The effect of ESG rating disagreement on stock liquidity: evidence from China Open Access

As Environmental, Social, and Governance (ESG) considerations increasingly shape investment decisions, ESG ratings have emerged as a critical tool for evaluating corporate sustainability (Halbritter & Dorfleitner, 2015). However, ESG ratings often differ significantly across rating agencies, even for the same firm, creating uncertainty for investors and researchers alike (Berg, Kölbel, & Rigobon, 2022; Christensen, Serafeim, & Sikochi, 2022; Rossi, Byrne, & Christiaen, 2024). This divergence—commonly referred to as ESG rating disagreement—has sparked debate about its implications for capital markets. For example, Brandon, Krueger, and Schmidt (2021) found that ESG rating disagreement is positively correlated with stock returns, suggesting a risk premium for companies with high ESG rating disagreements, especially in terms of environmental dimension disagreements. Furthermore, Serafeim and Yoon (2023) discovered that although rating consensus can predict future ESG news and its market reaction, for companies with larger rating disagreements, this predictive ability weakens, thereby affecting the relationship between news and market reactions.

While prior research has explored how ESG rating disagreements affect stock returns, less is known about how these disagreements influence stock liquidity, an essential indicator of market efficiency and corporate financing conditions (Chordia, Roll, & Subrahmanyam, 2008; Chung & Hrazdil, 2010; Fang, Noe, & Tice, 2009). Liquidity determines how easily assets can be traded without affecting their price, making it vital for investors, firms, and regulators. Understanding whether ESG rating disagreement impairs or improves liquidity is both practically relevant and theoretically intriguing.

Although existing research has provided some insights into the potential impact of ESG rating on stock liquidity (Bazrafshan, 2023; He, Feng, & Hao, 2023; Zhang, Hao, Gao, Xia, & Zhang, 2024), the discussion in the current literature on this area remains relatively limited. Existing studies often focus on the impact of ESG rating disagreements on stock returns (Brandon et al., 2021; Serafeim & Yoon, 2023), with direct research on their impact on stock liquidity being relatively scarce.

Importantly, existing studies rarely articulate why and through which specific channels ESG rating disagreement should translate into liquidity outcomes. In particular, little is known about whether ESG rating disagreement affects liquidity primarily by triggering information production from market intermediaries, or whether its impact critically depends on firm-level institutional contexts that shape how ESG signals are interpreted. Understanding these mechanisms and conditions is essential, because ESG rating disagreement is inherently an information phenomenon: its market consequences depend not only on the presence of divergent signals, but also on how these signals are processed, amplified, or discounted by information intermediaries and investors under different institutional and financial environments.

This study aims to fill the gap by examining how ESG rating disagreements affect stock liquidity, particularly in the context of China's capital market. China provides a compelling empirical setting due to the increasing regulatory emphasis on ESG disclosure, the rapid growth of ESG investing, and the presence of heterogeneous institutional characteristics such as varied firm ownership structures and uneven analyst coverage. These unique features make the Chinese market a valuable testing ground for exploring how ESG information asymmetry interacts with investor behavior and liquidity.

Two competing theoretical perspectives offer different expectations about the relationship between ESG rating disagreement and stock liquidity. The information asymmetry view posits that rating disagreements increase market attention by drawing interest from analysts and institutional investors, thereby reducing information gaps and enhancing trading activities (Cheng, Chiao, Wang, Fang, & Yao, 2021; El Ouadghiri, Erragragui, Jaballah, & Peillex, 2022; Jiang, Kim, & Zhou, 2011; Wang, Li, San, & Gao, 2023). Conversely, the market disagreement theory suggests that conflicting ESG signals may heighten investor uncertainty, increase divergence in valuations, and raise transaction costs, which could reduce liquidity (Buraschi, Trojani, & Vedolin, 2014; Kim & Koo, 2023; Sadka & Scherbina, 2007).

From an information economics perspective, ESG rating disagreement does not directly alter trading costs or liquidity conditions; rather, it operates by reshaping the information environment in which investors form beliefs and execute trades. In this context, information intermediaries, such as financial analysts and formal research reports, play a pivotal role in transforming conflicting ESG signals into actionable information. Without such intermediaries, ESG rating disagreement may remain latent noise rather than an economically meaningful driver of trading activity.

Scholars argue that increasing the attention of analysts and research reports may trigger more trading activities and investor interest (Cheng et al., 2021; El Ouadghiri et al., 2022; Jiang et al., 2011). For instance, Aouadi and Marsat (2018) documented that ESG controversies are surprisingly associated with greater firm value by attracting more attention from analysts. Bazrafshan (2023) found that noisy information regarding ESG attracts the attention of institutional investors, leading to increased trading activities, while simplified ESG information draws in retail investors.

Motivated by these contrasting views, this paper investigates whether ESG rating disagreement influences stock liquidity through increased analyst and research report attention, and whether this relationship varies by firm ownership (e.g. state-owned vs. non-state-owned enterprises) and financial condition (e.g. loss-making vs. profitable firms).

Crucially, the liquidity implications of ESG rating disagreement are unlikely to be uniform across firms. Whether rating disagreement attracts trading or suppresses liquidity depends on firm-level institutional features that shape investor trust, risk perception, and information reliance. Ignoring such heterogeneity may obscure economically meaningful variation and lead to incomplete or even misleading inferences about the market consequences of ESG rating disagreement. These contextual factors are particularly relevant in China, where SOEs tend to have different access to capital and regulatory scrutiny, and loss-making firms often attract speculative trading. By investigating these moderating and mediating factors, the study contributes to a more nuanced understanding of ESG's market consequences in emerging markets.

