Firm-level climate risk exposure, ESG disclosure and stock liquidity: evidence from textual analysis

Global climate change is a pressing issue that affects people from all walks of life in the 21st century. Severe climate problems such as extreme heatwaves, hurricanes, drought and devastating flooding have caused substantial economic losses and pose a major systematic risk to the financial market (Battiston, Dafermos, & Monasterolo, 2021; Engle, Giglio, Kelly, Lee, & Stroebel, 2020). Additionally, the Paris accord’s aim of limiting global temperature rise to below 1.5°C has substantial implications for stock market behavior. Consequently, investors strongly demand firm climate-related disclosures to assess climate risk of informed portfolio decisions (Arian & Sands, 2024; Krueger, Sautner, & Starks, 2020). However, accurately measuring the climate risk exposure of each firm is often challenging because of the scarcity of firm-level data on climate risk exposure.

This study addresses the existing gap by constructing a novel metric for firm-specific climate risk exposure (FCRE), thereby enhancing our understanding of the vulnerabilities related to physical and transition risks stemming from climate change at the level of individual firms. To construct this FCRE, we initially created a set of climate risk exposure seed words from the seminal research of Li, Shan, Tang, and Yao (2020), Sautner, Van Lent, Vilkov, and Zhang (2023) and from annual reports of the Ministry of Environment, Forest and Climate Change, India. Second, we train Word2Vec, a state-of-the-art textual analysis model on management discussion and analysis (MD&A) reports, to expand climate risk seed words. Finally, using Python’s Natural Language Toolkit (NLTK) module, the study quantified FCRE by determining the frequency of the above-constructed climate risk exposure lexicon words in annual reports’ MD&A reports for each firm and year. In a recent study, Li et al. (2020) analyzed the climate risk exposure of Chinese listed companies through sophisticated textual analysis of earnings calls. They differentiated physical and transition risks and identified proactive responses to climate challenges. Similarly, Sautner et al. (2023) demonstrated that climate-related narratives in earnings conference calls could provide insights into a company’s vulnerability to climate change. Drawing on this seminal research, several studies demonstrated the negative consequences of firm-specific climate risk on corporate investment (Agoraki, Giaka, Konstantios, & Negkakis, 2024), capital structure (Li & Zhang, 2023), leverage adjustment (Zhou & Wu, 2023), green innovation (Tian, Chen, & Dai, 2024), cash holdings (Heo, 2021), firm value (Ongsakul, Papangkorn, & Jiraporn, 2023) and stock price and its crash risk (Faccini, Matin, & Skiadopoulos, 2023; Jung & Song, 2023). However, despite the burgeoning focus on the correlation between climate risks and corporate outcomes, there exists a dearth of empirical evidence that quantitatively assesses the impact of FCRE on stock liquidity. Thus, our study aims to address the existing void in knowledge by constructing novel textual-based FCRE using a state-of-the-art natural language processing (NLP) model for Indian-listed firms and examining its effect on firm stock liquidity.

Stock market liquidity is essential for the efficient functioning of financial markets and impacts price efficiency, transaction cost, stock return and financial stability (Chordia, Roll, & Subrahmanyam, 2001; Pástor & Stambaugh, 2003). Liquidity also directly impacts investors’ perceptions by impounding information into asset prices (Amihud, Mendelson, & Pedersen, 2006). Thus, managers and investors care about the firm’s stock liquidity.

Compelling theoretical and empirical evidence supports the notion that climate risk is becoming a substantial determinant of stock market liquidity. Notably, recent asset pricing models emphasize the significance of climate change and environmental issues as long-run risk factors (Bansal, Ochoa, & Kiku, 2017), and Bolton and Kacperczyk (2021) indicate that polluting firms are associated with greater risk exposure, leading investors to demand higher risk premiums to compensation the exposure to environmental or carbon emission risk. Pástor, Stambaugh, and Taylor (2021) (PST), using sustainable investing theory, argued that initial lower returns for green firms, driven by climate concerns as a safer hedge, but green stocks could eventually outperform brown stocks due to unexpected shifts in demand due to investor preference and consumer demand for green products. Thus, the growing demand for environmentally friendly stocks, driven by their reduced climate risk vulnerability, is expected to increase their value through a risk-premium mechanism, potentially enhancing trading activities (Amihud, 2002; Chordia et al., 2001). However, Saravade and Weber (2024) assert that greenwashing in environmental reporting poses the greatest challenges to the market in terms of transaction cost or lack of liquidity, which may impede the effective functioning of green bond markets. Thus, researchers propose a dynamic climate risk hedging strategy using textual analysis of climate news and environmental, social and governance (ESG) score-based portfolios (Engle et al., 2020; Faccini et al., 2023). This suggests that ESG-transparent firms may prevent climate change risk, leading to increased stock liquidity by reducing systematic and idiosyncratic risks and attracting a wider range of investors (Chen & Xie, 2022; Meng-tao, Da-peng, Wei-qi, & Qi-jun, 2023). In addition, Wang, Li, San, and Gao (2023) discovered that an ESG rating could potentially decrease information asymmetry and improve stakeholders’ support, which enhances stock liquidity. He, Du, and Yu (2022) revealed that implementing a comprehensive ESG disclosure score could significantly decrease firms’ stock price crash risk by improving firms’ information transparency and reducing managers’ misconduct behavior. Moreover, an effective ESG rating encourages the reduction of corporate carbon emissions and promotes synergetic green innovation by addressing financial constraints and increasing managers’ ecological consciousness (Li & Xu, 2024; Tan & Zhu, 2022).

