Next Article in Journal
Proposal of an Integrated Method of Unmanned Aerial Vehicle and Artificial Intelligence for Crack Detection, Classification, and PCI Calculation of Airport Pavements
Previous Article in Journal
New Energy Demonstration City Policy and Corporate Green Innovation: From the Perspective of Industrial and Regional Spillover Effect
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Climate Risk Disclosure and Financial Analysts’ Forecasts: Evidence from China

1
School of Economics and Management, China University of Petroleum (East China), Qingdao 266580, China
2
School of Business, Qingdao University of Technology, Qingdao 266520, China
*
Author to whom correspondence should be addressed.
Sustainability 2025, 17(7), 3178; https://doi.org/10.3390/su17073178
Submission received: 23 February 2025 / Revised: 28 March 2025 / Accepted: 31 March 2025 / Published: 3 April 2025
(This article belongs to the Section Air, Climate Change and Sustainability)

Abstract

:
This study examines whether climate risk disclosure (CRD) matters to financial analysts in China. Using textual analysis to measure CRD, we find that CRD is negatively related to analyst forecast error and dispersion, supporting the information hypothesis. We also find that information disclosure quality (e.g., earnings quality) and external monitoring (e.g., long-term institutional investor) may moderate this relationship. Mechanism analysis indicates that lower information asymmetry and more climate-related on-site visits are potential channels through which CRD influences analyst forecast properties. Furthermore, the above relationship is more pronounced in regions with higher climate awareness, carbon-intensive industries, and state-owned enterprises, and the relationship is primarily driven by transition risk disclosure (TCRD) rather than physical risk disclosure (PCRD). Our findings, which remain valid after addressing various robustness and endogeneity concerns, have significant implications for regulators to standardize and enhance CRD practices.

1. Introduction

Climate risk, arising from extreme weather events, has emerged as one of the most significant challenges for firms worldwide (The Global Risks Report, 2020 (https://cn.weforum.org/publications/the-global-risks-report-2020/, accessed on 30 March 2025)). Previous research generally confirms that climate risk can cause physical damage to tangible assets, decreasing not only the value of those assets but also the potential economic benefits they could have generated [1,2]. Furthermore, climate disasters typically cause disruptions in business operations and supply chains [3,4], resulting in a loss of productivity and earnings [5]. The above-mentioned uncertainty imposed by climate risks in corporate activities increases the volatility of earnings and cash flow [6], making it difficult for financial analysts to forecast key financial metrics accurately, such as sales, margins, and cash flows, and leading to greater forecast error and dispersion [6,7]. Then, a question arises as to whether climate risk disclosure (hereafter CRD) matters to financial analysts’ forecasts. However, relatively little is known about such a question, with a few exceptions [8]. In this paper, we investigate whether and how CRD is related to financial analysts’ forecasts in China.
We posit that effective CRD in annual reports may mitigate the negative impact of climate risk on analysts’ forecasts [6,7]. The CRD in annual reports provides analysts with detailed insights into what climate risks a firm faces, how those risks affect its operations, and most importantly, how it responds to those risks. Therefore, accurate and reliable CRD may alleviate information asymmetry [9] and help analysts better assess the effect of climate risk on firms’ future business strategy, results of operations, and financial conditions, thereby enhancing their forecasting abilities [8]. Alternatively, CRD once manipulated may mislead analysts’ forecasts. The semi-voluntary nature of CRD in China means that managers have a high degree of discretion to determine whether and to what extent to disclose climate risks. In addition, given that CRD is primarily textual information, managers may engage in textual manipulation and “greenwashing” behaviors, thus increasing analysts’ forecasting challenges arising from climate risks.
Accordingly, whether CRD matters to analysts’ forecasts is ultimately an empirical question. To address this question, we develop two competing hypotheses, the information hypothesis and the opportunistic hypothesis, with opposite predictions regarding the influence of CRD on analysts’ forecasts, and test these conflicting hypotheses using a sample of Chinese A-share listed companies during the period 2007–2023. Following Lin and Wu [9], we employ textual analysis and machine language learning methods to measure CRD. Empirical analysis indicates that the higher the frequency of CRD in annual reports, the lower the analyst forecast error and dispersion. Our findings suggest that CRD in annual reports is meaningful to financial analysts and may compensate for their deteriorated forecast performance caused by climate risks [6,7]. To confirm the validity of these findings, we conduct several robustness tests, including adjusting CRD by its industry–year average, transforming CRD into its natural logarithm, and excluding samples during the COVID-19 pandemic period. We also employ the Heckman two-stage model, propensity score matching (PSM), instrument variable analysis (IV), and the PSM-DID model to address potential endogeneity concerns. After verifying the positive impact of CRD on analysts’ forecasts, we further investigate the moderating effects of information disclosure quality (e.g., earnings quality) and external monitoring (e.g., long-term institutional investors) on this relationship. Our findings indicate that earnings quality can enhance the credibility of CRD [10], and long-term institutional investors have a monitoring effect on CRD [11], thus further moderating the relationship between CRD and analysts’ forecasts. Mechanism analysis reveals that CRD improves analyst forecast accuracy by alleviating information asymmetry and increasing climate-related on-site visits. In addition, we conduct three heterogeneity analyses at the region, industry, and firm levels. In regions with more climate awareness, industries with more carbon emissions, and state-owned enterprises, the above relationship is more pronounced. Given that climate risk is typically distinguished into physical risk and transition risk (TCFD, 2017), we also examine whether physical risk disclosure (hereafter PCRD) and transition risk disclosure (hereafter TCRD) equally matter to analysts. Our findings suggest that TCRD can significantly enhance analyst forecast accuracy, whereas PCRD cannot.
Our study is related to, but not overlapping with, that of Ben-Amar et al. [8]. Ben-Amar et al. [8] focus on the effect of CRD on analysts’ forecasts in the U.S., while we highlight the issue in China. As the largest emerging economy, China exhibits substantial differences from developed countries, such as the U.S., in terms of its institutional framework and legal environment, which could inevitably influence analysts’ forecasts [6]. For example, the U.S. SEC has required listed firms to disclose material climate risks in their annual reports since 2010. In 2024, the SEC formally proposed relevant disclosure rules titled The Enhancement and Standardization of Climate-Related Disclosures: Final Rules to ensure that all SEC registrants provide investors with more reliable climate-related information in their registration statements and annual reports. However, no such specialized regulations have been issued in China. Although the SSE, SZSE, and BSE, three main stock exchanges in China, jointly issued The Guidelines for Self-Regulation of Listed Companies-Sustainability Reporting (Trial) in 2024, the guidelines have not yet been formally implemented. Moreover, only part of listed companies is required to disclose climate risks, and the disclosure requirements are neither as detailed nor as strict as those mandated by the SEC’s final rules (see Appendix A for more details). Therefore, given that CRD practice is still semi-voluntary in China, the findings of Ben-Amar et al. [8] may not be fully applicable to the Chinese setting, and our study is motivated by this gap.
Our study contributes to the literature in several ways. First, we extend research on CRD by examining its effect on analysts’ forecasts. The negative effects of climate risks at the firm level have been well documented in prior research [1,3,5,12]. Consequently, stakeholders such as investors have expressed a strong need for reliable and comparable CRD to facilitate their decision-making [13]. However, compared to the extensive body of literature on climate risk, research on CRD remains relatively limited. Relevant studies have examined the effect of institutional investors [11] and environmental protection tax reform [14] on CRD, and the impact of CRD on stock price crash risk [9], cost of equity [15], and its peer effect [16]. Ben-Amar et al. [8] focus on analysts’ forecasts, suggesting that CRD does not necessarily matter in the U.S. unless climate risks are financially material. In contrast, our study extends the findings of Ben-Amar et al. [8] by positing that CRD can generally enhance analysts’ forecasts in China. Moreover, we also find that this relationship is primarily driven by TCRD rather than PCRD, which contrasts with the study of Ben-Amar et al. [8]. As mentioned above, this discrepancy may be due to the different institutional backgrounds of CRD in China and the U.S. To the best of our knowledge, our study is the first to examine this question in China, thus providing new empirical evidence on this topic and enriching the existing literature.
Second, our study enriches the literature concerning the effects of non-financial disclosure on analysts’ forecasts. Prior research has demonstrated that CSR disclosure [17], carbon disclosure [18], and the disclosure of non-financial information in annual reports, such as forward-looking information [19] and risk-related information [20], can enhance analyst forecast accuracy. Nonetheless, these non-financial disclosures, being primarily textual, are also vulnerable to managerial manipulation [21], which can subsequently deteriorate analyst’s forecasts. Our findings suggest that although CRD in China is largely voluntary, it remains relatively reliable and relevant, thereby improving analyst forecast accuracy. In this regard, we extend previous mixed findings of non-financial disclosure by demonstrating the positive effect of CRD on analysts’ forecasts.
Finally, our study has significant implications for regulators to standardize and enhance CRD practices. As mentioned above, there are no formal climate-related disclosure regulations in China so far. However, given the ease of manipulation of non-financial disclosures, it is necessary for Chinese regulators to establish a comprehensive CRD system. This system should guide firms to disclose more reliable, consistent, and comparable climate-related information, thereby fully satisfying the needs of investors, analysts, and other stakeholders. For example, the U.S. has proposed guidance and rules regarding CRD in 2010 and 2024, and the U.K. has become the first to mandate listed firms to disclose climate risks following the TCFD’s recommendation in 2021, all of which can provide a general reference for Chinese regulators. Furthermore, our study can also provide similar practical insights for other emerging economies.

2. Related Literature and Hypothesis Development

We develop two competing hypotheses, the information hypothesis and the opportunistic hypothesis, to investigate the impact of CRD on financial analysts’ forecasts.