To empirically test our hypotheses, we used panel data on 4,065 A-share listed companies in China from 2009 to 2022. The baseline analyses reveal a significant positive association between ESG rating disagreements and stock liquidity. This finding indicates that although divergent ratings may initially create uncertainty, they ultimately stimulate market attention and trading activities, thereby improving liquidity. Robustness checks, including alternative model specifications and endogeneity tests, consistently confirm the reliability of our results.

Beyond the baseline relationship, our study uncovers the mechanisms and boundary conditions that explain how and when ESG rating disagreements shape market outcomes. Mediation analyses show that analyst attention and research report attention plays a central role in transmitting the effect of rating disagreements to liquidity, underscoring the importance of information intermediaries in capital markets. Further, moderation analyses demonstrate that ownership structure and financial condition significantly influence this relationship: the liquidity-enhancing effect of ESG rating disagreements is stronger for non-SOEs and loss-making firms.

This paper makes several key contributions. First, we advance the literature by providing systematic evidence on the direct relationship between ESG rating disagreements and stock liquidity in an emerging market context. While prior studies often regarded divergent ESG ratings as noise or a source of market uncertainty (Berg et al., 2022; Christensen et al., 2022), our findings suggest that rating disagreement can function as a valuable signal that attracts market attention and stimulates trading activity, thereby improving liquidity.

Second, we extend theoretical insights by identifying the mediating role of analyst and research report attention. Building on information intermediary theory (Healy & Palepu, 2001) and supported by evidence that analyst coverage and formal reports improve liquidity through enhanced information dissemination (Brauer & Wiersema, 2018; Dang, Doan, Nguyen, Tran, & Vo, 2019; Piotroski & Roulstone, 2004; Roulstone, 2003), we demonstrate how intermediaries transform conflicting ESG signals into tradable insights that reduce information asymmetry.

Third, we contribute to a more nuanced understanding of ESG–market dynamics by documenting the moderating effects of ownership type and financial condition. Consistent with research showing that corporate ownership structures alter the market relevance of ESG information (Dai & Wang, 2024; He et al., 2023; Wen, Agyemang, Alessa, Sulemana, & Osei, 2023) and that financial constraints heighten investor reliance on non-financial signals (Chen, Chen, Lobo, & Wang, 2011; Francis, LaFond, Olsson, & Schipper, 2004; Lang, Lins, & Maffett, 2012), we find that rating disagreements have stronger liquidity effects in non-SOEs and loss-making firms. These results highlight the conditional nature of ESG's impact and offer practical insights into when rating divergence matters most for capital markets.

Finally, this study provides important practical implications. For investors, our findings suggest that ESG rating disagreements can serve as trading signals, especially when amplified by analyst and research report attention, indicating periods of heightened liquidity and market activity. For corporate managers, the results underscore the importance of engaging proactively with analysts and ensuring transparent ESG communication, particularly for non-SOEs and financially constrained firms where rating divergence attracts stronger scrutiny. For policymakers and regulators, the evidence supports ongoing efforts to enhance ESG disclosure standards and promote rating consistency across agencies, while also highlighting the need for targeted oversight of firms most sensitive to ESG rating divergence. By linking ESG information quality with market efficiency, our study provides actionable guidance for market participants and regulators seeking to improve transparency, liquidity, and capital allocation in emerging markets.

The remainder of this paper proceeds as follows: Section 2 reviews the relevant literature and develops our hypotheses. Section 3 introduces the data, variables, and research design. Section 4 presents the empirical results, including robustness, mediation, and moderation analyses. Section 5 discusses the findings, highlights theoretical and practical implications, and concludes.

Building on the unresolved debate regarding whether ESG rating disagreement enhances or impairs stock liquidity, we further examine how this effect materializes and under what conditions it is most pronounced. Specifically, we focus on information intermediaries as key transmission mechanisms and firm-level institutional and financial characteristics as critical boundary conditions.

Recent literature has examined the effects of ESG ratings on firm value (Behl, Kumari, Makhija, & Sharma, 2022; Cellier & Chollet, 2016; Fatemi, Glaum, & Kaiser, 2018; Wong et al., 2021; Yu, Guo, & Luu, 2018), highlighting the growing influence of ESG factors in shaping investor perceptions and firm performance. However, significant divergence often exists among ESG ratings provided by different agencies for the same firm (Berg et al., 2022; Brandon et al., 2021; Christensen et al., 2022), which introduces information noise and interpretive uncertainty.

Previous literature related to stock liquidity highlights that liquidity is not only a key factor in measuring capital market efficiency (Chordia et al., 2008; Chung & Hrazdil, 2010) but also a significant driver of firm value (Fang et al., 2009). The study by Luo (2022) emphasized the positive role of ESG information disclosure in enhancing corporate liquidity, revealing the importance of investor confidence in market liquidity. This point was further confirmed in a study by He et al. (2023), who found that ESG performance significantly boosted the stock liquidity of listed companies, especially in non-state-owned enterprises, companies with weaker corporate governance, and companies with lower information transparency.

Theoretically, two contrasting views explain how ESG rating disagreements could influence stock liquidity. The negative impact perspective argues that rating discrepancies create confusion among investors about a firm's true ESG performance, leading to valuation disagreement (Buraschi et al., 2014; Sadka & Scherbina, 2007). In this context, investors may hesitate to trade due to increased uncertainty and risk aversion, particularly under conditions of poor ESG transparency or low confidence in ESG data reliability (Luo, Yan, & Yan, 2023). Additionally, rating disagreements may prompt investors to adopt more cautious investment strategies, thereby reducing transaction frequency and liquidity (Gyönyörová, Stachoň, & Stašek, 2023; Tan & Pan, 2023). This trend could become more pronounced during periods of market turbulence, as investors might adopt a more cautious stance towards market reactions and the reliability of information. This point is supported by the research of Engelhardt, Ekkenga, and Posch (2021), who found that during the COVID-19 crisis, European companies with high ESG ratings were associated with higher abnormal returns and lower stock volatility, suggesting that the stability and reliability of ESG could have a significant impact on investor confidence and market liquidity during turbulent times.