India is an ideal candidate for this research. Since India is the fifth largest economy with 8.2 % gross domestic product (GDP) in 2023–2024, the Indian stock market has experienced significant growth over the years, with its market capitalization to GDP ratio increasing from 17.83% in 1991 to 103.7% in 2022. However, this growth often increases carbon emissions and resource exploitation. As a result, India has become the third-largest carbon emitter, experiencing a staggering 156% increase from 2000 to 2021 (IEA, 2023). To address these challenges, India committed at COP26 to reduce carbon emissions by 1 billion tons, achieve 500GW of non-fossil energy capacity, meet 50% of energy needs with renewables by 2030 and reach net-zero emissions by 2070 (PIB Press Release, 2022). India is the top G-20 performer and one of the top seven worldwide climate change performers, according to the Climate Change Performance Index (Burck, Uhlich, Bals, Höhne, & Nascimento, 2024). Despite this, climate change poses multifaceted challenges to Indian firms and their market performance. Thus, it is valuable to analyze how climate risk exposure at the firm level affects stock liquidity in order to combat the challenges caused by climate change in the financial markets.

Based on the discussion above, we propose that an increase in the firm’s exposure to climate change risk, as assessed from the corporate reports of Indian-listed firms, imposes an additional idiosyncratic risk burden on investors and is linked with a decrease in stock liquidity. Further, this paper seeks to investigate the role of ESG disclosure and investor sentiment in mitigating the adverse effects of firm climate risk exposure on its stock liquidity.

Using a large sample of Indian-listed firms from the financial year (FY) 2003–2004 to 2022–2023, we found that FCRE is negatively associated with stock liquidity. This implies that heightened FCRE increases market uncertainty and additional idiosyncratic risk burdens, influencing investor trading behavior and the firm’s stock liquidity level. The relationship remains robust despite potential endogeneity concerns. This has been confirmed through a range of endogeneity tests, such as propensity score matching (PSM), the two-stage least squares instrumental variable (IV) approach and the two-step system generalized method of moments (GMM) estimator, as well as after incorporating additional control variables. Further, we observed that firms with high-quality ESG ratings could mitigate the adverse influence of FCRE on stock liquidity. Furthermore, our analysis reveals that firm-specific investor sentiment (FSIS) and institutional investor holdings (institutional ownership (IO)) serve to partially and fully mitigate the adverse effects of FCRE on stock liquidity, respectively. Our heterogeneity analysis reveals that this negative relationship is especially pronounced among firms characterized by low readability and high complexity in their financial statements, as well as during the introduction and decline phases of their life cycle. Additionally, we include the Paris Agreement as an exogenous shock and find that FCRE worsens stock liquidity after the agreement. This indicates that investors devalue and are less attracted to climate-vulnerable companies as they become more aware of the adverse consequences of climate change.

This paper makes contributions to the extant literature in four key ways. First, to the best of our knowledge, this study is the pioneer study to investigate the relationship between firm-specific climate risk and stock liquidity. This adds to the existing literature on liquidity in stock markets (Brogaard, Li, & Xia, 2017; Dang et al., 2022; Pástor, Stambaugh, & Taylor, 2022) by demonstrating the adverse impact of FCRE on stock liquidity, thereby introducing another substantial idiosyncratic risk factor of stock liquidity. Second, while the existing studies like Battiston et al. (2021) provide the macro-level assessment of climate risk, Sautner et al. (2023) measure firm-level data that are constrained by a short time frame and limited companies for particular countries due to the unavailability of earnings calls, especially in emerging markets. We bridge this gap by introducing a unique FCRE measure for Indian-listed companies, employing Word2Vec methodology on MD&A reports to assess firm-specific climate risk. Third, this study provides robust evidence that higher sustainable practices proxied by ESG disclosure ratings significantly moderate the stock liquidity-reducing impact of climate risk. This evidence supports the stakeholder theory of Freeman, Harrison, Wicks, Parmar, and de Colle (2010), the signaling theory and ongoing research on ESG implications on stock market outcomes (He, Feng, & Hao, 2023; Meng-tao et al., 2023). Finally, our analysis shows a non-linear relationship between FCRE and stock liquidity, moderated by the information environment and investor sentiment. Therefore, this research contributes to the literature on information asymmetry and the behavioral dimensions of stock liquidity (Dang et al., 2022; Debata, Dash, & Mahakud, 2018; Glosten & Milgrom, 1985; Wang, Mbanyele, & Muchenje, 2022). Overall, our study sheds light on the impact of managers’ perception of environmental risk on financial market trading dynamics, specifically in relation to stock liquidity outcomes.

The subsequent sections of this paper will be presented as follows. Section 2 reviews the pertinent literature and develops a testable hypothesis. Section 3 outlines the research design, including the sample selection and variable construction procedure. Section 4 reports the baseline results, endogeneity checks, moderating factors effects, heterogeneity tests and additional tests. Section 5 concludes.