2.1. The Information Hypothesis

Given the negative impacts of climate risks on firms, including the damage to tangible assets [1,2], disruptions to business operations and supply chains [3,4], and a loss of productivity and earnings [5], investors often incorporate climate risks into their investment decisions [22] and tend to downgrade the valuation of firms with higher climate risk exposure [23]. Similarly, creditors may impose stricter loan terms [24] or extend fewer loans [25,26] in response to climate risks. According to the signaling equilibrium theory, companies, as the information-advantaged party, can mitigate the information asymmetry for investors through effective information disclosure [9]. Therefore, it is crucial for firms to provide more reliable and relevant climate-related disclosures, particularly regarding the impacts of climate risks on their operations and the strategies they are implementing to mitigate these risks. Such enhanced disclosure can significantly alleviate information asymmetry between managers and external stakeholders, such as investors and creditors [9], helping them restore trust from these stakeholders. As for financial analysts, Zhang and Kanagaretnam [6] note that climate risks can increase the volatility of earnings and cash flows, making it more challenging to provide accurate forecasts [7]. We posit that CRD can also reduce information asymmetry between managers and analysts, thereby mitigating the negative impact of climate risks on analysts’ forecasts [6,7]. This proposition is consistent with Hope [27], who states that a greater quantity of financial disclosure can enhance analyst forecast accuracy.
Furthermore, both public and private information are critical for analysts’ forecasts [28,29]. We infer that the availability of public climate-related information in annual reports may motivate analysts to seek additional private information for verifying the reliability of public information. Cheng et al. [30] and Han et al. [31] demonstrated that on-site visits can facilitate analysts’ private information acquisition and significantly enhance their forecast accuracy. Building on this notion, we suppose that analysts may conduct more climate-related on-site visits to evaluate the real impact of climate risks on firms’ future business strategy, results of operations, and financial conditions, thereby providing more accurate forecasts.
Overall, if the information hypothesis prevails, the public disclosure of climate risk may not only alleviate information asymmetry between managers and analysts but also facilitate analysts’ private information acquisition from more climate-related on-site visits. Together, these factors can enable analysts to make more accurate forecasts. Thus, we propose the information hypothesis as follows:
H1. 
CRD can enhance analyst forecast accuracy.

2.2. The Opportunistic Hypothesis

The regulatory enforcement against textual disclosure violations remains relatively rare, to some extent, providing implicit institutional protection for listed companies engaging in textual manipulation. Recent studies confirm that managers may engage in such manipulation to influence investor perception. For example, Huang et al. [21] suggest that managers use tone management to mislead investors, while Wang and Wang [32] highlight that tone management serves as a complement to accrual manipulation. Additionally, managers may manipulate the readability of textual information to hide bad news [33], conceal accounting manipulation [34], receive more abnormal compensation [35], and mitigate the risk of control transfer [36].
Given that CRD primarily consists of textual content, it is particularly vulnerable to such manipulation. Notably, although China introduced CRD guidelines in 2024, these guidelines have yet to be formally implemented and do not apply to all listed firms, indicating that regulatory enforcement against CRD violations remains relatively limited. In practice, many listed companies exhibit inadequate or inappropriate CRD in their annual reports, including omitting material climate risks that should be disclosed, providing only qualitative analyses without substantive assessments or quantitative data, and engaging in climate-related “greenwashing”. Such “greenwashing” behaviors typically involve downplaying the operational impacts of physical climate risks or overstating the effectiveness of activities undertaken to mitigate transition risks, such as falsifying carbon emissions data. For instance, Inner Mongolia Erdos High-Tech Materials Co., Ltd. was the first firm to be publicly disclosed for carbon emissions data fraud in China’s national carbon emissions trading market by falsifying its 2019 carbon emissions data. Therefore, if CRD involves such textual manipulation, it can undoubtedly increase analysts’ forecasting challenges arising from climate risks [6,7].
Overall, if the opportunistic hypothesis prevails, managers might engage in CRD manipulation to mislead analysts, resulting in either a negative or negligible effect of CRD on analysts’ forecasts. Thus, we propose the opportunistic hypothesis as follows:
H2. 
CRD cannot enhance analyst forecast accuracy.
It is worth noting that even if CRD is accurate and reliable, a negative relationship between CRD and analyst forecast accuracy cannot be entirely ruled out. Climate-related risk disclosed in annual reports is entirely different from other information because of its risk orientation and uncertainty [6,20]. Compared to known risks, unknown risks are more likely to affect asset pricing [20]. As a form of unknown risk, climate risk may trigger market anxiety, reduce investors’ confidence, and lead analysts to overestimate its potential impact on business operations. Moreover, CRD may increase information uncertainty, which has been known to impair analyst forecast accuracy [29]. Consistent with these views, the risk orientation and uncertainty inherent in CRD may also induce less accurate analysts’ forecasts even if CRD is accurate and reliable. However, a positive relationship between CRD and analyst forecast accuracy could help exclude this alternative explanation, thereby supporting the information hypothesis.

3. Research Design

3.1. Data and Sample

Our initial sample includes all A-share non-financial listed companies in China from 2007 to 2023. The sample period begins in 2007, coinciding with the formal implementation of China’s new accounting standards system on 1 January 2007. After excluding missing observations, the ultimate sample contains 24,564 observations. CRD data are obtained from annual reports, and data for calculating analysts’ forecasts and other financial variables are from CSMAR and CNRDS, two widely used databases in China. All variables are winsorized at the top and bottom 1% levels.

3.2. Regression Model

We construct the following regression model to investigate the effect of CRD on financial analysts’ forecasts:
F E R R O R i , t ( F D I S P i , t ) = α 0 + α 1 C R D i , t 1 + α i C o n t r o l s i , t + I n d u s t r y i , t + Y e a r i , t + ε i , t
where the dependent variables FERROR and FDISP are the proxy for analyst forecast accuracy. The variable of interest CRD is the proxy for CRD. The information hypothesis (H1) is supported if α1 is significantly negative, while the opportunistic hypothesis (H2) is supported if α1 is significantly positive or insignificant. We also control the industry fixed effect (Industry) and year fixed effect (Year) in Model (1).

3.3. Independent Variable

Our main independent variable is CRD. Following Lin and Wu [9], we use textual analysis and machine language learning methods to measure CRD at the firm level. The main steps include determining the seed words of climate risks, expanding the seed words using the Word2Vec model, extracting information from the text of annual reports, and identifying the climate-related words in annual reports. Ultimately, the climate-related word set comprises 47 seed words and 138 augmented words after referring to several experts’ opinions (see Appendix B for more details on these words). CRD is calculated as the ratio of the climate risk words to the total words in annual reports, as shown in Equation (2).
C R D i , t = C l i m a t e   r i s k   w o r d s W o r d s   i n   a n n u a l   r e p o r t s
To verify the validity of CRD, we also compare its annual trend with that of the SNSI ESG ratings following Du et al. [37]. Firms with higher ESG ratings tend to have better internal governance and disclosure mechanisms, making them more likely to disclose climate-related risks in their annual reports. Moreover, the environmental dimension (“E”) of ESG typically includes climate-related metrics such as carbon emissions, energy efficiency, and climate risk management. Therefore, firms with more efficient CRD are expected to exhibit better ESG performance, particularly in the “E” dimension. Specifically, the SNSI ESG ratings consist of nine levels—C, CC, CCC, B, BB, BBB, A, AA, and AAA—which are assigned values from one to nine in ascending order. As presented in Figure A1 of Appendix C, the overall trend of CRD closely aligns with that of the SNSI ESG ratings over time, supporting the validity of CRD as a measure of climate risk disclosure.

3.4. Dependent Variable

Financial analyst forecast accuracy is proxied by forecast error (FERROR) and dispersion (FDISP), as presented in Equations (3) and (4). FERROR is measured as the average absolute difference between the actual EPS (EPS) and forecasted EPS (FEPS), scaled by the stock price (P) of last year. FDISP is calculated as the standard deviation of forecasted EPS (FEPS), divided by the stock price (P) of last year. Since we focus on the effect of CRD in year t − 1 on analysts’ forecasts for year t, as presented in Model (1), FEPSi,t is the average EPS for year t as forecasted in the period from the release of the annual report for year t − 1 to that for year t. The larger the values of FERROR and FDISP, the lower the accuracy of analysts’ forecasts.
F E R R O R i , t = E P S i , t F E P S i , t P i , t 1
F D I S P i , t = S t d F E P S i , t P i , t 1

3.5. Control Variable

Following Zhang and Kanagaretnam [6] and Ben-Amar et al. [8], we include various control variables in Model (1), including firm size (SIZE), profitability (ROA), leverage (LEV), sales growth (GROWTH), board size (BOD), proportion of independent directors (ID), duality of chairman and CEO (DUAL), shareholdings of controlling stockholders (TOP1), audit firm type (Big4), and the number of analysts following the firm (FCOVER). The definitions of all variables are presented in Table 1.

4. Empirical Analyses

4.1. Descriptive Statistics

Table 2 presents the descriptive statistics, providing an overview of the main variables. The mean value of CRD is 0.007, with a minimum value of 0.001 and a maximum value of 0.027, indicating significant variation in CRD among sample firms. The descriptive statistics of FERROR, FDISP, and other control variables are consistent with prior research.

4.2. Baseline Results

Table 3 presents the baseline results. Our main interest is CRD, the coefficients of which are significantly negative at the 1% level (coef. = −0.183, t-stat. = −5.77; coef. = −0.065, t-stat. = −4.17), indicating that a one-standard-deviation increase in CRD is associated with a decrease of 4.36% (2.95%) in the standard deviation of analyst forecast error (dispersion). Our findings suggest that CRD in annual reports is meaningful for analysts and can partially offset some of the forecasting challenges associated with climate risks [6,7], which is consistent with the information hypothesis (H1).

4.3. Robustness Tests

To confirm the validity of these results, we conduct three robustness tests. First, since there may be industry differences in CRD, we calculate the adjusted CRD (STDCRD) as the difference between the CRD and its’ industry–year average following Plumlee et al. [38], and then re-run the regression of Model (1) with STDCRD. As shown in columns (1) and (2) of Table 4, our baseline results are robust to such alternative measures of CRD.
Second, we take the natural logarithm of CRD, calculating LNCRD as the natural logarithm of climate risk words plus 1. The results presented in columns (3) and (4) are also consistent with the baseline findings.
Third, to mitigate the impact of the COVID-19 pandemic on analysts’ forecasts, we exclude samples from 2019 to 2020. Overall, columns (5) and (6) indicate that our baseline findings are unlikely to be affected by the sample period.

