0.093) | |||||
| 0.067*** (0.009) | 0.083*** (0.010) | 0.087*** (0.012) | |||||
| Tl | 0.173* (0.095) | −0.186 (0.121) | −0.245* (0.135) | 0.220** (0.087) | 0.191** (0.096) | −0.175 (0.121) | −0.231* (0.135) |
| Cflow | 3.373*** (0.245) | 4.511*** (0.297) | 4.895*** (0.325) | 3.100*** (0.220) | 3.364*** (0.245) | 4.505*** (0.298) | 4.888*** (0.325) |
| lnSale | −0.337*** (0.019) | −0.567*** (0.026) | −0.593*** (0.026) | −0.327*** (0.017) | −0.339*** (0.019) | −0.569*** (0.026) | −0.595*** (0.030) |
| Age | 0.239*** (0.035) | −0.076 (0.054) | −0.062 (0.056) | 0.223*** (0.033) | 0.243*** (0.035) | −0.073 (0.054) | −0.058 (0.056) |
| Bdnum | −0.032*** (0.007) | −0.027*** (0.009) | −0.030*** (0.010) | −0.028*** (0.006) | −0.031*** (0.007) | −0.027*** (0.009) | −0.030** (0.010) |
| Stock | −0.922*** (0.085) | 1.816*** (0.114) | 1.878*** (0.132) | −0.793*** (0.076) | −0.930*** (0.086) | 1.816*** (0.1140) | 1.876*** (0.1310) |
| −0.515*** (0.057) | −0.577*** (0.064) | −0.589*** (0.074) | −0.663*** (0.083) | ||||
| 0.057*** (0.007) | 0.061*** (0.008) | 0.073*** (0.009) | 0.077*** (0.010) | ||||
| _cons | 10.858*** (0.459) | 15.429*** (0.546) | 16.450*** (0.611) | 10.024*** (0.394) | 10.743*** (0.448) | 15.270*** (0.533) | 16.291*** (0.597) |
| Industry | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| N | 22010 | 22010 | 22010 | 22010 | 22010 | 22010 | 22010 |
| R2 | 0.275 | 0.344 | 0.354 | 0.261 | 0.274 | 0.344 | 0.353 |
Robustness test II
| Variables | (1) Q1 | (2) Q1 | (3) Q1 | (4) Q1 |
|---|---|---|---|---|
| ESG1 | −0.199*** (0.052) | −0.486*** (0.063) | −0.498*** (0.063) | −0.322** (0.099) |
| ESG12 | 0.023*** (0.006) | 0.053*** (0.008) | 0.054*** (0.008) | 0.032** (0.013) |
| Tl | 0.339*** (0.102) | 0.253*** (0.088) | 0.222** (0.090) | −0.107 (0.128) |
| Cflow | 1.494*** (0.188) | 3.038*** (0.222) | 3.088*** (0.226) | 3.782*** (0.324) |
| lnSale | −0.361*** (0.026) | −0.317*** (0.018) | −0.318*** (0.018) | −0.217*** (0.021) |
| Age | 1.128*** (0.125) | 0.180*** (0.034) | 0.177*** (0.035) | 0.180*** (0.048) |
| Bdnum | −0.030*** (0.010) | −0.027*** (0.006) | −0.022*** (0.006) | −0.021** (0.009) |
| Stock | −1.023*** (0.129) | −0.832*** (0.076) | −0.777*** (0.077) | −0.559*** (0.121) |
| _cons | 7.682*** (0.683) | 9.912*** (0.408) | 9.901*** (0.418) | 7.392*** (0.547) |
| Industry | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes |
| Firm | Yes | No | No | No |
| City | No | Yes | Yes | Yes |
| Industry × Year | No | No | Yes | Yes |
| Industry × Year × City | No | No | No | Yes |
| N | 21631 | 22000 | 21910 | 12010 |
| R2 | 0.596 | 0.309 | 0.348 | 0.471 |
| Variables | (1) Q1 | (2) Q1 | (3) Q1 | (4) Q1 |
|---|---|---|---|---|
| −0.199*** (0.052) | −0.486*** (0.063) | −0.498*** (0.063) | −0.322** (0.099) | |
| 0.023*** (0.006) | 0.053*** (0.008) | 0.054*** (0.008) | 0.032** (0.013) | |
| Tl | 0.339*** (0.102) | 0.253*** (0.088) | 0.222** (0.090) | −0.107 (0.128) |
| Cflow | 1.494*** (0.188) | 3.038*** (0.222) | 3.088*** (0.226) | 3.782*** (0.324) |
| lnSale | −0.361*** (0.026) | −0.317*** (0.018) | −0.318*** (0.018) | −0.217*** (0.021) |
| Age | 1.128*** (0.125) | 0.180*** (0.034) | 0.177*** (0.035) | 0.180*** (0.048) |
| Bdnum | −0.030*** (0.010) | −0.027*** (0.006) | −0.022*** (0.006) | −0.021** (0.009) |