While ESG rating disagreement is often viewed as a source of information noise, a growing body of finance theory suggests that disagreement can also stimulate trading activity and enhance stock liquidity. From a market microstructure perspective, liquidity improves when heterogeneous beliefs motivate investors to trade on divergent interpretations of available information, thereby increasing market participation, trading volume, and market depth (Harris & Raviv, 2015; Hong & Stein, 2007). ESG rating disagreement naturally generates such heterogeneity by providing conflicting assessments of firms' sustainability performance across rating agencies.

First, disagreement among ESG ratings can increase investors' incentives to acquire and process information, leading to belief dispersion and information-based trading. When investors interpret ESG signals differently, they are more willing to trade against one another, which raises trading frequency and improves liquidity. This mechanism is consistent with theories of heterogeneous beliefs and disagreement-driven trading, which show that divergence in opinions can intensify trading activity and enhance liquidity by encouraging participation from both informed and speculative traders (Banerjee, 2011; Diether, Malloy, & Scherbina, 2002; Kandel & Pearson, 1995).

Second, ESG rating disagreement may attract arbitrageurs and short-term traders who seek to exploit temporary mispricing arising from inconsistent ESG assessments across rating agencies. Such disagreement-induced mispricing creates incentives for arbitrage trading, which accelerates price discovery and increases immediacy in trading. Prior studies demonstrate that arbitrage activity and speculative trading play an important role in improving liquidity by tightening bid–ask spreads and increasing market depth (Holden & Subrahmanyam, 1992; Shleifer & Vishny, 1997).

Third, conflicting ESG ratings can stimulate additional information production and dissemination, as market participants attempt to reconcile divergent ESG signals. Increased public information availability reduces information asymmetry faced by liquidity providers, lowers adverse-selection risk, and facilitates liquidity provision. This channel aligns with classic microstructure models showing that greater information transparency improves liquidity by mitigating adverse selection (Easley & O'hara, 2004; Glosten & Milgrom, 1985; Kyle, 1985). Consistent with this view, empirical evidence indicates that intensified information production and investor attention, such as analyst activity and public disclosures, are associated with improved stock liquidity and market depth (Healy & Palepu, 2001; Piotroski & Roulstone, 2004; Roulstone, 2003).

Taken together, the literature suggests competing predictions regarding the liquidity implications of ESG rating disagreement. While disagreement may increase uncertainty and discourage trading, it may also stimulate information-based trading and arbitrage activity that enhance liquidity. Accordingly, we propose the following non-directional hypothesis.

ESG rating disagreement increases the complexity and ambiguity of firms' sustainability information, raising investors' information-processing costs. This heightened uncertainty generates stronger demand for professional interpretation, making financial analysts a key conduit through which ESG disagreement is translated into market-relevant signals. By issuing forecasts, recommendations, and ESG-related commentary, analysts help reconcile conflicting ESG assessments, reduce information asymmetry, and stimulate information-based trading, thereby enhancing stock liquidity (Livnat & Zhang, 2012; Piotroski & Roulstone, 2004).

Drawing on information intermediary theory, analyst attention mitigates information asymmetry and enhances market transparency, thereby facilitating trading activity and liquidity (Healy & Palepu, 2001). In the specific context of ESG rating disagreement, analysts serve as critical information filters: they help investors reconcile conflicting ESG cues and make well-informed trading decisions.

Furthermore, empirical research supports a robust link between analyst activity and market liquidity. For instance, Roulstone (2003) documents that greater analyst following is associated with improved liquidity, reflected in narrower bid–ask spreads and deeper market depth, and that analyst characteristics lead to changes in liquidity measures. In this sense, analyst attention serves as a transmission mechanism through which ESG rating disagreement reshapes the information environment and affects liquidity outcomes. Synthesizing these theoretical and empirical insights, we propose.

Beyond routine analyst following, research reports represent a more formalized and comprehensive form of information production (Brauer & Wiersema, 2018; Hirst, Koonce, & Simko, 1995). When ESG ratings diverge, investors may require structured, in-depth analyses, such as scenario evaluations and cross-firm comparisons, to interpret conflicting signals, particularly for longer-term valuation and risk assessment.

Theoretically, research reports act as an extension of the information intermediary role played by analysts: they package complex signals into digestible, strategic analyses that enhance market transparency. This improved clarity addresses investor uncertainty directly and supports more informed and confident trading decisions. Dang et al. (2019) provide global evidence showing that heightened analyst coverage, including through formal reports, improves stock liquidity by offering more accessible and trustworthy public information. More recent findings reinforce that media coverage and formal disclosures are linked to improved liquidity metrics (Huynh & Dang, 2025).

Consequently, in the presence of ESG rating disagreement, which introduces ambiguity and inverts investor confidence, increased issuance and attention to research reports help reduce uncertainty, improve transparency, and enable more efficient trading. This process supports greater liquidity through narrower spreads, deeper orders, and higher trading volume. Accordingly, research report attention represents a distinct and complementary information channel through which ESG rating disagreement can influence stock liquidity. Therefore, we posit the following hypothesis.

Ownership structure fundamentally shapes investor perception and the corporate information landscape, particularly when evaluating non-financial dimensions such as ESG performance (Villalonga, Tufano, & Wang, 2025). From a liquidity perspective, ownership structure shapes the baseline information environment and investors' reliance on ESG-related signals. For SOEs, implicit government backing and policy support reduce perceived downside risk, dampening the marginal informational content of ESG rating disagreement and muting its impact on trading behaviour (Geng & Pan, 2024). As such, ESG rating divergence may be perceived as less informative or consequential when it involves SOEs, who are often presumed to face lower reputational or financing risks. In contrast, non-SOEs rely more heavily on capital market perceptions for financing and valuation, making investors more sensitive to ESG-related uncertainty and disagreement, which can amplify trading responses and liquidity effects.