4.4. Heterogeneity Tests

To address potential endogenous concerns, we conduct several endogenous tests. On the one hand, firms with a higher frequency of CRD may also engage in high-quality financial disclosure, which could independently influence analysts’ forecasts. To mitigate this sample selection bias, we employ both Heckman’s two-stage model and propensity score matching (PSM). The results from the Heckman model, presented in columns (1) and (2) of Table 5, show that CRD is negatively related to analyst forecast error and dispersion, supporting the robustness of our baseline findings. For the PSM analysis, we split the sample into two groups: high CRD (firms with CRD above the upper quartile) and low CRD (firms with CRD below the lower quartile). Following Ni et al. [39], we set the caliper at the 0.5% level to ensure sufficient similarity between the two groups. The results in columns (3) and (4) indicate that CRD can also enhance analyst forecast accuracy even after controlling for other observable characteristics.
On the other hand, potential reverse causality may exist between CRD and analysts’ forecasts. To mitigate this concern, we first lag CRD by one period in Model (1). In addition, we also employ instrument variable (IV) analysis and the PSM-DID model for further investigation. Following Yang et al. [40], we use the Paris Agreement (PA) as an instrumental variable as it is closely related to a firm’s CRD but exogenous to error terms. Specifically, PA is a dummy variable that equals 1 for years after 2016 and 0 otherwise. The results of our main analysis reported in columns (5) and (6) remain unchanged. Furthermore, China, as one of the world’s largest industrial producers and carbon emitters, has committed to achieving carbon peaking by 2030 and carbon neutrality by 2060 since 2020, providing a natural quasi-experimental setting to construct a DID model. Firms in carbon-intensive industries, serving as the primary regulatory targets in achieving the dual-carbon goals, confront more pressure to reduce carbon emissions and are more likely to be expected to disclose climate-related information. Therefore, we classify these firms as the treatment group (TREAT = 1) and others as the control group (TREAT = 0). A dummy variable (POST) is introduced that equals 1 for years after 2020 and 0 otherwise. Then, we conduct 1:1 nearest-neighbor matching without replacement based on all control variables in Table 1 and estimate Model (5) using the matched sample. As present in columns (7) and (8), the coefficients of TREAT × POST are significantly negative at the 1% level, suggesting that reverse causality is unlikely to drive our earlier findings.
F E R R O R i , t ( F D I S P i , t ) = μ 0 + μ 1 T R E A T i , t + μ 2 P O S T i , t + μ 3 D I D i , t + μ i C o n t r o l s i , t + I n d u s t r y i , t + Y e a r i , t + ε i , t

4.5. Tests of Moderating Effect

In this section, we further investigate the moderating effects of information disclosure quality (e.g., earnings quality) and external monitoring (e.g., long-term institutional investors) on the relationship between CRD and analysts’ forecasts.

4.5.1. Moderating Effect of Earnings Quality

It has been well demonstrated in prior research that quantitative information and textual information often serve as a complement to each other. For example, Wang and Yuan find that earnings quality can enhance the credibility of textual information [10], while other scholars confirm that textual manipulation is often used by managers to complement earnings manipulation [32,34]. Based on these insights, we posit that high earnings quality may signal the high quality of climate risk-related textual information in annual reports, and analysts who incorporate high-quality CRD into their earnings forecasts can improve forecast accuracy.
Therefore, to examine the potential moderating effect of earnings quality on the relationship between CRD and analysts’ forecasts, we use the negative absolute value of discretionary accruals as the proxies for earnings quality (EQ) and interact EQ with CRD to construct Model (6). To address the multicollinearity problem between the interaction term (CRD × EQ) and CRD, we follow Balli and Sørensen [41] and decenter the interaction term (CRD × EQ). The coefficients of CRD × EQ in columns (1) and (2) of Table 6 are significantly negative at the 1% and 5% levels, respectively, indicating that earnings quality can strength the positive effect of CRD on analysts’ forecasts, consistent with our expectations. These findings reveal that high-quality information disclosure is a critical prerequisite for the effectiveness of CRD, echoing the findings of Wang and Yuan [10] and further supporting the information hypothesis that effective CRD may compensate for the negative impact of climate risk on analysts’ forecasts.
F E R R O R i , t ( F D I S P i , t ) = β 0 + β 1 C R D i , t 1 + β 2 E Q i , t 1 + β 3 C R D i , t 1 × E Q i , t 1 + β i C o n t r o l s i , t + I n d u s t r y i , t + Y e a r i , t + ε i , t

4.5.2. Moderating Effect of Long-Term Institutional Investors

Institutional investors can influence corporate decision-making not only by exercising exit threats through “voting with their feet” but also by actively participating in corporate governance through “voting with their hands”. Prior research has confirmed institutional investors’ critical role in shaping CRD. For example, Ilhan et al. [13] reveal that CRD is considered as crucial as financial reports by institutional investors. Moreover, Cohen et al. [11] propose that institutional investors call for more CRD. However, institutional investors differ significantly in their monitoring roles due to variations in investment philosophies and preferences. Compared with short-term institutional investors who engage in high-frequency trading to pursue immediate returns, long-term institutional investors tend to focus more on firms’ sustainable development. As a result, they may exert stronger external monitoring on firms’ CRD, and long-term institutional investor ownership may also strengthen the positive impact of CRD on analysts’ forecasts.
Following Zhou et al. [42], we distinguish between long-term and short-term institutional investors based on their turnover rates. Specifically, institutional investors are sorted into three groups according to their average turnover rates. The group with the lowest turnover rate is classified as long-term institutional investors, while the group with the highest turnover rate is classified as short-term institutional investors. Then, we use the shareholding ratios of long-term institutional investors to measure LINS and interact LINS with CRD to construct Model (7). The coefficient of CRD × LINS in column (3) of Table 6, which is also decentered following Balli and Sørensen [41], is significantly negative at the 1% level, suggesting that long-term institutional investors may monitor firms to disclose more reliable climate-related information, which can mitigate the information asymmetry between managers and analysts, and further help analysts provide more accurate forecasts. These findings are consistent with our expectations and support the information hypothesis from the information asymmetry perspective.
F E R R O R i , t ( F D I S P i , t ) = λ 0 + λ 1 C R D i , t 1 + λ 2 L I N S i , t 1 + λ 3 C R D i , t 1 × L I N S i , t 1 + λ i C o n t r o l s i , t + I n d u s t r y i , t + Y e a r i , t + ε i , t

5. Mechanism Analyses

Having confirmed the effect of CRD on financial analysts’ forecasts, we attempt to provide further evidence by investigating the possible channels through which CRD works. As noted in the information hypothesis, CRD may not only alleviate information asymmetry by disclosing more detailed climate-related information about the firm [9] but may also lead financial analysts to conduct more climate-related on-site visits to obtain additional private information, which may facilitate them to better estimate the impact of climate risks on firms’ future business prospects and financial conditions. In line with this argument, CRD can enhance financial analysts’ forecasts by alleviating information asymmetry and increasing climate-related on-site visits. In this section, we adopt a “three-step approach” for the mechanism analysis [40].

5.1. Mechanism Analysis of Information Asymmetry

Following Hou and Moskowitz [43], we employ stock price delay (DELAY) as a proxy for information asymmetry. Table 7 presents the mechanism analysis related to information asymmetry. Columns (1) and (4) are consistent with our baseline findings in Table 3, indicating that CRD in annual reports can reduce the analyst forecast error and dispersion. The coefficient of CRD in column (2) is −0.777 and significantly negative at the 1% level, suggesting that CRD reduces stock price delay and alleviates information asymmetry, consistent with the findings of Lin and Wu [9]. Furthermore, the coefficients of DEALY in columns (3) and (5) are significantly positive at the 1% level, while the coefficients of CRD remain significantly negative at the 1% level, revealing that information asymmetry partially mediates the relationship between CRD and analyst forecast accuracy. Overall, these findings provide further support for the information hypothesis. On the one hand, CRD provides analysts with more firm-specific climate-related information and mitigates the information asymmetry between managers and analysts, through which CRD can further enhance analyst forecast accuracy. On the other hand, our study also aligns with prior research emphasizing the value of non-financial disclosure in facilitating analysts’ forecasts. For example, Dhaliwal et al. [17] posit that non-financial disclosures, such as CSR reports, may complement financial disclosures and assist analysts in making more accurate forecasts.

5.2. Mechanism Analysis of Analysts’ Climate-Related On-Site Visits

To conduct the mechanism analysis of analysts’ climate-related on-site visits, we measure SURVEY as the number of climate-related on-site visits divided by 100 [31]. Similarly, the results presented in Table 8 imply that CRD can induce more analysts’ climate-related on-site visits, through which CRD can further reduce forecast error and dispersion. These findings are consistent with the notion that on-site visits may facilitate analysts’ private information acquisition [30,31] and such private information can help analysts further verify the reliability of public information disclosed by firms. Overall, the public disclosure of climate risks, corroborated with additional private information from climate-related on-site visits, can assist analysts in making more accurate forecasts, further supporting the information hypothesis.

6. Further Analyses

To further support our main findings, we conduct heterogeneity analyses at the region, industry, and firm levels, respectively. We also distinguish climate risks into two types and examine whether PCRD and TCRD equally matter to analysts.

6.1. Region Heterogeneity Analysis: Regions with High Climate Awareness vs. Regions with Low Climate Awareness

Based on the regional variation in climate risk [44], climate awareness may also vary across regions. Specifically, in regions with high climate awareness, analysts may have a high demand for CRD, thereby prompting firms to disclose more comprehensive and detailed climate-related information in their annual reports. This, in turn, can assist analysts in making more accurate forecasts. However, in regions with less climate awareness, climate-related information may not matter to analysts and other types of information may be more essential in their forecasting models. Consequently, climate awareness can cause regional heterogeneity in CRD’s effect on analysts’ forecasts.
To investigate such region-level heterogeneity, we divide the sample into two groups based on whether the regional climate searching index, a proxy for climate awareness, exceeds the annual median. The results in Table 9 suggest that CRD can enhance analysts’ forecasts in all regions. However, the effect is relatively significant in regions with higher climate awareness, which is consistent with our expectations.