Climate risk exposure pertains to companies’ susceptibility to the potential physical and transition risks associated with climate change, such as severe weather conditions, regulation changes and shifts in consumer preferences toward sustainability (Engle et al., 2020). Over the past few years, an increasing amount of research has delved into the negative impacts of climate risk exposure on various aspects of companies, such as financial performance, corporate investment, corporate governance and market reaction (Agoraki et al., 2024; Ongsakul et al., 2023; Sautner et al., 2023; Zhou & Wu, 2023). Contemporary research definitively demonstrated that climate risk significantly impacts the pricing of different financial assets like stocks, bonds and house property (Bernstein, Gustafson, & Lewis, 2019; Faccini et al., 2023; Ilhan, Sautner, & Vilkov, 2021; Painter, 2020; Seltzer, Starks, & Zhu, 2022). For instance, Bolton and Kacperczyk (2021) discovered a correlation between companies’ CO2 emissions (and changes in emissions) and their stock returns, indicating that firms with higher emissions tend to earn higher returns. Investors expect compensation for their investment in companies with a significant carbon footprint. In a subsequent investigation utilizing a more extensive sample of companies, Bolton and Kacperczyk (2023) validated this evidence. Therefore, institutional investors view climate risk as a severe and long-term problem, suggesting a growing awareness of environmental risks among market participants (Krueger et al., 2020). Choi, Gao, and Jiang (2020) also documented that investors pay closer attention to firms’ carbon emissions and react negatively to climate-related events, leading to changes in stock prices. Nevertheless, the financial market faces difficulties in accurately assessing climate risks because of several challenges, such as a lack of risk-sharing mechanisms, uncertainties in evaluating climate risks and policy responses and insufficient information accessible to investors (Alam, Mohamad Tahir, Y.H, Saif-Alyousfi, & Pahlevi, 2024; Bansal et al., 2017; Engle et al., 2020). This reveals that market inefficiencies may influence stock liquidity in pricing climate risk, as investors may face challenges in accurately valuing climate-exposed firms.

Theoretically, Fama’s (1970) efficient market hypothesis posits that a semi-strong form of efficient market stock prices reflects all publicly available information. Thus, new climate risk disclosures can adjust stock prices as investors update their beliefs and trading strategies accordingly. Firms disclosing climate risks may benefit from reduced information asymmetry, lowering their cost of capital, improving market reputation and enhancing market value. In contrast, firm exposure to climate risk can significantly create vulnerabilities in the market and an extra risk burden for investors through different channels. Firstly, extreme climate events like droughts, heatwaves and floods directly cause disruptions in operations and supply chains, reducing productivity and, thus, stock valuation (Chen & Yang, 2019; Pankratz, Bauer, & Derwall, 2023). Secondly, regulatory interventions, such as emission trading schemes, increase compliance and financing costs, further increase market uncertainty and lower stock liquidity (Cepni, Şensoy, & Yılmaz, 2024; Ginglinger & Moreau, 2023; Painter, 2020; Ramadorai & Zeni, 2024). Thirdly, climate-related issues enhance the demand for investment capital to ensure net-zero emission adaptation. However, firms face financial constraints because investors require risk premiums for these green innovation projects (Agoraki et al., 2024; Fang, 2024). Lastly, stock returns and liquidity are impacted by climate risk uncertainty, as trading activities are influenced by changes in demand and supply, as indicated by higher trading volumes linked to larger stock price movements (Campbell, Grossman, & Wang, 1993).

Research by Bolton and Kacperczyk (2021), Pedersen, Fitzgibbons, and Pomorski (2021) and Hsu, Li, and Tsou (2023) suggests that climate risks are priced based on carbon emissions, making high-emission firms perceived as riskier investment options. More importantly, Pástor et al. (2021) proposed a theoretical model suggesting that high-emission firms (brown firms) may offer higher expected returns than others (green firms) because investors utilize green assets to hedge against climate-related risks. However, Pástor et al. (2022) observed in a subsequent study that stocks of environmentally friendly or green firms tend to perform better over the long term than those of brown firms. This outperformance is attributed to the increasing consumer demand for eco-friendly products and investors’ preference for “greeniums” and greater ESG performance stocks.

Considering the potential detrimental outcomes of climate risk, as discussed above, we argued that firms’ exposure to climate risk can affect stock liquidity through heightened uncertainty, investor aversion to risk and regulatory uncertainty. Uncertainty causes risk-averse investors to circumvent climate-exposed stocks, leading to disruptions in liquidity and hindering the price discovery process (Baker, Bradley, & Wurgler, 2011; Bansal et al., 2017; Easley & O’Hara, 2010).

Despite the proliferation of scholarly works on climate finance, there is a dearth of research on how FCRE impacts firms’ stock liquidity. Hence, our research addresses this gap by investigating the impact of FCRE on stock liquidity. As a result, we formalize our first hypothesis as follows:

ESG has garnered widespread attention in the capital market due to growing concerns about environmental sustainability and corporate social responsibility. ESG disclosure rating offers additional information to investors to enhance their ability to monitor the manager’s commitment to sustainability and encourage firms to pursue green innovation, especially amid rising climate risk exposure (Chen, Kuo, & Chen, 2022; Chen & Xie, 2022; Tan & Zhu, 2022). Additionally, ESG factors foster a positive corporate image among stakeholders, accumulating reputation capital that acts as insurance when the company encounters adverse shocks (Lins, Servaes, & Tamayo, 2017). Furthermore, these ratings encourage the reduction of corporate carbon emissions by easing financial constraints and solving agency issues (Li & Xu, 2024). Thus, corporates are developing sustainable and transparent business strategies that attract more responsible investors and generate long-term socio-economic value by reducing information asymmetry, tax avoidance, cost of capital, idiosyncratic risk and stock price crash risk (Hoi, Wu, & Zhang, 2013; Li, Tsang, Zeng, & Zhou, 2021; Adeneye, Kammoun, & Ab Wahab, 2023; Chen & Xie, 2022; Feng, Goodell, & Shen, 2022).