Empirical studies reinforce this differential responsiveness. He et al. (2023) demonstrated that ESG ratings significantly enhance stock liquidity in Chinese A-share firms, particularly among firms with weaker political connections. Similarly, Dai and Wang (2024) document that the influence of ESG rating disagreement on corporate risk-taking is significantly greater for non-SOEs, further suggesting that ownership type amplifies the sensitivity to ESG signal discord. Supporting this, broader research on ESG, ownership, and firm performance indicates that the interplay between ESG and corporate outcomes often varies by ownership concentration or structure. For instance, Wen et al. (2023) demonstrated that ownership concentration significantly moderates the impact of ESG performance on debt financing outcomes, a related context where governance structure alters how ESG is capitalized by markets.

Taken together, these theoretical and empirical insights suggest that firm ownership type moderates how ESG rating disagreements influence stock liquidity: non-SOEs, without state anchoring, may suffer more severe liquidity effects when ESG signals conflict. Thus, ownership type conditions the extent to which ESG rating disagreement translates into liquidity changes. Therefore, we hypothesize.

Financial condition is a fundamental determinant of how investors process corporate information and make trading decisions. Firms with poor financial performance, particularly loss-making firms, are generally perceived as riskier, opaquer, and more susceptible to external financing constraints (Chen et al., 2011; Francis et al., 2004). According to information asymmetry theory, when investors face uncertainty about firm quality, they rely more heavily on available signals, including non-financial disclosures such as ESG ratings, to update their beliefs and trading strategies (Healy & Palepu, 2001; Verrecchia, 2001).

In the context of ESG rating disagreements, financial conditions may shape how such disagreements are interpreted by the market. For profitable firms, strong earnings performance can itself act as a credible signal of firm quality, thereby reducing investors' dependence on ESG-related signals. For firms with weak financial performance, ESG rating disagreement interacts with elevated uncertainty about firm viability, intensifying investor attention and speculative trading (Bushee & Miller, 2012; Dhaliwal, Radhakrishnan, Tsang, & Yang, 2012). In such settings, ESG-related disagreement becomes a salient signal that attracts information production and trading activity, thereby strengthening its liquidity impact.

Prior studies support this mechanism. For example, Lang et al. (2012) show that non-financial disclosures and transparency matter more in environments with weaker firm fundamentals, since external observers must rely more on alternative information channels. Similarly, Kim and Verrecchia (1994) argue that when the cost of processing information is high, investors allocate more attention to signals that can potentially reduce uncertainty. Thus, for firms with weak financial conditions, ESG rating disagreements are more likely to be perceived as salient information, attracting analyst attention, enhancing information production, and ultimately improving stock liquidity. Accordingly, financial condition serves as a critical boundary condition that governs how strongly ESG rating disagreement influences stock liquidity. Based on the above theoretical reasoning, we propose the following hypothesis.

In this study, A-share listed companies in the Chinese capital market from 2009 to 2021 are selected as research samples. We filter the original sample as follows: First, companies in abnormal trading status due to special treatment (ST or *ST) are excluded; second, due to the uniqueness of the financial and insurance industry, listed companies in this sector are excluded; and last, samples missing key variable data are removed. Finally, we obtain 35,890 observations from 4,065 companies. To minimize the impact of outliers on the research results, all continuous variables are winsorized at the 1% and 99% tails.

Regarding data sources, the ESG-related data for this study mainly comes from the Hua Zheng ESG ratings provided by WIND database, Bloomberg database, SynTao Green Finance ESG ratings, Hexun website, and the CNRDS database. Other financial and market data are sourced from the CSMAR database. All statistical analyses were conducted in the Stata 16.0 software environment.

We follow the methodology of Avramov, Cheng, Lioui, and Tarelli (2022), Berg et al. (2022), quantifying ESG rating disagreement through the standard deviation of ESG ratings provided by five rating agencies. The process is as follows: First, in the initial data processing phase, ESG rating results from Hua Zheng, Bloomberg, SynTao Green Finance, Hexun, and CNRDS for listed companies are converted into a scoring system to ensure consistency and comparability between different rating agencies. Second, a sorting process is conducted where the ESG scores assigned by each rating agency to companies are ranked annually, with rankings based on the magnitude of the scores, and companies with the same score sharing the same rank. The third step involves standardizing the rankings of companies rated by each agency using the range standardization method to eliminate differences between various rating scales. Finally, pairwise rating disagreements are constructed by calculating the standard deviation of standardized rankings between every two rating agencies for each company, serving as a proxy for rating disagreement. For each company, ten pairs of rating agency combinations are formed, and the average of these pairwise rating disagreements is calculated as the company's annual ESG rating disagreement (EsgDis1) ^[1]. Additionally, another measure of corporate ESG rating disagreement (EsgDis2) is obtained by directly calculating the standard deviation of standardized rankings from five rating agencies.

Stock liquidity is the dependent variable in this study, reflecting the ease of buying and selling stocks in the stock market and serving as one of the key indicators for measuring the efficiency of capital markets. Good stock liquidity can not only reduce transaction costs but also improve the efficiency of price discovery, which is significant for both investors and listed companies (Amihud & Mendelson, 1986). To comprehensively assess the impact of stock liquidity, following Amihud (2002), we derive the stock illiquidity index (Illiquidity) using equation (1) as detailed below:

In Equation (1), $|Ri,t,d|$ is the absolute return rate of stock i on the d^th trading day in year t, and $Volumei,t,d$ represents the trading volume of stock i on that same day. $Daysi,t$ indicates the annual trading frequency of stock i. Essentially, $|Ri,t,d|/Volumei,t,d$ captures the stock's return change per unit of trading volume on the d^th trading day of year t. A higher illiquidity index suggests that each unit of trade exerts a more significant influence on the stock's price, indicating increased transaction costs for investors and reduced liquidity of the stock. To address the potential issue of high skewness and kurtosis in the raw illiquidity measure, following Roosenboom, Schlingemann, and Vasconcelos (2013), we subsequently computed the stock liquidity index (Liquidity) through Equation (2).