6.2. Industry Heterogeneity Analysis: Carbon-Intensive Industries vs. Non-Carbon-Intensive Industries

Huang et al. [1] suggest that the impact of climate risk varies across industries, with carbon-intensive industries being particularly vulnerable to climate risk caused by extreme weather events [37]. Based on this review, CRD in carbon-intensive industries may be more essential for analysts to help them assess the impact of climate risk on firms’ future business strategy, results of operations, and financial conditions, thereby assisting them in forecasting earnings more accurately.
To examine industry-level heterogeneity, we divide the sample into two groups: carbon-intensive and non-carbon-intensive industries. Following Du et al. [37], we consider mining, manufacturing, electric power, thermal power, gas, water generation and supply, construction, transportation, warehousing, and postal services to be carbon-intensive industries. The coefficients of CRD in carbon-intensive industries, presented in Table 10, are significantly negative at the 1% level, whereas those in non-carbon-intensive industries are not statistically significant. These findings provide valuable implications for Chinese regulators in the pursuit of dual-carbon goals, highlighting the feasibility to standardize and strengthen CRD practices, particularly in carbon-intensive industries.

6.3. Firm Heterogeneity Analysis: State-Owned Enterprises vs. Non-State-Owned Enterprises

In China, there are notable differences in policy implementation between state-owned enterprises (SOEs) and non-state-owned enterprises (non-SOEs). Specifically, SOEs maintain a close relationship with the government and often enjoy priorities to obtain subsidies and tax incentives, thereby enabling them to respond more effectively to government policies in return for these resource advantages. In contrast, non-SOEs, which generally possess weaker political connections, are more market-oriented in their decision-making processes. In light of this, SOEs tend to be more proactive in disclosing climate-related information [14,40], especially following the issuance of the Guidelines for Self-Regulation of Listed Companies-Sustainability Reporting (Trial), which can facilitate analysts in utilizing such information for more accurate earnings forecasts.
To investigate firm-level heterogeneity, we explore whether the impact of CRD on analysts’ forecasts responds differently to firm ownership. As shown in Table 11, the absolute coefficients of CRD in SOEs are significantly larger than those in non-SOEs, indicating that the positive effect of CRD on forecast accuracy is more pronounced in SOEs. These findings are consistent with prior research highlighting the positive action of SOEs in CRD [14,40]. Moreover, they also suggest that the engagement of non-SOEs in implementing CRD policies needs to be further enhanced.

6.4. Effect of PCRD and TCRD on Analysts’ Forecasts

Climate risk is typically distinguished into physical risk and transition risk. The former is related to the physical impacts of climate change, while the latter is the transition risks related to low-carbon transformation. We further examine the impact of PCRD and TCRD on analysts’ forecasts. The results in Table 12 indicate that TCRD can enhance analyst forecast accuracy, while PCRD cannot. It also suggests that the positive impact of CRD on analysts’ forecasts is primarily driven by TCRD rather than PCRD, which contrasts with the study of Ben-Amar et al. [8] and provides new empirical evidence on this topic.
We attribute the inconsistency between our findings and those of prior studies to the different levels of informational content provided by PCRD and TCRD. Physical risks, such as floods, droughts, or heatwaves, often exhibit common patterns across regions and industries, allowing analysts to obtain relevant information through alternative channels (e.g., government reports, media coverage, or satellite data), thereby reducing the marginal value of firm-specific PCRD for analysts’ forecasts. In contrast, transition risk is more firm-specific and closely related to a firm’s strategic responses to climate policies and regulatory changes. Through TCRD, firms can provide analysts with insights into how they are adjusting current and future strategies, business models, and resource allocations to address climate-related risks and opportunities. These disclosures may also include details on specific transition plans such as process improvements or equipment upgrades and the progress of their implementation, thereby providing analysts with richer firm-level information that enhances earnings forecast accuracy.

7. Conclusions

7.1. Conclusion and Discussion

Recently, the negative effect of climate risks on financial analysts’ forecasts has received increasing academic attention [6,7]. Building on this stream of research, we further investigate whether CRD can compensate for analysts’ deteriorated forecast performance caused by climate risks. Following Lin and Wu [9], we employ textual analysis and machine language learning methods to measure CRD and examine its relationship with analysts’ forecasts using a sample of Chinese A-share listed companies during the period 2007–2023. We document that CRD is negatively related to analyst forecast error and dispersion, and that this effect is moderated by information disclosure quality (e.g., earnings quality) and external monitoring (e.g., long-term institutional investor). These findings support the information hypothesis, suggesting that CRD in annual reports is valuable for financial analysts and can mitigate some of their forecasting challenges related to climate risks [6,7]. The positive moderating effect of earnings quality is consistent with those in prior studies highlighting that earnings quality enhances the credibility of textual information [10]. Meanwhile, it also implies that high-quality information disclosure serves as a critical foundation for ensuring the effectiveness of CRD. In addition, long-term institutional investor ownership also strengthens the positive impact of CRD on analysts’ forecasts, indicating that long-term institutional investors, who focus more on firms’ sustainable development, can exert stronger external monitoring on CRD, thereby mitigating the information asymmetry between managers and analysts and ultimately facilitating analysts’ forecasts.
Having demonstrated the positive effect of CRD on analysts’ forecasts, we further investigate its potential mechanisms. Our empirical analysis suggests that CRD may not only alleviate information asymmetry by disclosing more detailed firm-specific information but also lead analysts to conduct more climate-related on-site visits to obtain private information. Together, these factors may help analysts better estimate the impact of climate risks on a firm’s future earnings. Hence, lower information asymmetry and more climate-related on-site visits are two possible channels through which CRD influences analysts’ forecasts. Heterogeneity analyses at the region, industry, and firm levels indicate that the above relationship is particularly pronounced in regions with high climate awareness, carbon-intensive industries, and state-owned firms. Specifically, an analyst’s demand for CRD in regions with high climate awareness is much higher than that in other regions, thus leading firms to disclose more climate-related information. Moreover, carbon-intensive industries are particularly vulnerable to climate risks compared to other industries, so CRD in carbon-intensive industries may be more essential for analysts’ forecasts. Moreover, SOEs’ engagement in implementing CRD policies is more active than that of non-SOEs, which can better facilitate analysts’ demand for climate-related information. Overall, all of these heterogeneity findings provide further support for the information hypothesis. Ultimately, we examine the impact of both PCRD and TCRD on analysts’ forecasts and find that the positive impact of CRD on analysts’ forecasts is primarily driven by TCRD rather than PCRD, which completely contrasts with the study of Ben-Amar et al. [8]. This divergence is largely due to the richer firm-specific information content embedded in TCRD, which provides more valuable insights for analysts in assessing firms’ strategic responses to climate risks.

7.2. Policy Implications

As climate risk becomes a global concern, the standardization and enhancement of CRD have been increasingly emphasized by regulators and institutions worldwide. For example, both the U.S. SEC and ISSB have issued several specialized CRD-related rules since 2010. Although China also introduced CRD-related guidelines in 2024, these guidelines have not yet been formally implemented and are not specifically designed for CRD. Moreover, the guidelines do not apply to all listed firms, and the requirements of the disclosure content and format remain relatively loose. For instance, firms are not mandated to disclose Scope 1, Scope 2, and Scope 3 GHG emissions. While both qualitative and quantitative disclosure are encouraged in the guidelines, quantitative analysis is not mandatory. These limitations indicate that China still has a long way to go in establishing a comprehensive CRD system that guides firms to disclose more reliable, consistent, and comparable climate-related information.
Our findings provide several implications for the development of such a system. For example, effective CRD practices cannot rely solely on market forces but also need to be complemented by regulatory intervention and governmental support. On the one hand, we find that long-term institutional investors exhibit a monitoring effect on promoting CRD, highlighting the power of the market in CRD practice. On the other hand, the positive effect of CRD on analysts’ forecasts is more pronounced in SOEs than in non-SOEs, reflecting the stronger responsiveness of SOEs to policy directives and emphasizing the advantage of governmental intervention in CRD practice. Furthermore, as our results indicate that the effect of CRD on analysts’ forecasts is primarily driven by TCRD rather than PCRD, regulators should place greater emphasis on encouraging firms to disclose more transition risk-related information, especially quantitative information, to help investors and analysts better evaluate the impact of climate risks on firms’ future business prospects and financial conditions.

7.3. Future Research Prospects

While this study provides valuable insights into the impact of CRD on analysts’ forecasts, several prospects remain for future research. First, as CRD is typically influenced by institutional environments and regulatory intensity, both of which vary across countries and regions, future research could conduct cross-country comparative studies to explore the heterogeneous effects of CRD on analysts’ forecasts, thereby enhancing the generalizability of the current findings. Second, this study relies on textual analysis of annual reports to measure CRD, which may not fully capture issues such as false disclosures or “greenwashing”. Future studies could incorporate carbon emissions data, third-party ESG ratings, or employ advanced natural language processing (NLP) techniques to assess the credibility and accuracy of CRD. Such extensions would enrich our understanding of how climate-related information affects the decision-making of capital market participants.

Author Contributions

Conceptualization, Y.L.; methodology, Y.L.; software, Y.L.; validation, J.H.; formal analysis, Y.L.; investigation, Y.L.; resources, J.H.; data curation, Y.L.; writing—original draft preparation, Y.L.; writing—review and editing, J.H.; visualization, J.H.; supervision, J.H.; project administration, J.H.; funding acquisition, Y.L. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Natural Science Foundation of Shandong Province (grant number ZR2023QG111) and the Fundamental Research Funds for the Central Universities (grant number 23CX06041A).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The data can be requested from the corresponding author upon reasonable request.

Conflicts of Interest

The authors declare no conflicts of interest.