Freeman’s stakeholder theory contends that adhering to ESG principles promotes corporate governance, prevents financial irregularities and increases firm value by addressing the needs of all stakeholders, including customers, suppliers, employees and society (Freeman et al., 2010). According to signaling theory, ESG reporting conveys managers’ commitment to sustainability, disclosure transparency and risk management, boosting the firm’s reputation and attracting more investors (Chen & Xie, 2022; Wong & Zhang, 2022). Based on these theories, existing research suggests that providing ESG information can potentially enhance stock liquidity by reducing corporate risk, addressing information asymmetry and enhancing corporate reputation, especially by building confidence among investors (He et al., 2023; Meng-tao et al., 2023). This, in turn, can lead to increased investor confidence and their trading preference toward sustainable companies.

In summary, we believe that companies with heightened ESG ratings can reduce the adverse effects of firm-level climate risk on a firm’s stock liquidity by catering to the needs of stakeholders, fostering green transformation, enhancing long-term value and reducing information asymmetry through enhancing investors’ trust and effective climate governance. Therefore, we propose our second hypothesis:

Our sample comprises all listed firms in the National Stock Exchange of India (NSE) spanning the period from FY 2003–2004 to 2022–2023 (i.e. from 1^st April to 31^st March)[¹]. Following Chordia et al. (2001) and Dang et al. (2022), we employed the following stock selection criteria in our study. First, we required that each stock be continuously traded throughout the sample period, ensuring the availability of daily trading information. Second, we excluded stocks that were not actively traded, as well as firms lacking MD&A reports and relevant data throughout the entire sample period, which facilitated the construction of a balanced panel dataset, resulting in an initial pool of 544 firms out of a total of 2,206 NSE-active listed companies. Third, following the guidelines set forth by Brogaard et al. (2017), we further refined our sample by excluding financial companies due to their distinct capital structures, disclosure practices and regulatory frameworks. This stringent process culminated in a final dataset comprising 9,320 firm-year observations across 466 Indian firms, spanning the FY from 2003–2004 to 2022–2023.

We obtained our sample data from different sources to construct all our variables. Specifically, (1) individual stock liquidity proxies and other market-related data originate from the Bloomberg database, (2) FCRE is constructed using text analysis on the MD&A portion of annual reports, which is manually retrieved from the Bombay Stock Exchange ^[2], and (3) accounting-based control variables are collected from the Center of Monitoring Indian Economy database.

We winsorize all continuous variables at the 1% and 99% levels to minimize the potential impact of outliers.

We construct the FCRE, which involves a three-step process. First, we built a set of seed words related to climate risk exposure from Li et al. (2020), Sautner et al. (2023) and the annual reports published by the Ministry of Environment, Forest and Climate Change, India. Second, we implement a machine learning model, specifically the continuous bag-of-words (CBOW) model within the Word2Vec framework, which is utilized to undergo training on MD&A reports. This trained model is subsequently employed to expand the seed words pertaining to climate risk by assessing the similarity between the seed words and the words present in the pre-trained model.

Words are eliminated from the expanded lexicon that does not exactly reflect climate risk exposure by manually examining all words thoroughly. Additionally, Table A3 ( Appendix) shows the top 20 climate risk exposure keyword occurrences in the sample, providing a holistic picture of the frequency and importance of particular terms regarding climate risks in the sample period. Table A4 ( Appendix) provides the list of industry names corresponding to the companies in our study sample, along with their respective industry codes, as classified under the National Industrial Classification (NIC) – 2008 framework of India. Then, we validate our constructed FCRE measure by showing the summary statistics of the FCRE by year and industry in  Appendix Table A5 and display them by industry in Figure A1. Finally, following Li and Zhang (2023), the study utilizes Python’s NLTK library and a custom-designed word count algorithm, specifically to calculate the frequency of the above-constructed climate risk lexicon words within the MD&A content of annual reports for each firm and each year. This approach allows for a quantitative assessment of how frequently climate risk-related terms appear in the MD&A sections of annual reports, shedding light on how companies address and emphasize climate risks over different periods, which is noted as follows:

Stock liquidity is an elusive and hard-to-define concept. Liquidity is defined as the extent to which an investor can buy or sell substantial amounts of security without paying any huge trading cost and without causing a negative change in the value of the asset (Amihud et al., 2006). We use two different measures of stock liquidity to examine the relationship between FCRE and stock liquidity.