| Stock | −1.023*** (0.129) | −0.832*** (0.076) | −0.777*** (0.077) | −0.559*** (0.121) |
| _cons | 7.682*** (0.683) | 9.912*** (0.408) | 9.901*** (0.418) | 7.392*** (0.547) |
| Industry | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes |
| Firm | Yes | No | No | No |
| City | No | Yes | Yes | Yes |
| Industry × Year | No | No | Yes | Yes |
| Industry × Year × City | No | No | No | Yes |
| N | 21631 | 22000 | 21910 | 12010 |
| R2 | 0.596 | 0.309 | 0.348 | 0.471 |
Columns (1)–(3) of Table 5 replace the dependent variable with alternative definitions of Tobin’s Q (Q2–Q4), while keeping other variables unchanged. The results show that the ESG1 coefficient remains significantly negative at the 1% level, and the ESG12 remains significantly positive at the 1% level. This provides strong evidence that the U-shaped relationship between corporate ESG performance and capital-market value creation is robust to different definitions of the dependent variable.
Next, we replace the key explanatory variable with the annual median Sino-Securities ESG rating, denoted ESG2, and reestimate the model using Q1–Q4 as alternative dependent variables. The results, reported in Columns (4)–(7), show that ESG2 is significantly negative at the 1% level and ESG22 is significantly positive at the 1% level.
In addition, to address potential endogeneity concerns arising from omitted variables, we apply a more stringent set of fixed-effect specifications. The corresponding results are reported in Table 6.
Column (1) of Table 4–8 introduces firm fixed effects to eliminate the influence of time-invariant firm-level characteristics on capital-market value creation. Column (2) adds city fixed effects to control for time-invariant city-level heterogeneity. Column (3) further includes industry-by-year fixed effects on the basis of Column (2), thereby addressing industry-level factors that vary over time. Finally, Column (4) incorporates industry-by-year-by-city fixed effects, which provide the most stringent specification by simultaneously controlling for time-varying city-level factors as well as time-invariant heterogeneity at the industry–city level.
Mechanism analysis
| Variables | (1) FC | (2) lnGenpat |
|---|---|---|
| ESG1 | 0.038*** (0.007) | −0.045* (0.025) |
| ESG12 | −0.005*** (0.001) | 0.016*** (0.004) |
| Tl | −0.365*** (0.025) | 0.193*** (0.028) |
| Cflow | −0.059** (0.026) | 0.018 (0.075) |
| lnSale | −0.124*** (0.003) | 0.082*** (0.006) |
| Age | −0.045*** (0.004) | −0.154*** (0.015) |
| Bdnum | −0.004*** (0.001) | 0.015*** (0.003) |
| Stock | 0.083*** (0.011) | −0.033 (0.030) |
| _cons | 3.385*** (0.064) | −1.227*** (0.135) |
| Industry | Yes | Yes |
| Year | Yes | Yes |
| N | 18949 | 21485 |
| R2 | 0.641 | 0.177 |
| Variables | (1) | (2) lnGenpat |
|---|---|---|
| 0.038*** (0.007) | −0.045* (0.025) | |
| −0.005*** (0.001) | 0.016*** (0.004) | |
| Tl | −0.365*** (0.025) | 0.193*** (0.028) |
| Cflow | −0.059** (0.026) | 0.018 (0.075) |
| lnSale | −0.124*** (0.003) | 0.082*** (0.006) |
| Age | −0.045*** (0.004) | −0.154*** (0.015) |
| Bdnum | −0.004*** (0.001) | 0.015*** (0.003) |
| Stock | 0.083*** (0.011) | −0.033 (0.030) |
| _cons | 3.385*** (0.064) | −1.227*** (0.135) |
| Industry | Yes | Yes |
| Year | Yes | Yes |
| N | 18949 | 21485 |
| R2 | 0.641 | 0.177 |
Heterogeneity analysis results
| Variables | (1) State owned | (2) Private firms | (3) High-pollution | (4) Less-pollution | (5) Large size | (6) Small size |
|---|---|---|---|---|---|---|
| ESG1 | −0.143 (0.156) | −0.615*** (0.064) | −0.227** (0.097) | −0.645*** (0.074) | −0.149** (0.065) | −0.476*** (0.088) |
| ESG12 | 0.011 (0.020) | 0.068*** (0.008) | 0.020* (0.011) | 0.072*** (0.009) | 0.026*** (0.008) | 0.040*** (0.011) |
| Tl | 0.100 (0.231) | 0.224** (0.090) | 0.220 (0.154) | 0.204** (0.100) | −0.863*** (0.109) | 0.511*** (0.094) |
| Cflow | 2.680*** (0.597) | 3.116*** (0.222) | 3.022*** (0.468) | 3.133*** (0.245) | 3.286*** (0.251) | 2.695*** (0.264) |
| lnSale | −0.530*** (0.045) | −0.307*** (0.017) | −0.374*** (0.035) | −0.316*** (0.019) | −0.065*** (0.016) | −0.655*** (0.029) |
| Age | 0.089 (0.128) | 0.224*** (0.033) | 0.206*** (0.070) | 0.221*** (0.036) | −0.110*** (0.040) | 0.317*** (0.040) |
| Bdnum | −0.056* (0.030) | −0.025*** (0.006) | −0.016 (0.015) | −0.031*** (0.007) | −0.030*** (0.007) | −0.027*** (0.008) |
| Stock | −0.796*** (0.241) | −0.747*** (0.078) | −0.872*** (0.127) | −0.772*** (0.087) | 0.119 (0.092) | −1.403*** (0.099) |
| _cons | 14.287*** (0.943) | 9.778*** (0.417) | 10.422*** (0.892) | 10.107*** (0.451) | 4.185*** (0.398) | 16.820*** (0.633) |
| Industry | Yes | Yes | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes | Yes | Yes |
| N | 1670 | 20340 | 3480 | 18530 | 8980 | 13030 |
| R2 | 0.434 | 0.258 | 0.249 | 0.264 | 0.259 | 0.331 |
| ESG1(p value) | 0.000 | 0.000 | ||||
| ESG12(p value) | 0.000 | 0.190 | ||||
| Variables | (1) State owned | (2) Private firms | (3) High-pollution | (4) Less-pollution | (5) Large size | (6) Small size |
|---|---|---|---|---|---|---|
| −0.143 (0.156) | −0.615*** (0.064) | −0.227** (0.097) | −0.645*** (0.074) | −0.149** (0.065) | −0.476*** (0.088) | |
| 0.011 (0.020) | 0.068*** (0.008) | 0.020* (0.011) | 0.072*** (0.009) | 0.026*** (0.008) | 0.040*** (0.011) | |
| Tl | 0.100 (0.231) | 0.224** (0.090) | 0.220 (0.154) | 0.204** (0.100) | −0.863*** (0.109) | 0.511*** (0.094) |
| Cflow | 2.680*** (0.597) | 3.116*** (0.222) | 3.022*** (0.468) | 3.133*** (0.245) | 3.286*** (0.251) | 2.695*** (0.264) |
| lnSale | −0.530*** (0.045) | −0.307*** (0.017) | −0.374*** (0.035) | −0.316*** (0.019) | −0.065*** (0.016) | −0.655*** (0.029) |
| Age | 0.089 (0.128) | 0.224*** (0.033) | 0.206*** (0.070) | 0.221*** (0.036) | −0.110*** (0.040) | 0.317*** (0.040) |
| Bdnum | −0.056* (0.030) | −0.025*** (0.006) | −0.016 (0.015) | −0.031*** (0.007) | −0.030*** (0.007) | −0.027*** (0.008) |
| Stock | −0.796*** (0.241) | −0.747*** (0.078) | −0.872*** (0.127) | −0.772*** (0.087) | 0.119 (0.092) | −1.403*** (0.099) |
| _cons | 14.287*** (0.943) | 9.778*** (0.417) | 10.422*** (0.892) | 10.107*** (0.451) | 4.185*** (0.398) | 16.820*** (0.633) |
| Industry | Yes | Yes | Yes | Yes | Yes | Yes |
| Year | Yes | Yes | Yes | Yes | Yes | Yes |
| N | 1670 | 20340 | 3480 | 18530 | 8980 | 13030 |
| R2 | 0.434 | 0.258 | 0.249 | 0.264 | 0.259 | 0.331 |
| 0.000 | 0.000 | |||||
| 0.000 | 0.190 | |||||
5.3 Mechanism analysis
This study turns to the analyses of the proposed mechanisms, the results show in Table 7.
Column (1) of Table 7 uses the financing constraint index (FC) as the dependent variable, where higher values represent tighter financing constraints. The coefficient on ESG1 is positive, while the ESG12 is negative and statistically significant. This pattern indicates an inverted U-shaped association between corporate ESG performance and financing constraints, consistent with H2. At lower levels of ESG engagement, increases in ESG scores are accompanied by a rise in financing constraints, suggesting that firms face greater difficulty in securing external capital. This worsening of financing conditions peaks at moderate ESG levels, where investments in ESG activities may generate costs without sufficiently credible signals to investors. At higher levels of ESG performance, however, financing frictions are alleviated, implying that credible and sustained ESG commitment improves firms’ access to external financing and lowers the cost of capital. This mechanism highlights how high ESG performance enhances capital-market value creation by reducing financial frictions.
Column (2) of Table 7 examines green innovation, proxied by the logarithm of one plus the number of green patents (lnGenpat). Here, ESG1 is negative and ESG12 is positive and significant, providing evidence of a U-shaped relationship between corporate ESG performance and green innovation, in line with H3. At very low ESG scores, firms typically do not engage in green patenting activities, and initial improvements in ESG performance are associated with limited or even negative changes in innovative output. This result suggests that moderate ESG engagement may divert resources toward compliance or symbolic sustainability initiatives, thereby crowding out substantive innovation efforts. Once ESG performance surpasses a higher threshold, however, firms are more likely to integrate sustainability into their strategic orientation, which encourages systematic R&D investment and collaboration. As a result, high ESG performers achieve significantly greater green patenting activity, underscoring the role of ESG in fostering innovation-led value creation.
5.4 Heterogeneity analysis
We further examine whether the relationship between corporate ESG performance and capital-market value creation varies across different types of firms. Specifically, we reestimate the baseline regression by ownership type, industry pollution intensity and firm size. The results are reported in Table 8.
Column (1) shows that for state-owned enterprises (SOEs), ESG performance has no significant effect on capital-market value creation. In contrast, Column (2) indicates that for private firms, ESG performance is significantly negative in the linear term and significantly positive in the squared term, confirming a U-shaped relationship. This finding suggests that investors and market forces exert stronger influence on private firms, rewarding or penalizing them for ESG activities, while SOEs may be less sensitive to investor sentiment due to government ownership and alternative performance expectations.
Columns (3) and (4) report the results by industry classification. Firms in high-pollution industries display a much flatter U-shaped curve, with coefficients smaller in magnitude. By contrast, firms in less-polluting industries exhibit a steeper U-shaped relationship. These results imply that industry context conditions the ESG–value link. In industries with inherently high environmental risks, investors appear to set a higher credibility threshold before ESG investments translate into value gains, leading to a delayed turning point and flatter slope. In cleaner industries, moderate-to-high ESG performance more readily generates reputational benefits and operational efficiencies that are positively valued by the market.
Finally, Columns (5) and (6) present the results by firm size. Large firms exhibit a U-shaped relationship, but the decline at low ESG levels and the subsequent rise at high levels are less pronounced. This attenuation may reflect the fact that large firms have diversified resources and multiple valuation drivers that reduce the marginal influence of ESG. Small firms, by contrast, display a much sharper U-shaped curve. At low and moderate ESG levels, investors remain skeptical of their ESG initiatives, possibly viewing them as costly or distracting from growth opportunities. Only when small firms demonstrate exceptionally strong ESG commitment do they overcome these concerns, thereby enhancing market value.
6. Discussion
6.1 Conclusion
This study provides consistent evidence that corporate ESG performance and capital-market value creation are related in a U-shaped manner. Firms at either very low or very high levels of ESG engagement obtain higher market valuations, whereas firms at intermediate levels of ESG activity underperform. These results highlight that partial or symbolic engagement may impose costs without convincing investors of long-term commitment, while substantive ESG integration is eventually rewarded in the capital market.
Two mechanisms help explain this pattern. First, ESG performance has an inverted U-shaped effect on financing constraints. Moderate ESG efforts exacerbate financing frictions because additional expenditures and uncertain returns weaken short-term profitability and deter capital providers. At higher levels of ESG engagement, however, firms improve their risk profile and credibility, which reduces financing constraints and enhances access to external capital. Second, ESG performance has a U-shaped effect on green innovation. Firms with weak or moderate ESG commitment undertake little substantive innovation and may even divert resources away from R&D. By contrast, firms with strong ESG engagement embed sustainability into corporate strategy, which stimulates significant investment in green patents and technologies.
The results also reveal pronounced heterogeneity across firm characteristics. The U-shaped relationship is evident for private firms but not for state-owned enterprises, suggesting that investor-driven valuation effects are more salient where market discipline dominates. Industry context further conditions the ESG payoff: firms in less-polluting sectors display a steeper U-shape with an earlier turning point, while those in high-pollution sectors experience flatter benefits, implying that higher credibility thresholds must be met before ESG is valued by the market. Firm size also matters. Smaller firms face a deeper and longer value trough at intermediate ESG levels, requiring very high ESG commitment to gain investor trust. Larger firms, by contrast, experience a more modest U-shape, with benefits accruing at relatively lower ESG levels, reflecting their greater resource endowment and diversified valuation drivers.
6.2 Implications for practice and policy
The findings of this study offer several important implications for both practice and policy. From the perspective of corporate management, the results indicate that the creation of market value from ESG engagement requires a credible and sustained commitment. Firms that remain at a moderate level of ESG activity are likely to incur costs without reaping corresponding benefits, which can foster the perception that “ESG does not pay.” Once ESG initiatives are deeply integrated into corporate strategy, however, firms are rewarded through enhanced reputation, improved financing conditions and stronger innovation outcomes. This process is particularly demanding for smaller firms, which face greater skepticism from investors and therefore need more transparent and ambitious ESG engagement to overcome the initial value penalty.
For investors, the evidence highlights the importance of a nuanced assessment of ESG performance. The U-shaped relationship suggests that firms at intermediate ESG levels may be undervalued if they are close to reaching the threshold where the market begins to reward ESG engagement. Conversely, investors should remain cautious toward firms that pursue only symbolic or superficial ESG activities, because these may impose costs without enhancing value. In addition, the results show that ownership structure matters: market-based valuation effects of ESG are more evident in private firms, while ESG activities in state-owned enterprises may be driven primarily by compliance or political objectives, and thus are less likely to be reflected in short-term market valuation.
From a policy perspective, the analysis suggests that market forces alone may not provide sufficient incentives for firms to advance beyond moderate levels of ESG engagement. Firms in this “valley of death” struggle to obtain financing and market recognition despite making genuine improvements. Targeted interventions can help bridge this gap. Instruments such as green credit, subsidies or credit guarantees are particularly valuable in supporting firms that are in transition toward higher ESG performance. The evidence indicates that such measures should be directed primarily toward small private firms in environmentally intensive industries, as these firms face the steepest initial costs and barriers to recognition. Properly designed policies can accelerate their transition to the high ESG stage, where capital markets reward sustainability efforts with enhanced value creation.