Appendix A

Table A1. Climate-related disclosure regulations.
Table A1. Climate-related disclosure regulations.
RegulationThe Enhancement and Standardization of Climate-Related Disclosures: Final RulesInternational Financial Reporting Standards Sustainability Disclosure Standard 2—Climate-Related Disclosures (IFRS S2)Guidelines for Self-Regulation of Listed Companies—Sustainability Reporting (Trial)
Issuing AuthorityU.S. SECISSBSSE, SZSE and BSE of China
Applicable EntityCompanies listed in the U.S.Global entities adopting IFRS Sustainability StandardsListed companies in China
MandatorMandatory for all SEC registrantsMandatory at the jurisdictional level by local regulatorsMandatory only for companies included in the SSE 180 Index, STAR 50 Index, SZSE 100 Index, and ChiNext Index, and those listed both domestically and overseas
Formal RegulationYesYesNo (Trial)
Specialized RegulationYesYesNo
Framework BasisBased on TCFD, including governance, strategy, risk management, metrics, and targetsBased on TCFD, including governance, strategy, risk management, metrics, and targetsBased on TCFD, ISSB, and Chinese policies, including governance, strategy, impacts, risk, and opportunity management, metrics, and targets
Materiality PrincipleFinancial materialityFinancial materialityFinancial materiality and impact materiality
Climate-Related ContentMandatory disclosure of material financial impactsDisclosure of impacts on cash flow, assets, and liabilitiesDisclosure of climate-related risks, opportunities, and responses
Mandatory Disclosure of GHG EmissionsScope 1 and 2Scope 1, 2, and 3No
Qualitative or Quantitative AnalysisQualitative and quantitative analysesQualitative and quantitative analysesQualitative and quantitative analyses, but the latter with limited requirements
Mandatory Scenario analysisNoYesNo
Mandatory External AssuranceLimited assurance is required for Scope 1 and 2NoNo

Appendix B

Climate risk word set is presented in Table A2. In the process of expanding seed words using the Word2Vec model, words with similarity greater than or equal to 0.5 with the seed words are retained, and three types of words were manually deleted: (1) words identified in the form of numbers or letters; (2) words with informal or incorrect expressions; and (3) some neutral words, such as power station, sun, etc.
Table A2. Climate risk word set.
Table A2. Climate risk word set.
Word TypeWord Set
Seed wordNew energy, electric vehicles, wind energy, hydropower, nuclear energy, solar energy, photovoltaic, wind power generation, hydropower, nuclear energy, hydrogen energy, energy storage, carbon sink, forest, ocean, technology, innovation, battery, infrastructure, energy efficiency, carbon neutrality, carbon emissions peak, emission reduction, carbon trading market, climate change, ecology, air temperature, precipitation, drought, flood disasters, natural disasters, pollution control, exhaust gas, wastewater, dust, cleanliness, low-carbon, energy saving, environmental protection, efficiency, optimization, new type, industrial structure, energy structure, green, R&D, green bonds
Augmented wordNew energy, electric vehicles, wind energy, water energy, nuclear energy, solar energy, photovoltaic, wind power generation, hydropower, nuclear energy, hydrogen energy, energy storage, carbon sink, forest, ocean, technology, innovation, battery, infrastructure, energy efficiency, carbon neutrality, carbon emission peak, emission reduction, carbon trading market, climate change, ecology, air temperature, precipitation, drought, flood disasters, natural disasters, pollution prevention and control, exhaust gas, wastewater, dust, clean, low carbon, energy saving, environmental protection, high efficiency, optimization, new type, industrial structure, energy structure, green, research and development, green bonds, ammonia nitrogen, ammonia gas, environmental protection, heavy rain, sun exposure, sultry, warming, charging, odor, energy storage, standard emission, atmosphere, heavy fog, rainwater, nitrogen oxides, ground temperature, low energy consumption, electric vehicle, electricity, power battery, power system, freezing rain, rainy, sulfur dioxide, power generation, waste, wastewater treatment, dust, wind power, wind sensation, wind force, fluoride, high energy consumption, industrial wastewater, solid waste, solid waste, energy consumption, flood, environmentally friendly, air pressure, carbon reduction, energy consumption reduction, carbon reduction, snowfall, rainfall, energy saving and consumption reduction, energy saving, resource-saving type, slag formation, particulate matter, renewable, zero emissions, zero carbon, zero carbon emissions, hydrogen sulfide, hydrogen chloride, energy consumption, energy efficiency, energy, inverter, emissions, total emissions, oxygen emissions, climate, meteorological elements, heavy rain, hydrogen energy, fuel cell, heat diffusion, three wastes, heat dissipation, production wastewater, dual carbon, acid mist, carbon peak, weather, external discharge, exhaust gas, pollution, sewage, no dust emission, infiltration, smoke dust, pollution control, heavy metals, resource-saving type, total nitrogen, dust capacity.

Appendix C

Figure A1. Annual trend in CRD and third-party ESG ratings.
Figure A1. Annual trend in CRD and third-party ESG ratings.
Sustainability 17 03178 g0a1