Our first proxy is Amihud’s (2002) measure of liquidity (Amihud), which captures the price impact characteristics of liquidity. This is a widely used proxy of liquidity and has been used by Fong, Holden, and Trzcinka (2017), Brogaard et al. (2017) and Debata et al. (2018). The Amihud price impact is calculated as the logarithm of one plus the average ratio of the daily absolute return to the INR trading volume on day d for stock i over year t, where Dit is the number of trading days for stock i in year t following Cheung, Chung and Fung (2015) and Dang et al. (2022). We then multiplied −1 to convert illiquidity to liquid to make the analysis easier, which is noted as follows:

where $|Ridt|$ is the absolute stock return of stock i on day d of year t, $VOLDidt$ is the trading volume of stock i on day d of year t and $Dit$ is the number of days with available data for stock i in year t. we transform the ratio by taking its natural logarithm and then multiplying −1 to convert illiquidity to liquidity.

Second, we calculate the high-low spread (HLS) of stock construed by Corwin and Schultz (2012) and used by Marshall, Nguyen, and Visaltanachoti (2012), Cheng and Fang (2023) and Roy, Rao, and Zhu (2022). In order to untangle the variance and spread parts of the high-low price, Corwin and Schultz (2012) compute the summation of the squared log price for two successive days:

Where $HO$ represents the observed high price of a stock on day d and $LO$ represents the observed low price of a stock on day d. ln = Natural logarithm. $Ht,t+1O=$ It is the high price over the two consecutive days t and t+1. $Lt,t+1O=$ It is the low price over the two successive days’ t and t+1. Corwin and Schultz (2012) have derived a solution for the spread (HLS) by drawing from previous research conducted on high-low price ratios.

The HLS estimation is calculated for each two-day interval by utilizing the daily high and low prices obtained from the Bloomberg database. For each year, the bid-ask spread for each sample stock is computed by averaging all spreads throughout all two-day intervals. In order to convert the illiquidity measurement of spread to liquidity, it was multiplied by −1.

To investigate the impact of FCRE on stock liquidity, we implement the following baseline regression model:

Where $Liquidityi,t$ represents the stock liquidity of firm i in year t and is measured using the Amihud and HLS proxy. $FCREi,t−1$ denotes firm-specific climate risk exposure. $Xi,t−1$ is a vector of control variables that potentially impact stock liquidity (Dang et al., 2022), such as stock return volatility (RVOL), the inverse of the share price (IPRC), firm size (SIZE), leverage (The LEV), research and development expenses (R&D) and IO. We also control industry $(δj$⁠) and time-fixed effects $(γt$⁠). $β0$ is the constant term. All the independent variables lagged one year. In addition, we include year- and industry-fixed effects to account for omitted variables bias and cluster the standard errors at the firm level to control for time-varying firm heterogeneity, serial correlations and heteroskedasticity in the error term. $εi,t$ is the error term. The detailed descriptions of all the variables are reported in Table A1 ( Appendix).

In order to investigate the moderating role of ESG disclosure in the FCRE–stock liquidity nexus, we implement the following modified baseline model:

Where ESG represents the environmental, social and governance rating score provided by the Bloomberg database.

Table 1 presents descriptive statistics for our sample’s key variables, such as stock liquidity (Amihud and HLS), FCRE and firm financial characteristics. The statistics show Amihud and HLS averages of −0.014 and −0.017, respectively. FCRE has a mean of 1.647 and a standard deviation of 1.043, indicating substantial climate exposure variations among companies.

Table 2 presents the correlation matrix, indicating a negative correlation between FCRE and stock liquidity proxies (Amihud and HLS). Furthermore, we observe that SIZE, R&D and IO positively correlate with our stock liquidity variables, whereas RVOL, IPRC and LEV have a negative correlation. In addition, we conducted variance inflation factor (VIF) diagnostic tests, which confirmed that our variables were not affected by multicollinearity issues, as their VIF values were less than two.

We present our baseline regression results (i.e. the relationship between FCRE and stock liquidity) in Table 3. Columns 1–2 report the regression output without control variables, while Columns 3–4 exhibit the regression output with the inclusion of all control variables as specified in Eq. (3). The results show that FCRE is significantly and negatively associated with both Amihud (β1 = −0.0029, t = −22.53 without control variables; β1 = −0.0021, t = −17.83 with control variables) and HLS (β1 = −0.0008, t = −6.67 without control variables; β1 = −0.0006, t = −2.84 with control variables). This supports our 1st Hypothesis (H1).

In addition to their statistical significance, these findings are also of considerable economic importance. For example, as results shown in Column 3, an increase of one standard deviation in FCRE (104.38%) leads to a decrease of about 0.22% (=1.043^*(−0.0021) in Amihud. The observed decrease can be interpreted as approximately 15.71% (=0.22/−0.014) of the average Amihud throughout sample firms. These results indicate that heightened corporate exposure to climate risks reduces stock liquidity. This aligns with the PST theory and the findings of Ongsakul et al. (2023), Pástor et al. (2022) and Krueger et al. (2020), which suggest that investors prioritize companies’ environmental governance, specifically climate risk management practices, leading to stock price devaluation for those with inadequate strategies and diminish interest towards environmentally suboptimal or “brown” stocks.

With regard to control variables, we observe that SIZE, R&D and IO are positively and significantly associated with stock liquidity proxies, whereas RVOL, IPRC and The LEV are significantly negatively related. The plausible reason is that higher RVOL and IPRC harm liquidity by depressing stock prices, increasing trading costs and risks and reducing market participation. On the other hand, SIZE positively affects liquidity, as larger firms are more attractive to trading activity because of lower operational risk and more information availability. R&D also boosts liquidity as investors like innovation and will be more attracted to trading those growth potential stocks. These results are consistent with the prior literature (Boubaker, Gounopoulos, & Rjiba, 2019; Dang et al., 2022; Wang et al., 2022).

To investigate the moderating role of ESG disclosure in mitigating the adverse effects of climate risk exposure on stock liquidity, this model includes the same control variables incorporated in the baseline model, as provided in the model (8). Table 4 presents the results of the moderating role of ESG. As indicated in Table 4, the results show that the coefficient of the interaction term (FCRE*ESG) is significantly positive for both liquidity proxies (Amihud: β3 = 0.0017, p < 0.01; HLS: β3 = 0.0024, p < 0.01). This indicates that the adverse impact of FCRE on stock liquidity is mitigated or alleviated by high-quality ESG disclosure practices. These results support 2nd Hypothesis (H2) of our study.

The probable reason could be that the companies promoting and disclosing ESG practices can adequately convey the manager’s environmental risk management strategies and provide other valuable non-financial insights to investors. This reduces information asymmetry and investor exaggeration regarding climate-related apprehension and ultimately improves stock liquidity (Krueger et al., 2020; Meng-tao et al., 2023). Moreover, our findings are aligned with Chen and Xie (2022) and Wang et al. (2023), as they documented that ESG disclosure can instill stakeholders’ trust, facilitate efficient market dynamics, attract diverse investors, which boosts investor trading activities, and ultimately increase stock liquidity. The findings also support Freeman et al.'s (2010) stakeholder theory and signaling theory, as these suggest that ESG disclosure reduces the agency cost and satisfies the demand of all types of corporates stakeholders, which improves the corporate competitive edge and enhances the long-term socio-economic value of corporates.

First, we implemented PSM, as suggested by Rosenbaum and Rubin (1983), to help mitigate the endogeneity issues associated with self-selection bias. PSM is implemented by categorizing the panel into two groups, such as the treated group (high FCRE) and the control group (low FCRE), as per the median value as the threshold. Next, we conduct a logit regression analysis on the dichotomous variable (1 for the treated group and 0 for the control group) using all control variables in our baseline regressions. We also included industry and year dummies in the sample. Thereafter, we calculated the predicted propensity score and employed nearest-neighbor matching. Table 5 (Columns 1 and 2) presents the regression results for the matched sample. The consistent findings of the PSM analysis indicate that potential endogeneity concerns are unlikely to impact our conclusion.

Additionally, Table A2 demonstrates a balancing test by comparing the treated and control groups before and after PSM and t-statistics that quantify the differences. The balancing test provides strong evidence that PSM effectively achieved balance in the variable distributions for both groups.

In this subsection, we address the endogenous problem of our model by employing the IV method, specifically the two-stage least squares (2SLS IV) regression model. This approach mitigates potential endogeneity problems, including omitted variables, reverse causality and measurement errors. Following Ongsakul et al. (2023), we re-estimate the baseline model utilizing the annual sectoral level (2-digit NIC classification) average of FCRE as the IV. This choice of IV is consistent with the notion that firms within the same industry experience shared external factors and operate under a comparable environment. Firms operating within similar sectors, particularly those in the energy or utilities domains, encounter common climate-related risks, including regulatory shifts and transition difficulties that influence their strategies for mitigating climate risks. This sectoral average eliminated the endogeneity problem because it captured the average risk in different industries rather than the specific managerial anticipated risks for individual firms. This analysis highlights exogenous factors driving climate risk, offering a robust framework for assessing how firms respond to climate-related challenges.

Column (3) of Table 5 displays the first-stage regression results with the dependent variable FCRE, showing that the IV FCRE (industry average) is significantly positive, as expected. Columns (4) and (5) present the second-stage regression results. The FCRE (instrumented) coefficients remain negative and statistically significant for both stock liquidity proxies, similar to the outcomes observed in the baseline regressions. We validated the IV using diagnostic tests, including the Anderson canon. corr. Lagrange Multiplier (LM) statistic (532.867, p-value = 0.000) and the Cragg–Donald Wald F statistic (564.69), which exceeds the Stock and Yogo (2002) threshold at the 10% level. These results confirm the instrument’s validity and robustness.

In this subsection, we execute the dynamic two-step system GMM estimator to examine the causal relationship between FCRE and stock liquidity. This method offers robustness against issues like unobserved heterogeneity, reverse causality and dynamic endogeneity within the explanatory variables (Roodman, 2009). Table 5 reports the results of the system GMM panel estimation.

As in Columns (6)-(7) of Table 5, the results show that the FCRE is negative and significantly related to stock liquidity (Amihud and HLS). Additionally, the statistical insignificance of the Autoregressive 2 (AR2) and Hansen J-statistics indicates no autocorrelation and validates the instruments in our models. These results are consistent with the findings of the baseline regression in Table 3.

This subsection provides the results of the analysis of the moderating role of FSIS and institutional investor holdings percentage (IO) on FCRE-LIQUIDITY.

Investor sentiment is the key factor influencing market behavior and stock performance and is also an important factor driving stock liquidity (Debata et al., 2018). Baker and Wurgler (2006) assert that investors are often overconfident and optimistic and fail to look at underlying risks, especially when firms are keen on green issues or ESG practices (Matsumura, Prakash, & Vera-Muñoz, 2014). Hence, firms that claim to prioritize ESG factors are more likely to see higher demand for their shares and therefore higher trading volume, even if they are experiencing climate risks (Engle et al., 2020). On the other hand, negative investor sentiment was found to worsen the liquidity problem. Those investors who expect poor climate risk management from a firm may exit, causing a vicious cycle of illiquidity (Krueger et al., 2020).

Institutional investors typically prioritize long-term investment strategies and demonstrate a commitment to sustainability, recognizing the enduring systematic risks associated with climate change. This awareness may help alleviate the adverse impacts of FCRE on stock liquidity. The research indicates that institutional investors focused on ESG factors tend to maintain their investments in companies facing significant climate risks, potentially leading to a decrease in stock volatility (Krueger et al., 2020; Lopez-de-Silanes, McCahery, & Pudschedl, 2024). Moreover, institutional investors are increasingly calling for robust climate risk disclosures. This demand enhances the quality of climate governance and streamlines investment decisions, ultimately reducing adverse selection costs and narrowing the bid-ask spread (Flammer, Toffel, & Viswanathan, 2021; Ilhan, Krueger, Sautner, & Starks, 2023).

To test these moderating effects, we first form the FSIS as defined by Aboody, Even-Tov, Lehavy, and Trueman (2018) by taking the average of overnight returns over the fiscal year and multiplying the result by 250, which is the estimated number of trading days in a year.

Table 6 shows the moderating roles of FSIS and IO on the relationship between FCRE and stock liquidity. In Columns (1) and (2), the moderating role of FSIS is presented, while in Columns (3) and (4), the moderating role of IO is presented.

In Columns (1)-(2), FSIS exhibits a significantly positive effect on both stock liquidity measures (Amihud: β = 0.0028, p < 0.01; HLS: β = 0.0011, p < 0.05). When FSIS is added to the model, FCRE remains negative and significant, though its impact weakens slightly (Amihud: −0.0019 vs. −0.0021; HLS: −0.0004 vs. −0.0006). However, the interaction term between FCRE and FSIS (FCRE*FSIS) is negatively and significantly associated with stock liquidity. This suggests that although FSIS can partially alleviate the negative effect of FCRE, it is insufficient to fully counteract the adverse liquidity impacts stemming from climate risk. Thus, positive investor sentiment may not be strong enough to completely offset the perceived risks of climate exposure.

Columns (3) and (4) reveal that the interaction between FCRE and IO (FCRE*IO) is significantly positive in both liquidity proxies (Amihud: β = 0.0144, p < 0.01; HLS: β = 0.0009, p < 0.01). This result highlights the importance of institutional investors in enhancing the stability of stock liquidity. Because these investors have a long-term horizon and are focused on sustainability, they reduce the adverse effects of FCRE by pressurizing effective climate risk management strategies (Flammer et al., 2021; Krueger et al., 2020; Lopez-de-Silanes et al., 2024).

This section illustrates the role of the information environment on the nexus between FCRE and LIQUIDITY. To examine these heterogeneity tests, we divided the sample into four categories based on the median values of two information quality measures: readability and complexity of the firm’s MD&A report (i.e. high readability and low readability and high complexity and low complexity). The aforementioned measures have been widely used in previous literature to evaluate the quality of the corporate information environment (Li, 2008; Loughran & McDonald, 2023; Luo, Li, & Chen, 2018; Su, Zhai, & Liu, 2023).

Readability of firm financial disclosures is essential to offset the adverse effects of information asymmetry on stock liquidity. Due to complex financial reports, investors have a hard time understanding a company’s financial health and risk management strategies. Furthermore, managers may obscure mediocre results or hide adverse information by delivering complex disclosures (Bloomfield, 2008). It increases perceived risk, and thus, investors are less trusting of these managers’ earning management practices, especially in an uncertain environment (Lo, Ramos, & Rogo, 2017; You & Zhang, 2009). Thus, they are more likely to demand risk premiums and have wider bid-ask spreads (Boubaker et al., 2019). Therefore, we argue that the linguistic readability and complexity of financial reporting are proxies for the quality of a firm’s MD&A report information.

The findings are presented in Table 7. Columns (1)-(4) demonstrate that the negative effect of FCRE on stock liquidity (Amihud and HLS) is substantial and particularly evident in the case of low-readability stocks, while the impact is comparatively less significant for high-readability stocks. In a similar vein, Columns (5)-(8) illustrate that the negative coefficient of FCRE is significantly more pronounced for high complexity stocks, while it is comparatively less for low complexity stocks. This suggests that firms providing clearer and less complex MD&A disclosures and demonstrating enhanced information transparency have bolstered investor confidence, mitigated the costs linked to climate risks and helped companies maintain liquidity despite unpredictable environmental challenges. This outcome aligns with the empirical evidence (Boubaker et al., 2019; Luo et al., 2018) and supports the information asymmetry theories associated with stock liquidity, including those proposed by Diamond and Verrecchia (1991) and Glosten and Milgrom (1985).

Table 7 presents the findings. Columns (1)-(4) show that the negative impact of FCRE on stock liquidity (Amihud and HLS) is significant and pronounced for low-readability stocks and lower magnitude for high-readability stocks. In the same vein, Columns (5)-(8) show that the negative coefficient of FCRE is much more pronounced for high complexity stocks and less so for low complexity stocks. The probable reason could be that firms with greater information transparency have enhanced investor confidence, reduced the costs linked to climate risks and assisted companies in preserving liquidity amidst unpredictable environmental challenges. These findings are consistent with the empirical evidence (Boubaker et al., 2019; Luo et al., 2018) and the information asymmetry theories of stock liquidity, such as those of Diamond and Verrecchia (1991) and Glosten and Milgrom (1985).

In this section, we investigate the role of the CLC in the relationship between FCRE and stock liquidity by classifying CLC according to Dickinson’s (2011) measure. Table 8 shows the results of CLC heterogeneity. Panel A (Amihud) and Panel B (HLS) models display the results at each company’s life cycle stage. The findings reveal a nonmonotonic relationship between FCRE and stock liquidity across different stages. Interestingly, the coefficient magnitude of FCRE is lower during the growth and mature stages. A plausible explanation is that growth and mature firms demonstrate higher levels of profitability, market stability, increased engagement in corporate social responsibility, lower idiosyncratic volatility and earnings management practices that enhance investor optimism, ultimately eroding the negative impact of FCRE on stock (Bakarich, Hossain, Hossain, & Weintrop, 2019; Dickinson, 2011; Hasan & Habib, 2017).

Overall, this analysis supports Miller and Friesen’s (1984) CLC theory by showing variability in climate management practices across CLC stages and which are more resilient during the growth and mature stages.

A company’s reaction to external shocks, such as changes in climate management regulations, might impact the stock market. We exploit the Paris Agreement, adopted in 2015 with the acceptance of 196 parties, making it a landmark international accord. It has significantly raised global awareness of climate change and may heighten investor attention toward climate-related risks (Ongsakul et al., 2023). To this hypothesis, we split the sample based on the Paris Accord (before and after 2015).

Table 9 indicates that the coefficients of FCRE are insignificant for the pre-Paris Accord, which reveals that the stock liquidity is more vulnerable to climate risk, possibly due to increased investor awareness of climate risk after the Paris Agreement, 2015.

We incorporate additional control variables into our baseline model to further address endogeneity concerns. These controls include firm profitability, measured through both accounting-based performance (return on assets) and market-based performance (Tobin’s Q), as well as firm age and product market concentration, captured by the Herfindahl–Hirschman Index.

The results presented in Table 10 demonstrate that FCRE remains negative and statistically significant for both stock liquidity measures, reinforcing the robustness of our baseline model findings.

This study examines the impact of novel textual-based FCRE on stock liquidity. Employing a comprehensive sample of 466 Indian listed firms from FY 2003–2004 to 2022–2023, our study results show that a rise of one standard deviation in FCRE is associated with a corresponding decline of 15.71% in average stock liquidity (Amihud). In addition, the study reveals that effective ESG disclosure can reduce the adverse effect of FCRE on stock liquidity. This suggests that ESG reduces information transparency and improves corporate climate governance, which can enhance firm stock liquidity. Additionally, our analyses indicate that the adverse impact of FCRE on stock liquidity is partially and fully moderated by FSIS and institutional holdings, respectively. Furthermore, our heterogeneity analysis reveals a nonmonotonic relationship across different levels of information disclosure quality and various stages of the CLC. This highlights the importance of behavioral factors, the theory of corporate information asymmetry and the life cycle theories related to stock liquidity. Our baseline results remain robust after addressing the endogeneity problems using the PSM, IV approach and two-step system GMM, as well as after including additional control variables. Overall, this study sheds light on how firm-level environmental risk perceptions affect financial market trading dynamics, particularly stock liquidity. Furthermore, it extends to the growing state-of-the-art textual analysis metrics of management disclosures by constructing the first-ever FCRE for Indian-listed companies.

The findings of our study have significant practical implications for a wide range of stakeholders, such as investors, managers and policymakers. First, all types of investors, including individual investors and fund managers, may find it valuable to understand the impact of climate change exposure on firm stock liquidity. This study can assist them in evaluating corporate value and making informed decisions about their portfolios by incorporating the firm-level climate change risk. After all, stock liquidity plays a crucial role in determining the potential returns and risks associated with investments. Second, it is crucial for managers to proactively evaluate and handle risks associated with climate change, incorporate climate factors into their decision-making procedures and establish environmentally conscious practices to enhance climate governance. In addition, policymakers seek to address climate change and promote socio-economic development through regulatory actions such as making stringent rules for reducing carbon footprint, climate risk disclosure and ESG transparency policies. Finally, environmental and social activist groups may leverage these findings to advocate for sustainable corporate practices.

Our analysis is subject to some limitations. The application of NLP techniques to extract climate risk from MD&A reports hinges on the precision of algorithms, which can potentially misinterpret nuanced or evolving terminology. Furthermore, the distinctive market structure and regulatory framework present in India could restrict the applicability of our results to other markets. Future research may benefit from integrating alternative data sources, such as sustainability reports and applying sophisticated methodologies like large language models (e.g. Bidirectional Encoder Representations from Transformers or GPT-based models). Analyzing cross-market comparisons and investigating moderating factors like corporate governance can enhance our comprehension of the influence of climate risk on financial markets worldwide.

Overall, our study offers valuable insights for various stakeholders and market participants, enabling informed decision-making and supporting the sustainable development of the economy.