To explore the mediating mechanisms, we include two variables.

• Analyst attention (Analyst): Measured as the natural logarithm of one plus the number of analysts issuing forecasts for a given firm in a given year. This measure is widely used in prior research (Aouadi & Marsat, 2018; Cheng et al., 2021) as a proxy for the level of information processing by financial analysts.

• Research report attention (Reports): Measured as the natural logarithm of one plus the number of research reports issued for each firm in each year, obtained from the CSMAR database. This follows the approach of Jiang et al. (2011) and El Ouadghiri et al. (2022), capturing the extent of firm-specific information diffusion through formal reports.

As discussed in the literature review, two moderating variables are introduced in this paper.

• Firm ownership type (SOE): A dummy variable that equals 1 if the company is state-owned, and 0 otherwise. Classification follows the definition provided by CSMAR and aligns with prior studies (He et al., 2023).

• Financial condition (Loss): A dummy variable equals 1 if the firm reports a negative net income in a given year, and 0 otherwise. This captures whether the firm is operating under financial distress or loss, based on standard accounting principles (Kim & Koo, 2023).

Following Brandon et al. (2021), Serafeim and Yoon (2023), Sun, Su, Cai, and Bai (2024), Liang, Sun, Xu, Xiong, and Cai (2024) and Wang, Wang, Dong, and Wang (2024), we incorporate a series of control variables to ensure the accuracy and reliability of our regression analysis, mainly focusing on the financial characteristics of the firm, the corporate governance structure, market participants, market performance and macroeconomic factors. These include: Company Size (Size), Financial Leverage (Leverage), Company Age (FirmAge), Cash Holdings (Cash), Return on Assets (ROA), Market-to-Book Ratio (MB), Board Size (Board), Proportion of Independent Directors (NumIndep), Dual Role (Dual), Executive Compensation (Salary), Institutional Ownership (IO), Audited by Big 4 Accounting Firms (Big4), Stock Returns (Ret), Stock Price Volatility (Sigma), and GDP Growth Rate (GdpGrowth).

By controlling these variables, we aim to capture the effect of ESG rating disagreements on stock liquidity more accurately, excluding other factors that may influence stock liquidity. All continuous variables were winsorized at the 1% and 99% tails to mitigate the impact of extreme values. The data were sourced from the CSMAR and WIND databases. Furthermore, in line with Wang et al. (2023), we include industry and year fixed effects in our baseline model, with specific variable definitions available in Appendix A.

To empirically test our hypotheses, we conducted multivariate regression analyses using panel data techniques. The baseline analysis is performed using industry- and year-fixed effects OLS regression, addressing unobserved heterogeneity across industries and over time. To assess mediation effects, we follow Baron and Kenny (1986) causal step approach to evaluate the significance of indirect effects through analyst and report attention. For moderation tests, we include interaction terms between ESG rating disagreement and the moderator variables (SOE, Loss) and examine their significance in predicting stock liquidity. All standard errors are clustered at the firm level to correct for autocorrelation and heteroskedasticity.

These methods were selected due to their wide acceptance in empirical accounting and finance literature, as well as their suitability for panel data with multiple cross-sectional and time-series observations. This design allows for replicability and transparency in the estimation process.

The general econometric model is specified as

where ESG rating disagreement is measured by ESG rating disagreement 1 (EsgDis1) and ESG rating disagreement 2 (EsgDis2); Liquidity denotes stock liquidity. Our primary focus is on the relationship between ESG rating disagreement and stock liquidity, namely the sign and magnitude of the coefficient α₁. All standard errors in the regression results have been adjusted for clustering at the firm level.

Table 1 provides a descriptive statistic of the main variables used in the regression analysis, including the dependent variable, independent variables, and a series of control variables. The average liquidity (Liquidity) is 3.329, with a standard deviation of 1.228, ranging from −1.468 to 6.423, and a median of 3.356, indicating a wide variation in stock liquidity among the samples. The average value of ESG rating disagreement 1 (EsgDis1) is 0.153, with a standard deviation of 0.094, a minimum of 0.004, a maximum of 0.412, and a median of 0.145, showing significant differences in ESG rating disagreement among different companies. The standard deviation for ESG rating disagreement 2 (EsgDis2) is 0.100, further emphasizing the presence of ESG rating disagreements. Additionally, the descriptive statistics of the control variables are consistent with existing research, indicating that the sample companies cover a wide range of sizes, financial conditions, and governance structures.

Table 2 reports the Pearson correlation analysis results for the main variables. The Pearson correlation coefficients between Liquidity and both EsgDis1 and EsgDis2 are significantly positive at the 1% level, essentially indicating a positive correlation between ESG rating disagreement and stock liquidity. This means that greater ESG rating disagreement is associated with higher stock liquidity, preliminarily supporting the research hypothesis H1.

In this paper, fixed-effect models are employed for regression analysis, with results presented in Table 3. Columns (1) and (3) demonstrate a significant positive correlation between ESG rating disagreements (EsgDis1 & EsgDis2) and stock liquidity when only controlling for year and industry fixed effects. Specifically, the coefficients for EsgDis1 are 0.213 and 0.217, both significant at the 1% level; the coefficients for EsgDis2 are 0.238 and 0.242, also significant at the 1% level. This indicates that greater ESG rating disagreements are associated with higher stock liquidity. Regarding control variables, company size (Size) and return on assets (ROA) show a significant positive correlation with stock liquidity, while financial leverage (Leverage), company age (FirmAge), and board size (Board) show a significant negative correlation, aligning with findings in the existing literature. Additionally, the dual roles of CEO and chairman (Dual) and executive compensation (Salary) positively impact stock liquidity, whereas institutional ownership (IO) and auditing by Big 4 accounting firms (Big4) negatively affect liquidity.

Columns (2) and (4) further control for provincial fixed effects based on columns (1) and (3), showing that the positive correlation between ESG rating disagreements and stock liquidity remains robust. This suggests that the positive impact of ESG rating disagreements on enhancing stock liquidity remains significant even when considering regional differences. Overall, the regression results in Table 3 support Hypothesis H1, indicating a positive correlation between ESG rating disagreements and stock liquidity.

From a theoretical perspective, this finding is consistent with market microstructure and heterogeneous-belief theories, which predict that disagreement among information signals can stimulate information-based trading and improve liquidity by increasing market participation and depth (Harris & Raviv, 2015; Hong & Stein, 2007). Rather than merely increasing noise, ESG rating disagreement appears to function as an information catalyst that encourages belief dispersion and trading activity, supporting the view that disagreement can enhance, rather than impair, market liquidity under certain conditions.

To verify the robustness of our main research findings, we conducted a battery of robustness tests. The results of each test panel are displayed in Table 4.

In Panel A, we employ an alternative model specification A to re-examine the relationship between ESG rating disagreements and stock liquidity. By adjusting Liquidity to the current year and considering the previous year's control variables (Controls_t−1), the results continue to show a significant positive correlation between ESG rating disagreements and stock liquidity. Specifically, the coefficients for EsgDis1 in columns (1) and (3) are 0.141 and 0.144, respectively, both significant at the 1% level; the coefficients for EsgDis2 in columns (2) and (4) are 0.169 and 0.173, respectively, also significant at the 1% level. These results further confirm the preliminary conclusion that ESG rating disagreements have a positive impact on stock liquidity.

In Panel B, we utilize an alternative model specification B, examining the impact of the previous year's ESG rating disagreements (EsgDis1_t−1 & EsgDis2_t−1) as explanatory variables on the current year's stock liquidity. The results indicate that the positive correlation between ESG rating disagreements and stock liquidity remains robust when conducting regression analysis with the ESG rating disagreements from the preceding period. This demonstrates that the impact of ESG rating disagreements on stock liquidity has a certain degree of persistence.

To overcome the potential influence of individual effects on the research conclusions, Panel C introduces firm fixed effects. After controlling for firm fixed effects, the positive correlation between ESG rating disagreements (EsgDis1 & EsgDis2) and stock liquidity remains significant. This indicates that even after controlling for firm characteristics that do not change over time, the positive impact of ESG rating disagreements on stock liquidity still exists, further enhancing the robustness of the research results.

In Panel D, we further control for stock turnover (Turnover_t) to examine the relationship between ESG rating disagreements and stock liquidity. The results show that the positive correlation between ESG rating disagreements and stock liquidity remains significant even after controlling for stock turnover. This suggests that the impact of ESG rating disagreements on stock liquidity is not solely realized through increased transaction frequency but may also involve more complex mechanisms.

In summary, through the robustness tests across these four panels, the main finding of this paper—that there is a significant positive correlation between ESG rating disagreements and stock liquidity—has been further validated. These robustness tests not only strengthen the credibility of the research conclusions but also provide new perspectives for future research.

To address potential endogeneity issues, we employ the instrumental variable estimation method to ensure more accurate and reliable estimation results for the impact of ESG rating disagreements on stock liquidity. Following the methodology outlined by Breuer, Müller, Rosenbach, and Salzmann (2018), and El Ghoul, Guedhami, Kim, and Park (2018), this study adopts industry-year mean of ESG rating disagreement (IV_IndMean) and city-year mean of ESG rating disagreement (IV_CityMean) as instrumental variables for ESG rating disagreements (EsgDis1 & EsgDis2). The rationale behind choosing such an instrumental variable is grounded in its compliance with the two essential conditions for ideal instrumental variables: its relevance to the endogenous variable and its independence from the random error term. Table 5 presents the estimation results using the instrumental variable method, including both the first- and second stage regression analyses.

In the first stage, the results show that both the industry-year mean and city-year mean of ESG rating disagreements are significantly positively related to the firm's ESG rating disagreements, with IV_IndMean coefficients of 0.743 (p < 0.01) and 0.772 (p < 0.01), and IV_CityMean coefficients of 0.911 (p < 0.01) and 0.961 (p < 0.01). This indicates a significant correlation between our chosen instrumental variables and the explained variable, satisfying the relevance condition for instrumental variables.

In the second stage regression, we use the predicted values of ESG rating disagreements from the first stage to explain stock liquidity. The results indicate a significant positive correlation between ESG rating disagreements (EsgDis1 & EsgDis2) and stock liquidity, with the impact coefficients of EsgDis1 on stock liquidity being 0.220 (p < 0.05) and 0.217 (p < 0.05), and the impact coefficients of EsgDis2 on stock liquidity being 0.238 (p < 0.01) and 0.242 (p < 0.01). These results further support H1 that ESG rating disagreements have a positive impact on stock liquidity.

To further explore the mechanisms through which ESG rating disagreements affect stock liquidity, we examine analyst attention (AnalystAtt) and research report attention (ReportAtt) as mediating variables (El Ouadghiri et al., 2022; Jiang et al., 2011). Following Baron and Kenny (1986)'s causal step approach, we explicitly tested the four conditions for mediation, with results reported in Table 6.

• Step 1: The independent variable (ESG rating disagreement) significantly affects the dependent variable (stock liquidity). In Panel A, EsgDis1 has a positive and significant effect on liquidity (coefficient = 0.200, p < 0.01). Similarly, in Panel B, EsgDis2 is positively related to liquidity (coefficient = 0.221, p < 0.01).

• Step 2: The independent variable significantly influences the mediators. As shown in Column (1), EsgDis1 is positively associated with AnalystAtt (coefficient = 0.091, p < 0.10), and EsgDis2 has an even stronger effect (coefficient = 0.138, p < 0.05). ESG rating disagreements are also positively associated with ReportAtt (coefficients = 4.045 and 5.046, both p < 0.01).

• Step 3: The mediators significantly affect the dependent variable while controlling for the independent variable. AnalystAtt is positively related to liquidity (coefficients = 0.107 and 0.057, both p < 0.01). Similarly, ReportAtt significantly improves liquidity (coefficients = 0.005 and 0.004, both p < 0.01).

• Step 4: The direct effect of ESG rating disagreement on liquidity is reduced once the mediators are included in the model. For example, the coefficient of EsgDis1 declines from 0.200 (p < 0.01) to 0.196 (p < 0.01) when AnalystAtt and ReportAtt are added. Likewise, the coefficient of EsgDis2 decreases from 0.221 (p < 0.01) to 0.216 (p < 0.01). This reduction indicates partial mediation.

Taken together, these results confirm that analyst attention and research report attention serve as important channels through which ESG rating disagreements enhance stock liquidity. These mediation results are consistent with information intermediary theory (Healy & Palepu, 2001), which posits that analysts and formal research outputs play a central role in transforming complex or conflicting information into decision-relevant signals.

Our findings suggest that ESG rating disagreement increases the demand for professional interpretation, leading to greater analyst scrutiny and research reporting, which in turn reduces information asymmetry and facilitates informed trading. This mechanism provides a clear economic rationale for how ESG rating disagreement is transmitted into higher stock liquidity.

To delve deeper into the heterogeneity of the impact of ESG rating disagreements on stock liquidity, we conducted cross-sectional analyses, focusing on the moderating effects of firm ownership type (state-owned enterprises, or not) and financial condition (loss-making, or not) on this relationship. Table 7 reports the results of the heterogeneity analyses.

According to the results in Table 7, the positive impact of ESG rating disagreements (EsgDis1 & EsgDis2) on stock liquidity shows significant differences between state-owned enterprises (SOEs) and non-state-owned enterprises. Specifically, the positive impact of ESG rating disagreements on stock liquidity is more pronounced for SOEs than for non-SOEs, with coefficients for EsgDis1 and EsgDis2 of 0.263 and 0.304, respectively (both significant at the 1% level). However, when considering the interaction terms between ESG rating disagreements and firm ownership type, the coefficients for SOE×EsgDis1 and SOE×EsgDis2 are −0.146 and −0.192 (both significant at the 5% level), indicating that the characteristics of state-owned enterprises may weaken the positive impact of ESG rating disagreements on stock liquidity. These results provide support for hypothesis H4.

This pattern is consistent with theories emphasizing the role of the institutional context in shaping information sensitivity. For SOEs, implicit government support and political connections reduce investors' reliance on ESG-related signals, weakening the liquidity response to rating disagreement. In contrast, non-SOEs face greater market discipline, making ESG disagreement more informative and liquidity relevant.

In the context of firm financial condition, Table 7 further reveals the unique role of loss-making enterprises (Loss) in the relationship between ESG rating disagreements and stock liquidity. The positive impact of ESG rating disagreements on stock liquidity is more significant in loss-making enterprises than in profitable ones, with the coefficients for EsgDis1 and EsgDis2 in loss-making enterprises being 0.171 and 0.208, respectively (both significant at the 1% level). Moreover, the coefficients for the interaction terms between loss status and ESG rating disagreements (Loss×EsgDis1 & Loss×EsgDis2) are 0.321 and 0.248, respectively (significant at the 5% and 10% levels), indicating that the positive impact of ESG rating disagreements on liquidity is even stronger in loss-making enterprises, thereby supporting research hypothesis H5.

This finding aligns with information asymmetry and signaling theories, which suggest that when firm fundamentals are weak, investors place greater weight on non-financial signals to assess risk and future prospects (Lang et al., 2012; Verrecchia, 2001). For loss-making firms, ESG rating disagreement amplifies uncertainty and attracts heightened investor and analyst attention, intensifying trading activity and strengthening the liquidity effect.

The results of these cross-sectional analyses indicate that the impact of ESG rating disagreements on stock liquidity exhibits clear heterogeneity, significantly influenced by firm ownership type and financial condition. The specific attributes of state-owned enterprises may suppress the positive effects of ESG rating disagreements, while the financial distress status of loss-making enterprises may exacerbate the positive impact of ESG rating disagreements on liquidity. These findings offer new perspectives on how ESG rating disagreements affect stock liquidity across different types of enterprises, highlighting the importance of considering firm heterogeneity in analyzing the impact of ESG rating disagreements.

This study investigates the effect of ESG rating disagreements on stock liquidity in the Chinese capital market, with a focus on the underlying mechanisms and contextual moderators. Using panel data on 4,065 A-share listed firms from 2009 to 2021, we document that ESG rating disagreements significantly enhance stock liquidity, suggesting that divergent ESG signals attract investor attention and stimulate trading activity. Mediation analysis further reveals that analyst attention and research report attention are important transmission channels: rating disagreements increase intermediary scrutiny, which in turn facilitates more informed trading and improved market liquidity. Moreover, the effect is heterogeneous across firm characteristics. The liquidity-enhancing role of ESG rating disagreements is stronger in non-state-owned firms, which lack implicit state support, and in loss-making firms, where weak fundamentals heighten the salience of non-financial signals.

Taken together, our results provide a coherent theoretical narrative: ESG rating disagreement alters the information environment by increasing uncertainty and interpretive demand, which activates information intermediaries and intensifies trading, with the strength of this process depending on firms' institutional and financial characteristics.

By integrating insights from information asymmetry theory and information intermediary theory, this study enriches the ESG literature by showing that rating disagreement, often considered as “noise” (Berg et al., 2022; Christensen et al., 2022), in fact has informational value in shaping investor behavior. The findings highlight the dual role of ESG signals: they not only reflect firms' sustainability performance but also influence market microstructure outcomes such as liquidity. These insights offer new perspectives for investors, analysts, and policymakers, underscoring the importance of ESG disclosure quality, rating transparency, and the need for standardization in the ESG evaluation industry.

Our research offers several important theoretical contributions:

First, it extends prior research on ESG and firm outcomes (Behl et al., 2022; Fatemi et al., 2018; Wong et al., 2021) by focusing on rating disagreements rather than absolute ESG levels. While earlier studies highlight how ESG performance influences firm value and liquidity (He et al., 2023; Luo, 2022), we show that the divergence of ratings itself conveys informational content that stimulates trading. This advances the view that ESG information asymmetry, rather than ESG quality per se, can shape market dynamics.

Second, by incorporating mediation mechanisms, this study highlights the role of financial analysts and research reports as critical information intermediaries (Brauer & Wiersema, 2018; Roulstone, 2003). Consistent with the information intermediary literature (Healy & Palepu, 2001; Livnat & Zhang, 2012), we show that analysts and reports transform rating disagreements into tradable insights, thereby improving liquidity. This mechanism-based approach enriches our understanding of how non-financial information affects capital markets, bridging the gap between ESG disclosure and market microstructure research.

Third, we contribute to the moderation literature by showing that ownership structure and financial condition alter the effectiveness of ESG rating disagreements. Consistent with the literature on ownership and disclosure (Dai & Wang, 2024; Villalonga et al., 2025; Wen et al., 2023), we find that non-SOEs are more sensitive to ESG disagreement, while loss-making firms attract greater attention when ESG ratings diverge. These findings underscore the conditional nature of ESG effects, supporting calls for more nuanced examinations of ESG impacts in heterogeneous institutional contexts (Dhaliwal et al., 2012; Lang et al., 2012).

Taken together, these contributions advance theoretical debates on ESG information, disclosure, and market outcomes. By demonstrating that rating disagreements act as signals that attract intermediary and investor attention, our study bridges ESG research with broader theories of information asymmetry, signaling, and capital market efficiency (Kim & Verrecchia, 1994; Verrecchia, 2001).

This study also holds significant practical relevance for investors, corporate managers, and policymakers:

First, the evidence of a positive and significant association between ESG rating disagreements and stock liquidity suggests that rating divergence itself can serve as a valuable trading signal. For investors, this highlights the need to carefully monitor ESG score discrepancies across agencies, as such disagreements may reflect heightened market attention and stimulate trading opportunities. Rather than dismissing divergence as noise, professional investors may strategically use these signals to identify periods of increased liquidity and exploit short-term trading opportunities (Avramov, Cheng, Lioui, & Tarelli, 2020; Pelizzon, Rzeznik, & Hanley, 2021).

Second, mediation results show that analyst and research report attention are critical channels through which ESG rating disagreements influence liquidity. This implies that market participants should not only observe rating divergence but also track how intermediaries respond. For institutional investors, monitoring the intensity of analyst coverage and the release of ESG-related research reports can provide actionable insights into liquidity conditions. For corporate managers, these results highlight the importance of engaging proactively with analysts and ensuring transparent ESG communication strategies. Doing so may attract positive analyst attention and mitigate uncertainty, thereby improving liquidity and potentially lowering financing costs. Regulators can also use this insight to strengthen the role of certified analysts and encourage high-quality ESG research reporting as a means of enhancing market transparency.

Third, moderation results underscore the heterogeneity of ESG disagreement effects. The finding that non-SOEs experience stronger liquidity impacts suggests that managers of such firms should be especially attentive to how inconsistent ESG signals are perceived. In the absence of state support, proactive ESG disclosure and consistency across reporting channels can reduce uncertainty and protect liquidity. Similarly, the stronger effect observed in loss-making firms suggests that financially distressed firms should carefully manage their ESG narratives, as divergent ratings are likely to attract heightened scrutiny from both investors and analysts. Policymakers may also consider targeted disclosure guidelines or rating oversight for such firms, recognizing that their market responses to ESG signals are particularly pronounced.

Despite its contributions, this study is subject to several limitations that provide opportunities for future inquiry.

First, the analysis is limited to Chinese A-share firms, which may constrain the generalizability of the findings. Future studies could extend the framework to other emerging and developed markets, or conduct cross-country comparisons to evaluate institutional differences (Berg et al., 2022; Christensen et al., 2022). Second, the measurement of ESG rating disagreement is based on the standard deviation across the five agencies. While this approach captures divergence, it may not fully reflect differences in methodology, weighting schemes, or disclosure reliance across rating providers. Future research could adopt alternative disagreement measures, including pairwise divergence or textual analysis of rating rationales, to validate and refine the construct. Third, while our study highlights mediation via analyst and report attention, other mechanisms may also explain the link between ESG divergence and liquidity. Similarly, additional moderators such as industry environmental sensitivity, governance quality, or cross-listing status merit further investigation. Finally, our quantitative approach, while robust, cannot fully capture the behavioral and strategic dimensions of how firms, rating agencies, and investors respond to ESG divergence. Future research could employ qualitative or mixed method designs to complement our findings and provide deeper insights into these dynamics.

All coauthors have made equal contributions to this paper.

The supplementary material for this article can be found online.