References

  1. Huang, H.H.; Kerstein, J.; Wang, C. The Impact of Climate Risk on Firm Performance and Financing Choices. J. Int. Bus. Stud. 2018, 49, 633–656. [Google Scholar]
  2. Ding, R.; Liu, M.; Wang, T.; Wu, Z. The Impact of Climate Risk on Earnings Management: International Evidence. J. Account. Public Policy 2021, 40, 106818. [Google Scholar] [CrossRef]
  3. Carvalho, V.M.; Nirei, M.; Saito, Y.U.; Tahbaz-Salehi, A. Supply Chain Disruptions: Evidence from the Great East Japan Earthquake. Q. J. Econ. 2021, 136, 1255–1321. [Google Scholar] [CrossRef]
  4. Park, Y.; Hong, P.; Roh, J.J. Supply Chain Lessons from the Catastrophic Natural Disaster in Japan. Bus. Horiz. 2013, 56, 75–85. [Google Scholar] [CrossRef]
  5. Addoum, J.M.; Ng, D.T.; Ortiz-Bobea, A. Temperature Shocks and Establishment Sales. Rev. Financ. Stud. 2020, 33, 1331–1366. [Google Scholar] [CrossRef]
  6. Zhang, L.; Kanagaretnam, K. Climate Disasters and Analysts’ Earnings Forecasts: Evidence from the United States. Eur. Account. Rev. 2024, 1–28. [Google Scholar] [CrossRef]
  7. Kong, D.; Lin, Z.; Wang, Y.; Xiang, J. Natural Disasters and Analysts’ Earnings Forecasts. J. Corp. Financ. 2021, 66, 101860. [Google Scholar] [CrossRef]
  8. Ben-Amar, W.; Herrera, D.C.; Martinez, I. Do Climate Risk Disclosures Matter to Financial Analysts? Bus. Financ. Account. 2024, 51, 2153–2180. [Google Scholar] [CrossRef]
  9. Lin, B.; Wu, N. Climate Risk Disclosure and Stock Price Crash Risk: The Case of China. Int. Rev. Econ. Financ. 2023, 83, 21–34. [Google Scholar] [CrossRef]
  10. Wang, W.; Yuan, D. Earnings Quality and Forward-Looking Information Disclosures: Positive Complement or Negative Substitution? J. Audit Econ. 2014, 29, 48–57. (In Chinese) [Google Scholar]
  11. Cohen, S.; Kadach, I.; Ormazabal, G. Institutional Investors, Climate Disclosure, and Carbon Emissions. J. Account. Econ. 2023, 76, 101640. [Google Scholar] [CrossRef]
  12. Hsu, C.; Novoselov, K.E.; Wang, R. Does Accounting Conservatism Mitigate the Shortcomings of CEO Overconfidence? Account. Rev. 2017, 92, 77–101. [Google Scholar] [CrossRef]
  13. Ilhan, E.; Krueger, P.; Sautner, Z.; Starks, L.T. Climate Risk Disclosure and Institutional Investors. Rev. Financ. Stud. 2023, 36, 2617–2650. [Google Scholar] [CrossRef]
  14. Cui, X.; Ding, Q.; Yang, L.; Yu, C. Environmental Protection Tax Reform and Corporate Climate Risk Disclosure. J. Clean. Prod. 2024, 478, 143931. [Google Scholar] [CrossRef]
  15. Matsumura, E.M.; Prakash, R.; Vera-Muñoz, S.C. Climate-Risk Materiality and Firm Risk. Rev. Account. Stud. 2024, 29, 33–74. [Google Scholar] [CrossRef]
  16. Li, Y.; Wang, D.; Meng, D.; Hu, Y. Peer Effect on Climate Risk Information Disclosure. China J. Account. Res. 2024, 17, 100375. [Google Scholar] [CrossRef]
  17. Dhaliwal, D.S.; Radhakrishnan, S.; Tsang, A.; Yang, Y.G. Nonfinancial Disclosure and Analyst Forecast Accuracy: International Evidence on Corporate Social Responsibility Disclosure. Account. Rev. 2012, 87, 723–759. [Google Scholar] [CrossRef]
  18. Shao, J.; Wu, Y.; Ye, Z. Does Carbon Disclosure Levels Affect Analyst Forecasts? Evidence from China. Sustainability 2025, 17, 1152. [Google Scholar] [CrossRef]
  19. Muslu, V.; Radhakrishnan, S.; Subramanyam, K.R.; Lim, D. Forward-Looking MD&A Disclosures and the Information Environment. Manag. Sci. 2015, 61, 931–948. [Google Scholar] [CrossRef]
  20. Wang, X.; Li, Y.; Xiao, M. Do Risk Disclosures in Annual Reports Improve Analyst Forecast Accuracy? China J. Account. Stud. 2017, 5, 527–546. [Google Scholar] [CrossRef]
  21. Huang, X.; Teoh, S.H.; Zhang, Y. Tone Management. Account. Rev. 2014, 89, 1083–1113. [Google Scholar] [CrossRef]
  22. Krueger, P.; Sautner, Z.; Starks, L.T. The Importance of Climate Risks for Institutional Investors. Rev. Financ. Stud. 2020, 33, 1067–1111. [Google Scholar] [CrossRef]
  23. Zhang, Y.; He, M.; Liao, C.; Wang, Y. Climate Risk Exposure and the Cross-Section of Chinese Stock Returns. Fin. Res. Lett. 2023, 55, 103987. [Google Scholar] [CrossRef]
  24. Huang, H.H.; Kerstein, J.; Wang, C.; Wu, F. (Harry) Firm Climate Risk, Risk Management, and Bank Loan Financing. Strateg. Manag. J. 2022, 43, 2849–2880. [Google Scholar] [CrossRef]
  25. Ginglinger, E.; Moreau, Q. Climate Risk and Capital Structure. Manag. Sci. 2023, 69, 7492–7516. [Google Scholar] [CrossRef]
  26. Li, Y.; Zhang, Z. Corporate Climate Risk Exposure and Capital Structure: Evidence from Chinese Listed Companies. Fin. Res. Lett. 2023, 51, 103488. [Google Scholar] [CrossRef]
  27. Hope, O. Disclosure Practices, Enforcement of Accounting Standards, and Analysts’ Forecast Accuracy: An International Study. J. Account. Res. 2003, 41, 235–272. [Google Scholar] [CrossRef]
  28. Schipper, K. Analysts’ Forecasts. Account. Horiz. 1991, 5, 105–121. [Google Scholar]
  29. Amiram, D.; Landsman, W.R.; Owens, E.L.; Stubben, S.R. How Are Analysts’ Forecasts Affected by High Uncertainty? J. Bus. Financ. Account. 2018, 45, 295–318. [Google Scholar] [CrossRef]
  30. Cheng, Q.; Du, F.; Wang, X.; Wang, Y. Seeing Is Believing: Analysts’ Corporate Site Visits. Rev. Account. Stud. 2016, 21, 1245–1286. [Google Scholar] [CrossRef]
  31. Han, B.; Kong, D.; Liu, S. Do Analysts Gain an Informational Advantage by Visiting Listed Companies? Contemp. Account. Res. 2018, 35, 1843–1867. [Google Scholar] [CrossRef]
  32. Wang, H.; Wang, K. Accrual Management and Tone Management. Acct. Res. 2018, 4, 45–51. (In Chinese) [Google Scholar]
  33. Li, F. Annual Report Readability, Current Earnings, and Earnings Persistence. J. Account. Econ. 2008, 45, 221–247. [Google Scholar] [CrossRef]
  34. Lo, K.; Ramos, F.; Rogo, R. Earnings Management and Annual Report Readability. J. Account. Econ. 2017, 63, 1–25. [Google Scholar] [CrossRef]
  35. Wang, K.; Wang, H.; Li, D.; Dai, X. Complexity of Annual Report and Management Self-interest: Empirical Evidence from Chinese Listed Firms. J. Manag. World. 2018, 34, 120–132. (In Chinese) [Google Scholar] [CrossRef]
  36. Lu, D.; Song, X.; Gong, Y. Controlling Shareholder’s Shares Pledge and Readability of Annual Reports. Fin. Trade Res. 2020, 31, 77–96. (In Chinese) [Google Scholar] [CrossRef]
  37. Du, J.; Xu, X.; Yang, Y. Does Corporate Climate Risk Affect the cost of equity? Evidence from Textual analysis with Machine learning. China Rev. Fin. Stud. 2023, 15, 19–46. (In Chinese) [Google Scholar]
  38. Plumlee, M.; Brown, D.; Hayes, R.M.; Marshall, R.S. Voluntary Environmental Disclosure Quality and Firm Value: Further Evidence. J. Account. Public Policy 2015, 34, 336–361. [Google Scholar] [CrossRef]
  39. Ni, Y.; Chen, Z.; Li, D.; Yang, S. Climate Risk and Corporate Tax Avoidance: International Evidence. Corp. Gov. Int. Rev. 2022, 30, 189–211. [Google Scholar] [CrossRef]
  40. Yang, X.; Wei, L.; Deng, R.; Cao, J.; Huang, C. Can Climate-Related Risks Increase Audit Fees?–Evidence from China. Fin. Res. Lett. 2023, 57, 104194. [Google Scholar] [CrossRef]
  41. Balli, H.O.; Sørensen, B.E. Interaction Effects in Econometrics. Empir. Econ. 2013, 45, 583–603. [Google Scholar] [CrossRef]
  42. Zhou, S.; Zhan, Q.; Hu, L. Can Institutional Investors Promote M&A Performance of State-owned Enterprises? Discussion on Institutional Investors Heterogeneity in China. Account. Res. 2017, 6, 67–74. (In Chinese) [Google Scholar]
  43. Hou, K.; Moskowitz, T.J. Market Frictions, Price Delay, and the Cross-Section of Expected Returns. Rev. Financ. Stud. 2005, 18, 981–1020. [Google Scholar] [CrossRef]
  44. Zhao, L.; Parhizgari, A.M. Climate Change, Technological Innovation, and Firm Performance. Int. Rev. Econ. Financ. 2024, 93, 189–203. [Google Scholar] [CrossRef]
Table 1. Variable definition.
Table 1. Variable definition.
VariableDefinition
FERRORThe average absolute difference between the actual EPS and forecasted EPS, divided by the stock price of last year.
FDISPThe standard deviation of forecasted EPS, divided by the stock price of last year.
CRDClimate risk disclosure, measured as the ratio of the climate risk words to the total words in annual reports.
SIZEThe natural logarithm of total assets.
ROAReturn on assets, calculated as net income divided by total assets.
LEVLeverage, calculated as total debts divided by total assets.
GROWTHSales growth, calculated as the change in revenue in the current year divided by revenue in the prior year
BODThe number of directors on the board.
IDThe proportion of the independent directors of the board.
DUALA dummy variable equals 1 if the CEO is also the chairman and 0 otherwise.
TOP1The proportion of shares held by the controlling stockholders.
BIG4A dummy variable equals 1 if the audit firm is one of the international “BIG Four” and 0 otherwise.
FCOVERThe natural logarithm of the number of analysts following the firm plus 1.
Table 2. Descriptive statistics.
Table 2. Descriptive statistics.
VariableNMeanSDMinp50Max
FERROR24,5640.0140.0210.0000.0070.139
FDISP24,5640.0090.0110.0000.0050.068
CRD24,5640.0070.0050.0010.0060.027
SIZE24,56422.4001.34020.00022.20026.500
ROA24,5640.0530.050−0.1200.0470.214
LEV24,5640.4330.2000.0560.4320.865
GROWTH24,5640.2330.449−0.4670.1503.040
BOD24,56410.1002.5305.0009.00018.000
ID24,5640.3810.0710.2500.3640.600
DUAL24,5640.2700.4440.0000.0001.000
TOP124,5640.3560.1520.0890.3380.754
BIG424,5640.0850.2790.0000.0001.000
FCOVER24,5642.1300.8610.0002.0803.850
Table 3. Baseline results.
Table 3. Baseline results.
Variables(1)(2)
FERRORFDISP
CRD−0.183 ***−0.065 ***
(−5.77)(−4.17)
SIZE0.003 ***0.002 ***
(19.65)(22.00)
ROA−0.032 ***−0.007 ***
(−10.08)(−4.47)
LEV0.011 ***0.006 ***
(11.85)(13.47)
GROWTH−0.001 *−0.000
(−1.87)(−0.30)
BOD−0.000 **−0.000
(−2.40)(−1.48)
IDRATE−0.003−0.002 *
(−1.52)(−1.78)
DUAL0.000−0.000
(0.80)(−1.17)
TOP1−0.008 ***−0.001 ***
(−8.34)(−3.17)
BIG4−0.003 ***−0.002 ***
(−6.75)(−6.63)
FCOVER−0.003 ***−0.000 ***
(−18.77)(−3.38)
Constant−0.039 ***−0.022 ***
(−11.97)(−14.01)
Industry FEYesYes
Year FEYesYes
N24,56424,564
R20.1050.107
F61.41062.270
Note: T-statistics reported in the parenthesis are based on standard errors clustered at the firm level. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 4. Results of robustness tests.
Table 4. Results of robustness tests.
Variables(1)(2)(3)(4)(5)(6)
FERRORFDISPFERRORFDISPFERRORFDISP
STDCRD−0.001 ***−0.000 ***
(−5.89)(−4.35)
LNCRD −0.001 ***−0.000 ***
(−6.10)(−4.06)
CRD −0.166 ***−0.050 ***
(−5.19)(−3.08)
SIZE0.003 ***0.002 ***0.003 ***0.002 ***0.003 ***0.002 ***
(19.67)(22.02)(19.79)(22.03)(19.34)(21.17)
ROA−0.032 ***−0.007 ***−0.032 ***−0.007 ***−0.032 ***−0.009 ***
(−10.08)(−4.48)(−10.25)(−4.58)(−10.11)(−5.58)
LEV0.011 ***0.006 ***0.011 ***0.006 ***0.011 ***0.006 ***
(11.86)(13.48)(11.65)(13.33)(11.47)(12.49)
GROWTH−0.001 *−0.000−0.001 *−0.000−0.000−0.000
(−1.87)(−0.29)(−1.76)(−0.24)(−1.34)(−0.71)
BOD−0.000 **−0.000−0.000 **−0.000−0.000 *−0.000
(−2.41)(−1.49)(−2.32)(−1.43)(−1.78)(−0.60)
IDRATE−0.003−0.002*−0.002−0.002*−0.001−0.002 *
(−1.53)(−1.78)(−1.34)(−1.65)(−0.79)(−1.95)
DUAL0.000−0.0000.000−0.0000.000−0.000
(0.82)(−1.16)(1.00)(−1.05)(0.42)(−1.12)
TOP1−0.008 ***−0.001 ***−0.008 ***−0.001 ***−0.006 ***−0.001 **
(−8.32)(−3.16)(−8.44)(−3.23)(−7.04)(−2.02)
BIG4−0.003 ***−0.002 ***−0.003 ***−0.002 ***−0.003 ***−0.002 ***
(−6.77)(−6.65)(−6.51)(−6.45)(−6.71)(−6.09)
FCOVER−0.003 ***−0.000 ***−0.003 ***−0.000 ***−0.003 ***−0.000 ***
(−18.77)(−3.38)(−18.56)(−3.25)(−16.98)(−3.12)
Constant−0.040 ***−0.023 ***−0.035 ***−0.021 ***−0.042 ***−0.023 ***
(−12.30)(−14.22)(−10.69)(−13.10)(−12.84)(−14.34)
Industry FEYesYesYesYesYesYes
Year FEYesYesYesYesYesYes
N24,56424,56424,56424,56421,39921,399
R20.1050.1070.1050.1070.1130.118
F61.45062.30061.50062.25060.54063.200
Note: Columns (1) and (2) are the results of adjusting CRD by its industry–year average. Columns (3) and (4) are the results of transforming CRD into its natural logarithm. Columns (5) and (6) are the results of excluding samples during the COVID-19 pandemic period. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 5. Results of heterogeneity tests.
Table 5. Results of heterogeneity tests.
Variables(1)(2)(3)(4)(5)(6)(7)(8)
FERRORFDISPFERRORFDISPFERRORFDISPFERRORFDISP
CRD−0.183 ***−0.065 ***−0.087 **−0.043 **−0.186 **−0.719 ***
(−5.78)(−4.13)(−2.14)(−2.16)(−2.04)(−15.26)
TREAT 0.0020.001
(1.04)(0.96)
POST 0.012 ***−0.002 **
(7.48)(−2.02)
TREAT × POST −0.005 ***−0.002 ***
(−5.70)(−3.39)
SIZE0.0010.005 ***0.003 ***0.001 ***0.004 ***0.002 ***0.004 ***0.002 ***
(0.27)(3.32)(11.98)(11.28)(22.84)(27.02)(18.11)(18.43)
ROA−0.016−0.029 ***−0.028 ***−0.003−0.0000.0000.0000.000
(−0.75)(−2.80)(−5.51)(−1.16)(−0.73)(0.04)(0.02)(0.40)
LEV0.014 ***0.0020.010 ***0.008 ***0.007 ***0.004 ***0.005 ***0.002 ***
(3.51)(1.03)(6.36)(10.03)(11.56)(12.08)(5.67)(5.16)
GROWTH−0.0020.002 **−0.001−0.000−0.000−0.000−0.000−0.000
(−1.04)(2.09)(−1.11)(−0.51)(−0.88)(−0.93)(−0.59)(−0.49)
BOD−0.000 *0.000−0.000 *0.000−0.000 ***−0.000 ***−0.000 **−0.000
(−1.93)(1.05)(−1.83)(1.06)(−3.60)(−2.70)(−2.33)(−1.44)
IDRATE0.004−0.010 **0.000−0.001−0.004 **−0.003 ***−0.005 *−0.003 *
(0.43)(−2.49)(0.16)(−0.49)(−2.06)(−3.36)(−1.77)(−1.88)
DUAL−0.0010.002 *0.000−0.001 ***0.001 **0.0000.001 **0.000
(−0.64)(1.92)(0.21)(−2.58)(2.05)(0.97)(2.00)(0.69)
TOP1−0.004−0.006 ***−0.010 ***−0.000−0.010 ***−0.003 ***−0.009 ***0.000
(−0.90)(−2.74)(−6.52)(−0.16)(−10.33)(−5.32)(−5.86)(0.34)
BIG40.000−0.006 ***−0.004 ***−0.002 ***−0.004 ***−0.003 ***−0.004 ***−0.002 ***
(0.01)(−2.89)(−5.12)(−3.79)(−7.29)(−9.27)(−5.22)(−4.83)
FCOVER−0.004 ***0.001−0.004 ***−0.000 *−0.004 ***−0.001 ***−0.005 ***−0.000 ***
(−4.25)(1.49)(−12.31)(−1.68)(−26.61)(−7.07)(−17.99)(−3.26)
IMR−0.0300.041 **
(−0.78)(2.15)
Constant0.028−0.114 ***−0.040 ***−0.014 ***−0.062 ***−0.037 ***−0.052 ***−0.026 ***
(0.33)(−2.68)(−6.52)(−5.22)(−17.18)(−20.00)(−11.59)(−11.71)
Industry FEYesYesYesYesYesYesYesYes
Year FEYesYesYesYesYesYesYesYes
N24,56424,5649402940224,56424,56412,73012,730
R20.1050.1070.0950.0960.0770.0850.0990.098
F60.14061.08020.85021.97066.53060.25028.91028.770
Note: Columns (1) and (2) are the second-stage results of Heckman’s two-stage model. Columns (3) and (4) are the results of PSM. Columns (5) and (6) are the results of IV analysis. Columns (7) and (8) are the results of the PSM-DID model. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 6. Results of moderating effects.
Table 6. Results of moderating effects.
Variables(1)(2)(3)(4)
FERRORFDISPFERRORFDISP
CRD−0.171 ***−0.065 ***−0.180 ***−0.064 ***
(−5.69)(−4.39)(−6.04)(−4.30)
EQ−0.002 ***−0.000 **
(−4.51)(−2.40)
CRD × EQ−0.401 ***−0.105 **
(−4.48)(−2.39)
LINS −0.053 ***−0.008 ***
(−10.96)(−3.55)
CRD × LINS −2.244 ***0.477
(−2.64)(1.13)
SIZE0.003 ***0.002 ***0.003 ***0.002 ***
(19.97)(22.14)(17.29)(21.00)
ROA−0.031 ***−0.007 ***−0.031 ***−0.007 ***
(−9.83)(−4.23)(−9.96)(−4.38)
LEV0.011 ***0.006 ***0.012 ***0.006 ***
(11.78)(13.50)(12.65)(13.71)
GROWTH−0.001 *−0.000−0.001 **−0.000
(−1.94)(−0.46)(−2.00)(−0.34)
BOD−0.000 **−0.000−0.000 **−0.000
(−2.46)(−1.43)(−2.50)(−1.51)
IDRATE−0.003−0.002*−0.003−0.002 *
(−1.39)(−1.74)(−1.52)(−1.81)
DUAL0.000−0.0000.000−0.000
(0.78)(−1.17)(0.73)(−1.20)
TOP1−0.007 ***−0.001 ***−0.008 ***−0.002 ***
(−8.18)(−3.03)(−9.36)(−3.50)
BIG4−0.003 ***−0.002 ***−0.004 ***−0.002 ***
(−6.64)(−6.57)(−7.07)(−6.78)
FCOVER−0.003 ***−0.000 ***−0.003 ***−0.000 *
(−18.83)(−3.54)(−13.87)(−1.90)
Constant−0.042 ***−0.024 ***−0.034 ***−0.022 ***
(−12.68)(−14.48)(−10.29)(−13.43)
Industry FEYesYesYesYes
Year FEYesYesYesYes
N24,24824,24824,56424,564
R20.1060.1070.1100.107
F59.49060.63061.71060.100
Note: Columns (1) and (2) are the moderating effect of earnings quality. Columns (3) and (4) are the moderating effect of long-term institutional investor ownership. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 7. Results of mechanism analysis: information asymmetry.
Table 7. Results of mechanism analysis: information asymmetry.
(1)(2)(3)(4)(5)
VariablesFERRORDELAYFERRORFDISPFDISP
CRD−0.163 ***−0.777 ***−0.153 ***−0.066 ***−0.063 ***
(−4.27)(−5.16)(−4.00)(−3.62)(−3.44)
DELAY 0.013 *** 0.004 ***
(6.94) (4.68)
SIZE0.003 ***−0.004 ***0.003 ***0.002 ***0.002 ***
(16.42)(−5.92)(16.73)(19.20)(19.40)
ROA−0.031 ***−0.016−0.031 ***−0.008 ***−0.008 ***
(−7.93)(−1.00)(−7.89)(−4.08)(−4.05)
LEV0.011 ***0.037 ***0.010 ***0.007 ***0.006 ***
(9.59)(8.22)(9.16)(11.99)(11.69)
GROWTH−0.001 **0.002−0.001 **−0.000−0.000
(−2.04)(1.16)(−2.11)(−1.59)(−1.63)
BOD−0.000 ***−0.000−0.000 ***−0.000 **−0.000 **
(−3.30)(−0.83)(−3.26)(−2.01)(−1.98)
IDRATE−0.003−0.013−0.003−0.001−0.001
(−1.58)(−1.46)(−1.51)(−1.27)(−1.22)
DUAL0.000−0.0020.000−0.000−0.000
(0.37)(−1.35)(0.44)(−1.07)(−1.02)
TOP1−0.007 ***0.016 ***−0.008 ***−0.001−0.001
(−6.70)(3.69)(−6.90)(−1.11)(−1.24)
BIG4−0.003 ***0.001−0.003 ***−0.002 ***−0.002 ***
(−5.37)(0.51)(−5.40)(−5.64)(−5.66)
FCOVER−0.003 ***−0.005 ***−0.003 ***−0.000 *−0.000 *
(−15.60)(−5.58)(−15.32)(−1.86)(−1.66)
Constant−0.030 ***0.175 ***−0.032 ***−0.020 ***−0.021 ***
(−7.41)(11.13)(−7.97)(−10.58)(−10.94)
Industry FEYesYesYesYesYes
Year FEYesYesYesYesYes
N17,89817,89817,89817,89817,898
R20.1030.2300.1050.1100.111
F49.780130.20049.87054.00053.290
Note: ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 8. Results of mechanism analysis: analysts’ climate-related on-site visits.
Table 8. Results of mechanism analysis: analysts’ climate-related on-site visits.
Variables(1)(2)(3)(4)(5)
FERRORSURVEYFERRORFDISPFDISP
CRD−0.183 ***0.403 ***−0.178 ***−0.120 ***−0.118 ***
(−5.77)(7.59)(−5.63)(−7.80)(−7.64)
SURVEY −0.011 *** −0.004 **
(−2.77) (−1.98)
SIZE0.003 ***−0.002 ***0.003 ***0.001 ***0.001 ***
(19.65)(−7.10)(19.51)(18.54)(18.49)
ROA−0.032 ***−0.015 ***−0.032 ***−0.008 ***−0.008 ***
(−10.08)(−2.91)(−10.13)(−4.99)(−5.02)
LEV0.011 ***−0.008 ***0.011 ***0.007 ***0.007 ***
(11.85)(−5.10)(11.75)(16.03)(15.92)
GROWTH−0.001 *0.001 *−0.001 *−0.000−0.000
(−1.87)(1.66)(−1.84)(−0.52)(−0.50)
BOD−0.000 **0.000−0.000 **−0.000−0.000
(−2.40)(0.63)(−2.39)(−1.13)(−1.12)
IDRATE−0.0030.010 ***−0.003−0.002 ***−0.002 ***
(−1.52)(3.29)(−1.47)(−2.66)(−2.61)
DUAL0.0000.003 ***0.000−0.000 ***−0.000 **
(0.80)(5.19)(0.89)(−2.61)(−2.53)
TOP1−0.008 ***−0.012 ***−0.008 ***−0.001 *−0.001 *
(−8.34)(−7.72)(−8.47)(−1.69)(−1.81)
BIG4−0.003 ***−0.000−0.003 ***−0.001 ***−0.001 ***
(−6.75)(−0.36)(−6.76)(−5.92)(−5.94)
FCOVER−0.003 ***0.006 ***−0.003 ***−0.000−0.000
(−18.77)(19.18)(−18.30)(−1.32)(−1.10)
Constant−0.039 ***0.032 ***−0.038 ***−0.016 ***−0.016 ***
(−11.97)(5.90)(−11.85)(−10.09)(−10.03)
Industry FEYesYesYesYesYes
Year FEYesYesYesYesYes
N24,56424,56424,56424,56424,564
R20.1050.1160.1060.0960.097
F61.41068.66060.31058.16056.990
Note: ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 9. Results of heterogeneity analysis at regional level.
Table 9. Results of heterogeneity analysis at regional level.
Variables(1)(2)(3)(4)
Regions with High Climate AwarenessRegions with Low Climate AwarenessRegions with High Climate AwarenessRegions with Low Climate Awareness
FERRORFERRORFDISPFDISP
CRD−0.222 ***−0.124 ***−0.068 **−0.046 **
(−3.97)(−2.87)(−2.52)(−2.36)
SIZE0.004 ***0.003 ***0.002 ***0.001 ***
(13.70)(12.65)(15.44)(14.90)
ROA−0.029 ***−0.035 ***−0.003−0.007 ***
(−5.17)(−7.75)(−1.04)(−3.47)
LEV0.015 ***0.007 ***0.008 ***0.004 ***
(9.33)(5.46)(10.68)(7.11)
GROWTH−0.001 ***0.000−0.0000.001 **
(−2.61)(0.96)(−1.51)(2.32)
BOD−0.000 ***−0.000 *−0.000 *−0.000
(−2.76)(−1.73)(−1.73)(−1.22)
IDRATE−0.007 **0.000−0.0030.000
(−2.14)(0.03)(−1.61)(0.28)
DUAL0.001−0.000−0.000−0.000
(1.08)(−0.33)(−0.30)(−1.16)
TOP1−0.008 ***−0.008 ***0.000−0.003 ***
(−5.18)(−6.35)(0.12)(−4.65)
BIG4−0.003 ***−0.004 ***−0.002 ***−0.002 ***
(−3.55)(−5.04)(−3.72)(−5.24)
FCOVER−0.003 ***−0.004 ***−0.000 *−0.000 ***
(−10.80)(−14.25)(−1.79)(−3.09)
Constant−0.049 ***−0.020 ***−0.033 ***−0.011 ***
(−8.90)(−4.06)(−12.52)(−4.78)
Industry FEYesYesYesYes
Year FEYesYesYesYes
N11,735955911,7309557
R20.1070.1140.1060.117
F32.59028.37032.20029.410
Empirical p-values0.072 *0.245
Note: Empirical p-values are calculated using bootstrapping with 1000 repetitions. ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 10. Results of heterogeneity analysis at industry level.
Table 10. Results of heterogeneity analysis at industry level.
Variables(1)(2)(3)(4)
Carbon-Intensive
Industries
Non-Carbon-Intensive IndustriesCarbon-Intensive
Industries
Non-Carbon-Intensive Industries
FERRORFERRORFDISPFDISP
CRD−0.184 ***−0.109−0.077 ***0.008
(−5.92)(−1.56)(−4.92)(0.22)
SIZE0.003 ***0.003 ***0.002 ***0.002 ***
(16.21)(11.00)(20.53)(10.62)
ROA−0.027 ***−0.038 ***−0.007 ***−0.005 *
(−7.43)(−5.90)(−3.60)(−1.72)
LEV0.011 ***0.004 **0.006 ***0.003 ***
(11.22)(2.40)(11.47)(3.64)
GROWTH−0.001−0.000−0.000−0.000
(−1.44)(−0.45)(−0.03)(−0.26)
BOD−0.000−0.000 **−0.000−0.000
(−1.63)(−2.10)(−0.75)(−0.84)
IDRATE−0.001−0.006−0.001−0.003 *
(−0.41)(−1.46)(−1.29)(−1.68)
DUAL0.0000.001−0.000−0.000
(0.42)(0.79)(−0.68)(−1.14)
TOP1−0.006 ***−0.010 ***−0.001 ***−0.001
(−6.27)(−5.69)(−2.77)(−0.86)
BIG4−0.003 ***−0.005 ***−0.002 ***−0.002 ***
(−5.93)(−4.59)(−6.21)(−3.59)
FCOVER−0.003 ***−0.004 ***−0.000 ***−0.000
(−15.56)(−10.28)(−3.70)(−0.71)
Constant−0.041 ***−0.047 ***−0.027 ***−0.023 ***
(−12.18)(−7.83)(−16.30)(−7.84)
Industry FEYesYesYesYes
Year FEYesYesYesYes
N18,199636518,1996365
R20.0960.1160.1110.069
F71.51030.71084.24017.370
Empirical p-values0.090 *0.000 ***
Note: ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 11. Results of heterogeneity analysis at firm level.
Table 11. Results of heterogeneity analysis at firm level.
Variables(1)(2)(3)(4)
SOEsNon-SOEsSOEsNon-SOEs
FERRORFERRORFDISPFDISP
CRD−0.253 ***−0.129 ***−0.122 ***−0.027
(−4.26)(−3.48)(−3.89)(−1.51)
SIZE0.003 ***0.003 ***0.002 ***0.002 ***
(11.76)(15.45)(13.04)(16.40)
ROA−0.035 ***−0.027 ***−0.012 ***−0.002
(−5.77)(−7.39)(−3.79)(−1.36)
LEV0.016 ***0.008 ***0.010 ***0.004 ***
(9.77)(6.98)(11.24)(7.90)
GROWTH−0.002 ***−0.000−0.001 **0.000
(−3.21)(−0.12)(−2.31)(1.30)
BOD−0.000 *−0.000−0.000−0.000
(−1.93)(−0.31)(−0.20)(−1.01)
IDRATE−0.004−0.004 *−0.002−0.002 **
(−1.23)(−1.87)(−0.88)(−2.41)
DUAL−0.0000.000−0.000−0.000 *
(−0.61)(0.24)(−0.87)(−1.88)
TOP1−0.004 **−0.008 ***−0.000−0.002 ***
(−2.57)(−6.58)(−0.08)(−3.52)
BIG4−0.004 ***−0.003 ***−0.002 ***−0.002 ***
(−5.25)(−4.39)(−4.66)(−5.41)
FCOVER−0.003 ***−0.003 ***−0.000 *−0.000 ***
(−10.41)(−15.92)(−1.69)(−3.38)
Constant−0.050 ***−0.036 ***−0.033 ***−0.015 ***
(−9.31)(−8.00)(−11.69)(−7.28)
Industry FEYesYesYesYes
Year FEYesYesYesYes
N934314,733934314,733
R20.1170.1120.1280.099
F26.81039.46029.62034.220
Empirical p-values0.022 **0.001 ***
Note: ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Table 12. Effect of PCRD and TCRD on analysts’ forecasts.
Table 12. Effect of PCRD and TCRD on analysts’ forecasts.
Variables(1)(2)(3)(4)
FERRORFDISPFERRORFDISP
PCRD−0.6880.265
(−1.23)(0.95)
TCRD −0.193 ***−0.071 ***
(−5.98)(−4.42)
SIZE0.003 ***0.002 ***0.003 ***0.002 ***
(19.17)(21.45)(19.66)(22.02)
ROA−0.031 ***−0.007 ***−0.032 ***−0.007 ***
(−9.94)(−4.35)(−10.08)(−4.48)
LEV0.011 ***0.006 ***0.011 ***0.006 ***
(11.78)(13.41)(11.85)(13.47)
GROWTH−0.001 **−0.000−0.001 *−0.000
(−2.16)(−0.54)(−1.87)(−0.29)
BOD−0.000 **−0.000−0.000 **−0.000
(−2.36)(−1.46)(−2.40)(−1.49)
IDRATE−0.003−0.001−0.003−0.002 *
(−1.38)(−1.63)(−1.53)(−1.78)
DUAL0.000−0.0000.000−0.000
(0.69)(−1.26)(0.80)(−1.17)
TOP1−0.007 ***−0.001 ***−0.008 ***−0.001 ***
(−8.23)(−3.10)(−8.34)(−3.18)
BIG4−0.003 ***−0.002 ***−0.003 ***−0.002 ***
(−6.44)(−6.32)(−6.75)(−6.64)
FCOVER−0.003 ***−0.000 ***−0.003 ***−0.000 ***
(−18.87)(−3.50)(−18.78)(−3.38)
Constant−0.038 ***−0.022 ***−0.039 ***−0.023 ***
(−11.62)(−13.67)(−11.98)(−14.03)
Industry FEYesYesYesYes
Year FEYesYesYesYes
N24,56424,56424,56424,564
R20.1040.1060.1050.107
F60.66061.88061.47062.320
Note: ***, **, and * indicate significance at 1%, 5%, and 10% levels, respectively.
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Liu, Y.; Han, J. Climate Risk Disclosure and Financial Analysts’ Forecasts: Evidence from China. Sustainability 2025, 17, 3178. https://doi.org/10.3390/su17073178

AMA Style

Liu Y, Han J. Climate Risk Disclosure and Financial Analysts’ Forecasts: Evidence from China. Sustainability. 2025; 17(7):3178. https://doi.org/10.3390/su17073178

Chicago/Turabian Style

Liu, Yaoyao, and Jie Han. 2025. "Climate Risk Disclosure and Financial Analysts’ Forecasts: Evidence from China" Sustainability 17, no. 7: 3178. https://doi.org/10.3390/su17073178

APA Style

Liu, Y., & Han, J. (2025). Climate Risk Disclosure and Financial Analysts’ Forecasts: Evidence from China. Sustainability, 17(7), 3178. https://doi.org/10.3390/su17073178

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop