ESG rating disagreements and stock price crash risk: can distraction, regional digital economy and corporate intelligence maturity play important roles?

Yiyuan Li (College of Business and Economics, Australian National University, Canberra, Australia)

Asian Journal of Accounting Research

ISSN: 2459-9700

Article publication date: 29 October 2024

599

Abstract

Purpose

ESG issues are gaining increasing attention from investors, but the environmental, social and governance (ESG) rating disagreement caused by different standards of rating agencies misleads investors' investment decisions. This can lead to an increased risk of stock price crashes, causing turbulence in the financial markets and reducing investors' confidence. The paper investigates whether ESG rating disagreement of the current period increases stock price crash risk and the mechanism to mitigate this impact.

Design/methodology/approach

With the sample of the listed companies of Shanghai and Shenzhen Stock Exchanges from 2010 to 2022 this paper examines the impact of ESG rating disagreement itself on stock price crash risk. Moreover, this paper examines the mechanisms by analyzing the moderating effect of distraction of investors; digital economy and corporate intelligence maturity.

Findings

This paper finds that ESG rating disagreement itself would amplify the stock price crash risk. When exploring the moderating effect of institutional investors' distraction, digital economic development level and corporate intelligence, the paper found that they would mitigate the impact of ESG rating disagreement on stock price crash risk. The relationship between ESG rating disagreement and stock price crash risk is more pronounced in the context of heavily-polluted, state-owned enterprises (SOEs) and enterprises with star analysts.

Originality/value

Currently, few articles discuss ESG rating disagreement, especially the impact of current ESG rating disagreement on stock price crash risk. This paper focuses on this topic and provides strategies to mitigate the impact of current ESG rating divergence on stock price crash risk.

Keywords

Citation

Li, Y. (2024), "ESG rating disagreements and stock price crash risk: can distraction, regional digital economy and corporate intelligence maturity play important roles?", Asian Journal of Accounting Research, Vol. ahead-of-print No. ahead-of-print. https://doi.org/10.1108/AJAR-04-2024-0148

Publisher

:

Emerald Publishing Limited

Copyright © 2024, Yiyuan Li

License

Published in Asian Journal of Accounting Research. Published by Emerald Publishing Limited. This article is published under the Creative Commons Attribution (CC BY 4.0) licence. Anyone may reproduce, distribute, translate and create derivative works of this article (for both commercial and non-commercial purposes), subject to full attribution to the original publication and authors. The full terms of this licence may be seen at http://creativecommons.org/licences/by/4.0/legalcode


1. Introduction

Nowadays, as sustainable development becomes an internationalized common goal, environmental, social and governance (ESG) disclosure and ESG rating are receiving increasing attention. Custom Market Insights has released the report, which illustrates that the global Environmental, Social and Governance Investing Market has achieved a valuation of USD 17.2 trillion by 2022 and analysts anticipate that the valuation will soar to USD 46.5 trillion by 2032 (Custom Market Insights, 2023). The company’s strong ESG performance is becoming increasingly important for attracting external investor investments (Kotsantonis et al., 2016). Due to variations in the criteria, the same company may receive different ESG ratings from different ESG rating agencies (Billio et al., 2021; Christensen et al., 2022). Different ESG ratings convey the confusing investment information for investors, undoubtedly increasing the opacity of companies to investors and introducing significant investment risks (Christensen et al., 2022; Fu and Li, 2023).

Avramov et al. (2022) found investors may be more inclined to seek higher returns to compensate for potential ESG-related risks when considering ESG uncertainty. Gibson Brandon et al. (2021) holds that the increased ESG rating disagreement may be interpreted as a factor contributing to uncertainty, leading to the demand for an uncertainty premium. Although there are studies about ESG rating disagreement and stock return (Avramov et al., 2022; Gibson Brandon et al., 2021; Yan, 2023), there is scarce research on the ESG rating disagreement and stock price crash risk.

Considering this context, the objective of this paper is to address this gap by examining the influence of ESG rating disagreement on the stock price crash risk of corporations. Past studies have indicated that information asymmetry between corporate management and external investors (Hutton et al., 2009), along with the agency costs of the company (Jin and Myers, 2006), can lead to the increased stock price crash risk.

The problem is approached through the examination of Chinese common stocks listed on the main board of the Shanghai and Shenzhen Stock Exchange from 2010 to 2022. The ESG rating data are sourced from six ESG rating agencies: Bloomberg, Wind, Russell, Hexun, SynTao Green Finance and Huazheng Index Information Service. Following Avramov et al. (2022), Hu et al. (2023), the ESG rating disagreement is calculated. Variables for the stock price crash risk are measured following the approach proposed by Cen (2023), using two variables, NCSKEW and DUVOL (Luo et al., 2023). While there are slight methodological differences in calculating these two variables, both of them effectively indicate the level of stock price crash risk. The calculation of ESG rating disagreement follows the methodology outlined by Hu et al. (2023), Luo et al. (2023). The regression results confirm that ESG rating disagreement could increase the stock price crash risk.

The second question is to address under what mechanism ESG rating disagreement reduces its impact on stock price crash risk. This paper explores the impact mechanisms of three moderating variables. The first moderating variable is the distraction of institutional investors. Following the approach proposed by Kempf et al. (2017), a variable is constructed to measure the distraction of institutional investors. Moreover, the level of digital development in the province could also be an important factor to have an impact on this process. According to the study (Gao et al., 2022; Ren et al., 2023), a variable is established to measure the level of digital economic development in the location where the company operates. The results indicate that a high level of regional digitization economy can reduce stock price crash risk. The third moderating variable is the maturity level of intelligence in the company. The intelligent upgrading of companies helps to increase company value and mitigate stock price crash risk (Kim et al., 2021; Nenavani et al., 2024). The companies have a high degree of intelligence experiencing lower stock price crash risk (Zhao et al., 2023b). The variable used to analyze the moderating effect is constructed according to the method proposed in previous research (Manita et al., 2020; Wu et al., 2022).

To test whether the positive impact of ESG rating disagreement on stock price crash risk varies based on the characteristics of the firm, a series of heterogeneity tests is conducted as the following analysis. It is observed that the impact of ESG rating disagreement on stock price crash risk is greater and more significant for heavily-polluted companies, state-owned companies and companies covered by star analysts.

This paper makes contributions to various aspects of the existing literature from three aspects.

First, previous research on the relationship between ESG and stock price crash risk has predominantly focused on ESG disclosure (Ming-quan et al., 2023) and performance (Chen et al., 2023; Fang, 2023; Kim et al., 2014). These articles have not considered the differences in ratings from different rating agencies. However, this paper complements this area of study by investigating ESG rating disagreement. By focusing on differences rather than disclosure and absolute numerical comparisons, insights can be gained into the impact of ESG discrepancies, which possess a confounding nature, on investment risk. Although the research of Dong et al. (2024) has found that the lagged ESG rating disagreement would mitigate stock price crash risk, their research has not explored the impact of ESG rating disagreement of current period on the stock price crash risk. This paper finds that ESG rating disagreement of current period could increase stock price crash risk and fill this gap.

Second, the contribution of this paper lies in exploring the impact of confusing information on stock price crash risk when such misleading information is present in the financial market. Previous studies on information leading to increased stock price crash risk primarily focused on the sudden disclosure of concealed negative information, resulting in a sharp change in investor sentiment, as seen in works (Shi, 2021; Tusheng et al., 2017). The conclusion of this paper is that, like false disclosure information, confusing information also increases the stock price crash risk.

The third contribution is about how investors should respond to such confusing investment signals. The previous perspective was based on information disclosure, that is, disseminating information to the financial market. This paper’s perspective, however, is based on information reception, that is, obtaining information from the financial market. As Sun and Zhao et al. discovered in their study (Sun and Xiao, 2018; Zhao et al., 2023a), it is not the more information disclosed, the better; excessive disclosure can lead to stock price collapse. Investors should respond to the excessive information volume with appropriate distraction. Investors can also decide whether to invest based on exogenous factors such as corporate intelligence and the level of regional digitization development, which are not directly related to the confusing information itself.

The rest of this paper is structured as follows. Section 2 presents the literature review and six hypotheses. Section 3 shows the dataset and the variables used. Section 4 describes the design of research and results. Section 5 presents the results of additional analysis. Section 6 is the conclusion.

2. Literature review and hypothesis development

In this section, the relevant literature is reviewed, and the main hypotheses are proposed.

2.1 ESG rating disagreement and stock price crash risk

ESG rating disagreement reflects information asymmetry, so if ESG information accurately reflected the company’s internal situation, ESG ratings would tend to be consistent (Liu et al., 2024). However, due to the high opacity of the company’s disclosures, there are obvious differences in ESG ratings. Most investors in capital markets rely on professional agencies for rating judgments, so ESG rating disagreement would affect investors' information and investment behavior (Wang et al., 2024). Although investors may still invest in companies with high ESG rating disagreement, they tend to remain vigilant and skeptical. Therefore, the negative performance or adverse news reports regarding the company can trigger a strong reaction from those investors who lack the investment confidence, such as quickly selling off their shares (Kim, 2023). In such cases, the risk of the company’s stock price crashing significantly increases. Then, the first hypothesis is as follows:

H1.

There is a positive relationship between ESG rating disagreement and stock price crash risk.

2.2 The mechanism

The paper explores three moderating mechanisms, which would mitigate the impact of ESG rating disagreement on the stock price crash risk.

2.2.1 Moderating effect of distraction of institutional investors

Because it is more difficult to observe and obtain information about most retailer investors' portfolios, whereas information about institutional investors' portfolios is more readily accessible, this paper examines the behavior of institutional investors. Since institutional investors generally have more expertise, they actively gather market information and maintain more diversified portfolios, the research conclusions may apply to retailer investors with these characteristics (Nguyen et al., 2020). The term “distraction of institutional investors” means that if an investor has investments in a company, and a noteworthy event occurs in another industry that is important to the investor’s portfolio, then the investor is more likely to be distracted from the company.

Institutional investors play a crucial role in the capital markets, and retail investors often make similar reactions based on the judgments and professional analyses of institutional investors, even mimicking their buying and selling behavior (Guan, 2022; Polat, 2020). If institutional investors are distracted and not focused on companies with more ESG rating disagreement, it means that even though retail investors may have doubts about the true situation of these companies and lack investment confidence, the absence of institutional investors' analysis and the lack of stock selling behavior from institutional investors will reduce the actual selling of stocks by retail investors due to negative information. Additionally, because institutional investors have shifted their attention to other sector stocks within their portfolios, they themselves will not significantly change their holdings in companies with high ESG rating disagreement (Schmidt, 2019). Therefore, the risk of a stock price crash will decrease. Based on this, hypothesis H2 is proposed.

H2.

With institutional investors' distraction plays a moderating role, the combined impact of ESG rating disagreement and distraction of institutional investors will lead to a negative shift in stock price crash risk.

2.2.2 Moderating effect of digital economic development level of the province

The maturity of digital infrastructure and technology in the region where a company is located contributes to the company delivering more authentic and accurate ESG reports (Ren et al., 2023). In regions with a high level of digital economic development, investors can utilize advanced digital technologies to search for real ESG information and analyze those information by themselves (Junyan et al., 2023). Therefore, they do not have to rely solely on the ratings provided by ESG rating agencies. Even if they still rely on these ratings, they can obtain additional information to assist their judgment when there are discrepancies between ESG ratings. This will increase their confidence in investing in the company and reduce doubts caused by ESG rating disagreements, which helps to mitigate stock price crash risk. Therefore, the third hypothesis of this paper is as follows:

H3.

With the digital economic development level of the province serving as a moderator, the joint effect of ESG rating disagreement and the digitization level of the province will result in a negative shift in stock price crash risk.

2.2.3 Moderating effect of corporate intelligence maturity of the firm

When a company’s smart infrastructure, including blockchain, the Cloud Computing, the Internet of Things (IoT) and Artificial Intelligence, is well-established with a high level of sophistication, it can positively impact the enhancement of ESG performance and contribute to the increase in the company’s value (Kim et al., 2021; Nenavani et al., 2024). The intelligence transformation of a company can drive technological innovation and reduce information asymmetry. This significantly diminishes the operational risks for the company (Liu and Wang, 2023). By disclosing information related to its intangible assets concerning digitalization and intelligence, a company can reduce its stock price crash risk. Based on the above perspective, hypothesis H4 is proposed.

H4.

With corporate intelligence maturity of the firm acting as a moderator, the combined influence of ESG rating disagreement and corporate intelligence maturity of the firm results in a negative alteration in stock price crash risk.

2.3 Further analyses on industry and ownership heterogeneity

First, companies with high levels of pollution, which would experience less distraction of investors due to their significant pollution attributes, experience a more pronounced enhancement in corporate value through ESG performance (Yin et al., 2023). Polluting industries will be more motivated to demonstrate environmentally friendly practices to the external environment through ESG performance (Hsu and Chen, 2023). Therefore, the impact of ESG rating disagreement on stock price crash risk is greater and more significant for heavily-polluted firms. To validate the points mentioned above, the following hypothesis is proposed.

H5.

The ESG rating disagreement of heavily-polluted firms will have a more pronounced positive impact on the stock price crash risk.

Second, as state-owned enterprises (SOEs) can more keenly grasp policy demands, the impact of ESG rating disagreement on the stock price crash is more significant for SOEs (Jin and Wu, 2023). Moreover, implicit social responsibilities are typically shouldered by SOEs. Non-state-owned firms are often driven by profit motives (Chen et al., 2020). Some factors of enterprise have the more pronounced impact on state-owned firms in terms of analysts’ coverage and equity financing costs (Chen et al., 2011). Therefore, the stock price crash risk of SOEs will experience notably correlated movements under the influence of ESG rating disagreement. These arguments lead to the last hypothesis:

H6.

A more notable positive effect on stock price crash risk is observed with the ESG rating disagreement of SOEs.

Third, this paper considers whether the company’s analysts include star analysts. The star analysts discussed in this paper are those recognized as such by the New Fortune magazine. Star analysts are better at utilizing information related to companies to make predictions (Desai et al., 2000). Star analysts not only find it easier to access company information, but they also amplify the impact of company-specific information on stock prices (Xu et al., 2013). For the companies with the coverage of star analysts, this paper believes that ESG rating disagreement will have a more significant impact on the risk of stock price crash.

H7.

Compared to companies without star analysts, the positive impact of ESG rating disagreement on the risk of stock price crash is more pronounced for companies with star analysts.

3. Data

3.1 Data sources

The sample includes all stock of listed companies on the main board of Shanghai and Shenzhen Stock Exchanges. The ESG rating data comes from six ESG rating agencies – Bloomberg, Wind, Russel, Hexun, SynTao Green Finance and Huazheng Index Information Service. Those ESG rating agencies are highly familiar with China market, and the ESG rating data they release is authoritative and widely utilized in previous research (Hu et al., 2023). Other data, including financial statement data, are from China Stock Market and Accounting Research Database (CSMAR) and WIND database. This paper utilizes Excel and Stata17 for data cleaning, processing and regression analysis.

The time span of the sample data is 13 years, from 2010 to 2022. The financial companies, the companies which are listed as “Special Treatment” by Shanghai and Shenzhen Stock Exchanges and the companies which have missing values are excluded from the sample. All continuous variables in the article have been winsorized.

3.2 Main variables

3.2.1 Dependent variable: stock price crash risk

Stock price crash risk refers to the presence of significant declines in stock prices, characterized by extreme downward movements (Cen, 2023). The variable constructed to measure stock price crash risk follows the approach (Cen, 2023). There are two variables used to measure the stock price crash risk. The first one is NCSKEW, and the other one is DUVOL.

Initially, the weekly specific return ri,T for firm i in week T is computed as follows. rm is the weighted market return. rm,T is the weighted market return in week T.

(1)ri,T=ci+β1,irm,T+2+β2,irm,T+1+β3,irm,T+β4,irm,T1+β5,irm,T2+ϵi,T

To measure Fi,T, which is the weekly return specific to the firm, the residual of above Equation (1) is used.

(2)Fi,T=ln(1+ϵˆi,T)

Using Fi,T, NCSKEW and DUVOL could be calculated.

(3)NCSKEWi,T=n(n1)32Fi,T3(n1)(n2)(Fi,T2)32

In the above equation, “n” is the quantity of trading weeks annually for firm i.

(4)DUVOLi,T=log((nu1)DownFi,T2(nd1)UpFi,T2)

nu” is the quantity of weeks during which Fi,T exceeds the annual average return. “nd” is the quantity of weeks during which Fi,T does not exceed the annual average return.

3.2.2 Main independent variable: ESG rating disagreement

Different ESG rating agencies employ varying scoring approaches, so all ESG rating scores are standardized to the range of 1–10 (Huang et al., 2022). The ESG rating disagreement is calculated following Avramov et al. (2022), Hu et al. (2023). First, to form pairwise combinations between ESG rating agencies. For one specific firm, the difference in ratings for each pair (ERD_pairwisei,t) is calculated as follows for each year:

(5)ERD_pairwisei,t=|ESG_scorei,a,tESG_scorei,b,t|2

ESG_scorei,a,t is the standardized ESG score for firm i in year t from “a”, which is one of the six agencies, whereas ESG_scorei,b,t is the standardized ESG score for firm i in year t from “b”, which is another agency. ERD_pairwisei,t could measure the ESG rating difference between the pair of agencies for the specific firm i in year t. Among the six ESG rating agencies, a maximum of 15 ERD_pairwisei,t can be formed.

Second, to calculate the mean of ERD_pairwisei,t, which would be the ESG rating disagreement for the specific firm i on year t.

(6)ERDi,t=n=1NERD_pairwisei,tN

In the formula, “N” represents the number of pairs for ESG rating agencies that have rated company i in year t, with a maximum value of 15.

3.2.3 Moderator variable

  • (1)

    Distraction of institutional investors

To construct variables measuring institutional investors’ distraction, following the approach (Kempf et al., 2017). Institutional investors would be more interested in positive news than negative news, since they face pressure from performance evaluations and their investors. So, when constructing the variable “Distractioni,q”, it only includes the distraction caused by the industries, which the firm i is excluded from, turn out to be the highest return.

(7)Distractioni,q=vFq1INDINDfwv,i,q1×wv,q1IND×ISqIND

“v” represents the investor, whereas “i” represents the specific firm. The “q” in the equation means the calendar quarter q. Fq1 means, in the end of quarter q−1, the column of institutional investors who invest in firm i. IND means the industries which belong to one of the 12 Fama-French industry and INDf is the industry which firm i belongs to.

wv,q1IND denotes the market-value weight of industry IND in investor v’s investment portfolio.

ISqIND is an indicator variable, which equals to 1 if, in a given quarter, the return of industry IND is the highest or lowest among all 12 Fama-French industries.

wv,i,q1 represents the weight of firm i in investor v's investment portfolio, taking into account both the market-value weight of shares held in investor v's portfolio for firm i and the ownership proportion of the total number of shares issued by the firm i. It cannot be simply based on market-value weight because an investor might hold a significant proportion of firm i’s all issued shares, but the overall market value of firm i in their portfolio may be relatively small.

(8)wv,i,q1=QPFweightv,i,q1+QPercOwnv,i,q1vFq1(QPFweightv,i,q1+QPercOwnv,i,q1)

PercOwnv,i,q1 is the proportion of shares of firm i owned by investor v, and PFweightv,i,q1 is the market value weight of firm f in investor i’s portfolio. The stocks held by investor i in quarter q−1 are sorted into quintiles, labeled QPFweightv,i,q1, according to their PFweightv,i,q1. The shareholders of firm f are sorted into quintiles, denoted as QPercOwnv,i,q1, based on their PercOwnv,i,q1.

  • (2)

    Digital economic development level of the province

Following the approach of Liu and Wang (2023), use R-studio to capture the number of occurrences of terms related to the digital economy, such as “digitization” and “digital economy”, in Baidu Index for each province from 2010 to 2022. Summing up the occurrences of these terms by province and year and dividing the sum by one million, a new variable – province level of digital economic development “Digital Economy” was created.

(9)DigitalEconomyprov,t=n=1N=365Termprov,t1,000,000

DigitalEconomyprov,t are the total occurrences of terms for one of the provinces in year t. The formula n=1N=365Termprov,t means to sum up the daily keyword search frequencies collected for year t of the specific province. This yields the total keyword search frequency for t year of the specific province. Dividing by 1,000,000 is to prevent the coefficient of this variable “Digital Economy” from becoming too small.

  • (3)

    Corporate intelligence maturity of the firm

The variable CIM (corporate intelligence maturity) is calculated by identifying intangible assets with names containing terms related to corporate intelligence (Leliaert et al., 2003; Rodov and Leliaert, 2002), sum the values of these intangible assets and divide it by total intangible assets and take the natural logarithm.

(10)CIMi,t=ln[Digitalintangibleassetsi,tTotalintangibleassetsi,t×100]

CIMi,t is the variable which measures the degree of corporate intelligence maturity of firm i in year t. Digitalintangibleassetsi,t represents the value of intangible assets which has been digitalized of firm i in year t, whereas Totalintangibleassetsi,t means the value of total intangible assets of firm i in year t.

3.2.4 Control variables

The control variables are selected following the previous studies (Chen et al., 2022). INST is the ownership percentage of institutional investors among all investors in the company. Mshare is the ownership percentage of the management among all investors. Opinion takes a value of 1 if the company’s financial report for the current year receives an unqualified audit opinion; otherwise, it takes a value of 0. BM ratio is the book-to-market ratio of the firm. Balance means equity balance, calculated as the sum of the ownership percentages of the second to fifth largest shareholders divided by the ownership percentage of the largest shareholder. Size of the company is measured using assets, and the assets are taken in the natural logarithm. ROS is the return of the stock of firm i in year t, and it is calculated following Bei et al. (2011). RV is the logarithm of the annualized monthly return standard deviation of the stock and it is calculated following Su (2015). The reason to include these two variables into the model is because past return can forecast future stock risk and firms with more volatility are likely to experience more crashes (Chen et al., 2001). The equation to calculate RV is as follows.

(11)RVi,j,t=ln[1Tt=1T(ri,j,t1Tt=1Tri,j,t)2]
ri,j,t represents the return of firm i on the j-th month within year t, and T represents the total number of months in each fiscal year.

The definitions of all variables are listed in Table 1.

Table 2 provides an overview of the key variables utilized in this study through summary statistics.

The mean value of NCSKEW and DUVOL is below zero which means, on average, firms would not face the stock price crash risk. The average ERD is 2.4567, which means the average ESG rating disagreement for a company is 2.4567. The average of distraction is 0.0027, which means on average, a firm faces the situation which its intuitional investors are distracted by the positive extreme return by other industries with the degree of 0.0027. The average value of variable digital economy is 0.0635, indicating that the average level of digital economic development in the province where the enterprise is located is 0.0635. The average value of CIM is 11.7689, indicating that the average maturity level of internal intelligence within the enterprise is 11.7689. An average firm also has 48.1992 for INST, 13.4478 for Mshare, 0.9827 for opinion, BM ratio for 0.6031, balance for 0.7507 and 22.4916 for size, 0.189 for ROS and −2.169 for RV.

4. Research design and main results

4.1 Research design

In the main regression analysis, to test hypothesis H1, the following model is constructed, and also used the fixed effects model to analyze their panel data (Jiang et al., 2023).

(12)NCSKEWi,t=α0+α1ERDi,t+α2SIZEi,t+α3INSTi,t+α4Msharei,t+α5Opinioni,t+α6BM ratioi,t+α7Balancei,t+α8ROSi,t+α9RVi,t+α10NCSKEWi,t1+YearFE+FirmFE+εi,t
(13)DUVOLi,t=β0+β1ERDi,t+β2SIZEi,t+β3INSTi,t+β4Msharei,t+β5Opinioni,t+β6BM ratioi,t+β7Balancei,t+β8ROSi,t+β9RVi,t+β10DUVOLi,t1+YearFE+FirmFE+εi,t

NCSKEWi,t and DUVOLi,t are the dependent variables which are used to measure the stock price crash risk. ERD is the independent variable defined earlier in the text. NCSKEWi,t1 and DUVOLi,t1 are used to capture the potential persistence of the third moment of stock returns following Chen et al. (2001). YearFE represents the year fixed effects. FirmFE means firm fixed effects. The double fixed effects model with both time and firm fixed effects provides comprehensive control when dealing with panel data (Jiang et al., 2023).

Then, the moderating effects of institutional investors’ distraction (H2), digital economic development level of the province (H3) and corporate intelligence maturity of the firm (H4) would be tested. The following models are constructed. Still using the panel dataset, the approach follows Li, who also use moderating effect in the research (Li et al., 2023a).

(12)NCSKEWi,t=α0+α1ERDi,t+α2Mi,t+α3ERDi,t×Mi,t+α4SIZEi,t+α5INSTi,t+α6Msharei,t+α7Opinioni,t+α8BM ratioi,t+α9Balancei,t+α10ROSi,t+α11RVi,t+α12NCSKEWi,t1+YearFE+FirmFE+εi,t
(13)DUVOLi,t=β0+β1ERDi,t+β2Mi,t+β3ERDi,t×Mi,t+β4SIZEi,t+β5INSTi,t+β6Msharei,t+β7Opinioni,t+β8BM ratioi,t+β9Balancei,t+β10ROSi,t+β11RVi,t+β12DUVOLi,t1+YearFE+FirmFE+εi,t
Mi,t is the variable which represents one of the three moderating variables. When conducting the following empirical research, Mi,t should be replaced by distraction, digital economy and CIM. ERDi,t×Mi,t is the interaction term between ESG rating disagreement and one of the three moderating variables. If the coefficients of these interaction terms (α3 and β3) are significantly negative, it can validate hypothesis H2, H3 and H4.

All standard errors in the regression results of the paper are robust standard errors (Robust SE).

4.2 Main result

The main result represents the results of H1, which is the primary regression of this paper. In Table 3, column (1) and (2) are the regression results which do not include control variables. Column (3) and (4) are the ones including control variables. ERD’s coefficient is significant at the 0.05 level of significance. With the inclusion of control variables, the ERD’s coefficients are 0.0256 and 0.0163 for regression (3) and (4), respectively. This means 2.85 and 2.70% increase in stock price crash risk, when it is measured by NCSKEW and DUVOL, respectively, associated with a one standard deviation increase in ERD. With the addition of control variables, the model’s explanatory power for stock price crash risk has slightly increased. All four of the ERD’s coefficients are significant, which suggests the positive effects of ESG rating disagreement on stock price crash risk. The bigger the ESG rating disagreement is, the higher the stock price crash risk would be. Therefore, it can be concluded that hypothesis H1 holds true.

4.3 Mechanism analysis

To further examine what could decrease the effect of ESG rating disagreement on stock price crash risk, the moderating role of the three moderating variables should be tested.

4.3.1 Distraction of institutional investors

In Table 4, the coefficients of the interaction term in both regressions are significant at the 1% level of significance, which are −3.6741 and −2.2113, respectively, for NCSKEW and DUVOL. This indicates one standard deviation increase of the interaction term would lead to −7.82% and −7.04% decrease in NCSKEW and DUVOL, respectively. Both ESG rating disagreement and distraction of institutional investors having positive coefficients indicate that they individually have a positive impact on stock price crash risk. However, negative coefficients of the interaction term suggest that the combined effect of ESG rating disagreement and distraction of institutional investors could mitigate stock price crash risk, which confirms the hypothesis H2.

4.3.2 Digital economic development level of the province

The following regression involves using digital economic development level of the province as the moderating variable. The moderating variable “Digital Economy” measures the annual digital economic development level of the province which the firm is located in. For businesses, being located in regions with a high level of development in the digital economy is more advantageous for utilizing these digital tools for corporate governance and risk management.

In Table 5, the coefficients of the interaction term are statistically significant negative, which indicates that one standard deviation increase of the interaction term would lead to −11.72% and −10.21% decrease in NCSKEW and DUVOL, respectively. The combined effect of digital economy and ESG rating disagreement mitigates stock price crash risk. This regression result confirms that hypothesis H3 is valid.

4.3.3 Corporate intelligence maturity

Enterprises with high levels of CIM can utilize digital technology to better identify, assess and respond to risks. This diminishes the risk of stock price crash associated with ESG rating disagreement. The regression analysis here is to analyze the moderating effect of the last moderating variable, that is, CIM.

The result in Table 6 show similar conclusion as the main result that ESG rating disagreement has a positive impact on stock price crash risk, with the coefficient of 0.0806 for NCSKEW and 0.047 for DUVOL, respectively. The combined effect of digital economy and ESG rating disagreement on stock price crash risk is negative, with the coefficients (−0.0046 and −0.0026) of interaction term significant at the 5% level of significance. This shows that one standard deviation increase of the interaction term would lead to −12.00% and −9.97% decrease in NCSKEW and DUVOL, respectively. Based on the analysis above, hypothesis H4 is assumed to be valid.

4.4 Robustness tests

A series of robustness tests was conducted to ensure the robustness of main findings under alternative settings.

First, replace stock price crash risk with profit volatility as the dependent variable. The lower the profit volatility, the lower the stock price volatility, and the smaller the stock price crash risk (Gang and Qi, 2021). Also, profit volatility could measure corporate resilience to risk (Zhu and Xu, 2022). Therefore, using the variable PV, which can measure financial and market risks of the company, as the dependent variable, replaces NCSKEW and DUVOL.

Profit volatility is defined as the three-year moving standard deviation of each company’s return on assets (ROA) adjusted by subtracting its annual ROA from the industry average ROA (Zhu and Xu, 2022). ROA for company i is the ratio of earnings before interest, taxes, depreciation and amortization (EBITDA) to total assets at the end of the corresponding year.

In Table 7, column (1) represents a regression without control variables, while the second column represents a regression with control variables. This positive and statistically significant impact of ESG rating disagreement on profit volatility indicates that ESG rating disagreement could positively affect financial and market risks of the company, which is consistent with the main result.

Second, the robustness tests below primarily involve altering the time span. In order to avoid interference from macroeconomic changes on the main conclusions of this paper, I adopted the robustness test approach similar to the approach used in the study (Shangkun et al., 2012). The selected three-time intervals have different start and end years, as well as varying time spans. The number of observations is relatively close across the three selected subsamples, which can help avoid some issues caused by the difference of the number of sample observations. By selecting three-time intervals, namely 2010–2017, 2013–2019 and 2015–2020, the aim is to examine whether the main conclusions are still robust. That is, whether the positive and significant impact of ESG rating disagreement on stock price crash risk remains unchanged despite variations in the time span.

The regressions still maintain firm fixed effects and year fixed effects in Table 8. When I explored the impact of ERD on NCSKEW, the coefficient of ERD is around 0.03, and when investigating the impact on DUVOL, the coefficient of ERD is around 0.02. In the following regressions, although the coefficient of ERD has changed, the impact of ESG rating disagreement on stock price crash risk is statistically significant and positive. Therefore, this indicates that the main result of this study is not affected by changes in the time span and remains robust.

Third, I altered the ESG rating agencies involved since they may assign different ESG scores to the same company, it can affect the calculation results of ESG rating disagreement. Because Huazheng and WIND started evaluating the company’s ESG rating in 2018, to avoid impacting the research results, the analysis here covers a period from 2018 to 2022, totaling five years. In Table 9, the ESG rating disagreement data used in the regressions in columns (1) and (2) is calculated using data from Bloomberg, Huazheng and WIND, in columns (3) and (4) is calculated using data from WIND, Huazheng and SynTao Green Finance, in column (5) and (6) is calculated using data from Bloomberg, Huazheng, SynTao Green Finance and WIND.

From the regression result, we could see that all the coefficients of ERD are positive, and they are all significant at the 5% significance level. This means the differences in rating data from various rating agencies do not affect the main conclusions of this paper. The impact of ESG rating divergence on stock price crash risk remains positive.

Last, this paper examines the impact of current ESG rating disagreement on stock price crash risk, while the research of Dong et al. (2024) focuses on the impact of lagged one-period ESG rating disagreement on stock price crash risk. Despite using different rating agencies to calculate ERD compared to Dong et al. (2024), this paper aims to explore whether the conclusion of Dong et al. (2024) still holds in this context. Therefore, the ERD is lagged by one period to investigate its impact on NCSKEW and DUVOL.

In Table 10, after lagging the ESG rating divergence by one period, we can see that the impact of ESG rating disagreement on stock price crash risk becomes negative, which is consistent with conclusion of Dong et al. (2024). This means one standard deviation increase of lagged ERD would lead to −2.37% and −2.93% decrease in NCSKEW and DUVOL, respectively. This indicates that current ESG rating disagreements indeed affect investors' confidence in the company, leading to overreactions to changes and disclosures, which in turn increases the risk of a stock price crash. This is because high information opacity in the current period causes investors to have insufficient understanding of the company’s information. However, after a lag period, investors would have gained a deeper understanding of the company, and their confidence in the company would have returned to normal levels. Moreover, because of the disagreement, investors will seek a more in-depth and comprehensive understanding of the company’s information, resulting in fewer sell-offs and short-term selling behaviors, thereby reducing the risk of stock price crashes of the companies. Therefore, the coefficient for the lagged one-period ERD is negative.

4.5 Endogeneity test

Although the main finding indicates that disagreement in ESG ratings can increase the risk of stock price crash, the results may be influenced by endogeneity issues. To address these concerns, three approaches are utilized: incorporating industry fixed effects, introducing new control variables and the two-stage least squares method.

First, I used industry fixed effect instead of firm fixed effect since considering some of the omitted variables that vary over industries but not over time. In column (1) and (2), the regression model includes both firm fixed effects and year fixed effects. For column (3) and column (4), the regression model includes both industry fixed effects and year fixed effects.

In Table 11, it could be concluded that even though the significance of ERD has decreased from 1% level of significance to 5% level of significance, there are still positive and statistically significant coefficients (0.0153 for NCSKEW as dependent variable and 0.0099 for DUVOL as dependent variable) for ERD after changing firm fixed effects with industry fixed effects. After using industry fixed effects, the R-square also decrease and this means the explanation power of the model has decreased. But these results could support the main result that ESG rating disagreement would increase stock price crash risk.

Second, the new control variables are added to addressing endogeneity issues. The explanations for the new variables are as follows. Board is the natural logarithm of the number of board members, representing the size of the board of directors. TMTPay refers to the natural logarithm of the total management compensation. Employee represents the natural logarithm of the number of employees. Indep represents the proportion of independent directors. SOE is a dummy variable, where SOEs are represented as 1, and non-SOEs are represented as 0. Bank is a dummy variable indicating ownership of bank stocks, where holding bank stocks is represented as 1, and not holding bank stocks is represented as 0. The following regression incorporates the new control variables along with the existing control variables. In column (1) and (2), there is only the group of original control variables. In column (3) and (4), the control variables include both original and new control variables.

In Table 12, after adding control variables, the coefficients of ERD are 0.026 for column (3) and 0.0163 for column (4) and they are significant at the 1% level of significance. ESG rating disagreement remains statistically significant. ESG rating disagreement still plays an increasing role in stock price crash risk. After adding new control variables, the t-value of ERD increased slightly, and the model’s explanatory power for NCSKEW and DUVOL also showed a modest improvement.

Last, the two-stage least squares regression is conducted. This study employs lagged ERD as the instrumental variable. The first column presents the first-stage regression results, and the second and third column shows the regression results for the second stage. In Table 13, in the first stage, the coefficient of lagged ERD is 0.0282 and it is significant at the 1% level of significance. This means the lagged ERD has negative effect on the independent variable. In the second stage, the coefficients of ERD of both column (2) and column (3) are positive and significant at the 5% level. This conclusion is consistent with the main regression results.

5. Additional tests

5.1 Industry heterogeneity

The statistical significance and extent of the impact of ESG rating disagreement on stock price crash risk may vary between heavily-polluted firms and lowly-polluted firms. To validate H5, firms in the sample were classified into heavily-polluted firms and lowly-polluted firms following the approach of Li et al. (2021).

From the result in Table 14, it is easily noticed that ESG rating disagreement is positive and statistically significant for heavily-polluted firms, which means one standard deviation increase of ERD would lead to 3.80 and 3.78% increase in NCSKEW and DUVOL, respectively. For heavily-polluted firms, ESG issues may have a greater impact on the operations and reputation of companies, but for non-heavily-polluted firms, ESG issues may be relatively less significant, resulting in ESG rating disagreement having a less significant impact on the stock price crash risk compared to the heavily-polluted firms. The conclusion is consistent with hypothesis H5.

5.2 Ownership heterogeneity

SOEs are generally perceived as being better positioned to assume social and environmental responsibilities, disagreement in ESG ratings for SOEs are likely to draw more scrutiny (Gu et al., 2019). The following regressions are conducted to test hypothesis H6.

The result in Table 15 validates hypothesis H6. For state-owned firms, ESG rating disagreement has the positive impact on the risk of stock price crash risk, and the effect is statistically significant at the 5% significance level, with the coefficients valuing at 0.0368 and 0.0249, respectively. This indicates that one standard deviation increase of ERD would lead to 4.09 and 4.13% increase in NCSKEW and DUVOL, respectively, in SOEs. ESG rating disagreement exerts a statistically significantly positive effect on the stock price crash risk for state-owned firms.

5.3 Coverage of star analysts

Star analysts are considered to make more accurate predictions and can bring greater returns to clients (Loh and Mian, 2006; Luo et al., 2020). The star analysts mentioned in this paper refer to those selected by the New Fortune magazine as star analysts.

Based on whether there are star analysts among the analysts of the company, this paper divides the sample companies into star analyst group and non-star analyst group. In Table 16, for the star analyst groups, the impact of ESG rating disagreement on the risk of stock price crash is positive and significant at the 5% level, which indicates that one standard deviation increase of ERD would lead to 3.46 and 2.87% increase in NCSKEW and DUVOL, respectively, in companies with star analysts. The regression results confirm hypothesis H7. For analysts with stronger information gathering and utilization abilities, ESG rating disagreement would create greater confusion.

6. Conclusion

This paper primarily explores the issue of whether ESG rating disagreement leads to the stock price crash risk and whether distraction of institutional investors, digital economic development level of the province and corporate intelligence maturity can play the moderating roles in this regard. The greater the ESG rating disagreement, the higher the opacity of the company. Investors are more likely to be confused by these divergent ESG signals and their assessment of the company’s investment value may also be prone to changes, thereby increasing the stock price crash risk.

The conclusions drawn from the empirical research in this paper can be summarized as follows. First, ESG rating disagreement could increase the stock price crash risk. Second, the distraction of institutional investors, digital economic development level of the province and CIM would mitigate stock price crash risk by playing the moderating roles. Third, for heavily-polluted firms, state-owned firms and firms with star analysts, the impact of ESG rating disagreement on stock price crash risk is more statistically significant.

The finding provides the novel prospective that investors can reduce investment risks by appropriately reducing their focus on companies with confusing ESG information. Additionally, investors can choose to invest in companies with high levels of intelligence maturity and located in regions with advanced digital economic development.

The inadequacy of this paper lies in only measuring the distraction of investors through institutional investors, without considering whether the distraction of other investors. Further research will consider the moderating effects of the entire investor body and collect data in this regard. Moreover, there may be moderating effects of other variables which could have an impact on this process. Therefore, further research would include whether government intervention and geopolitics will also have moderating effects.

Definition of variables

VariableDefinition
NCSKEWA variable which measures the stock price crash risk
DUVOLAnother variable which measures the stock price crash risk
ERDESG rating disagreement for the specific firm on that year
DistractionThe degree of institutional investors' distraction towards the firm
Digital economyThe variable measures the annual digital economic development level of the province which the firm is located in
CIMThe variable measures corporate intelligence maturity of the firm
INSTThe ownership percentage of institutional investors among all investors which multiplied by 1,000
MshareThe ownership percentage of the management among all investors which multiplied by 1,000
OpinionAn indicator variable, taking a value of 1 if the company’s financial report for the current year receives an unqualified audit opinion
BM ratioThe ratio of book value to market value of the firm
BalanceEquity balance is calculated as the sum of the ownership percentages of the second to fifth largest shareholders divided by the ownership percentage of the largest shareholder
SizeNatural logarithm of total assets
ROSReturn of the stock of firm i in year t
RVNatural logarithm of the annualized monthly return standard deviation of the stock

Source(s): This table is created by the author

Descriptive statistics

VariableObsMeanStd. Dev.MinMax
NCSKEW17,066−0.2770.7275−5.17075.0382
DUVOL17,066−0.18780.4864−2.56152.2777
ERD17,0662.45670.80905.8341
Distraction17,0660.00270.006100.0412
Digital economy17,0660.06350.05130.00020.1916
CIM17,06611.76896.6263023.2072
INST17,06648.199225.29090.0014157.0978
Mshare17,06613.447819.4083089.1771
Opinion17,0660.98270.130401
BM ratio17,0660.60310.25730.00981.5592
Balance17,0660.75070.62230.00564
Size17,06622.49161.349217.813228.6365
ROS19,4830.1890.588−0.82214.28
RV19,483−2.1690.458−3.9730.661

Source(s): This table is created by the author

Main result

(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0176**0.0122**0.0256***0.0163***
(0.0087)(0.0058)(0.0089)(0.0059)
Lagged NCSKEW or Lagged DUVOL −0.0907***−0.0967***
(0.0083)(0.0082)
INST 0.0045***0.0035***
(0.0006)(0.0004)
Mshare 0.0056***0.0041***
(0.001)(0.0007)
Opinion −0.1844***−0.1223***
(0.0475)(0.0315)
BM_ratio −0.2215***−0.0629*
(0.0521)(0.0345)
Balance 0.0444**0.0288**
(0.0213)(0.0141)
size −0.0205−0.0452***
(0.0174)(0.0115)
ROS −0.1332***−0.1012***
(0.0134)(0.0089)
RV −0.2752***−0.1783***
(0.0188)(0.0124)
_cons−0.3192***−0.2176***−0.48510.3207
(0.0218)(0.0146)(0.3906)(0.259)
Observations17,06617,06616,78216,782
R-squared0.22970.23040.27570.2806

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Mechanism analysis - distraction

(1)(2)
NCSKEWDUVOL
ERD0.0354***0.0222***
(0.0094)(0.0063)
Distraction11.248***6.8684***
(2.887)(1.9142)
ERD × Distraction−3.6741***−2.2113***
(1.155)(0.7659)
Control variablesYESYES
_cons−0.51180.3051
(0.3905)(0.2589)
Firm FEsYESYES
Year FEsYESYES
Observations16,78216,782
R-squared0.27660.2813

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Mechanism analysis – digital economy

(1)(2)
NCSKEWDUVOL
ERD0.0599***0.0374***
(0.013)(0.0086)
Digital economy−0.0601−0.083
(0.6154)(0.4081)
ERD × Digital economy−0.6246***−0.3637***
(0.1645)(0.1091)
Control variablesYESYES
_cons−0.66170.2009
(0.4175)(0.2769)
Firm FEsYESYES
Year FEsYESYES
Observations16,78216,782
R-squared0.27750.2818

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Mechanism analysis – CIM

(1)(2)
NCSKEWDUVOL
ERD0.0806***0.047***
(0.0157)(0.0104)
CIM0.0117***0.0055***
(0.003)(0.002)
ERD × CIM−0.0046***−0.0026***
(0.0011)(0.0007)
Control variablesYESYES
_cons−0.53510.2709
(0.3946)(0.2619)
Firm FEsYESYES
Year FEsYESYES
Observations16,78216,782
R-squared0.27790.2823

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Robustness test – PV

(1)(2)
PVPV
ERD0.0012***0.0008**
(0.0004)(0.0004)
Control variablesNoYES
_cons0.0303***−0.0242
(0.0011)(0.0186)
Firm FEsYESYES
Year FEsYESYES
R-squared0.50710.5504

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Robustness test – altering the time intervals

2010–20172013–20192015–2020
(1)(2)(3)(4)(5)(6)
NCSKEWDUVOLNCSKEWDUVOLNCSKEWDUVOL
ERD0.031***0.0179**0.0313***0.0185**0.0444***0.0271***
(0.0111)(0.0075)(0.0113)(0.0075)(0.013)(0.0086)
Control variablesYESYESYESYESYESYES
_cons−0.12250.793**−0.86340.6513−2.2168***−0.7634
(0.5912)(0.4016)(0.6238)(0.4145)(0.7721)(0.5106)
Firm FEsYESYESYESYESYESYES
Year FEsYESYESYESYESYESYES
R-squared0.3480.35450.33190.33490.35720.3581

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Robustness test – altering the control variables

Bloomberg, Huazheng and WINDWIND, Huazheng and SynTaoBloomberg, Huazheng, SynTao and WIND
(1)(2)(3)(4)(5)(6)
NCSKEWDUVOLNCSKEWDUVOLNCSKEWDUVOL
ERD0.108**0.0642**0.0675**0.0481**0.1104***0.0633**
(0.0483)(0.0327)(0.0337)(0.0227)(0.0392)(0.0264)
Control variablesYESYESYESYESYESYES
_cons2.66360.7638−1.7446−0.9132−1.8936−0.9652
(2.8117)(1.905)(1.7581)(1.1826)(1.7577)(1.1831)
Firm FEsYESYESYESYESYESYES
Year FEsYESYESYESYESYESYES
R-squared0.30170.31720.22230.20500.22390.2055

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Robustness test – lagged ERD

(1)(2)
NCSKEWDUVOL
Lagged ERD−0.0209**−0.0173**
(0.0104)(0.007)
Control variablesYESYES
_cons−1.5739**−0.6772
(0.6408)(0.4337)
Firm FEsYESYES
Year FEsYESYES
R-squared0.2970.3036

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Endogeneity test – industry fixed effect

(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0256***0.0163***0.0153**0.0099**
(0.0089)(0.0059)(0.0076)(0.005)
Control variablesYESYESYESYES
_cons−0.48510.3207−0.2741**0.3229***
(0.3906)(0.259)(0.1353)(0.1009)
Firm FEsYESYESNONO
Industry FEsNONOYESYES
Year FEsYESYESYESYES
Observations16,78216,78216,78216,782
R-squared0.27570.28060.08990.1041

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Endogeneity test – altering the control variables

(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0256***0.0163***0.026***0.0163***
(0.0089)(0.0059)(0.009)(0.006)
Original controlsYESYESYESYES
Added new controlsNoNoYESYES
_cons−0.48510.3207−0.39930.1891
(0.3906)(0.259)(0.4696)(0.3108)
Firm FEsYESYESYESYES
Year FEsYESYESYESYES
Observations16,78216,78216,78216,782
R-squared0.27570.28060.27820.2832

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Endogeneity test – two-stage least squares

(1)(2)(3)
ERDNCSKEWDUVOL
lagged_ERD0.0282***
(0.0099)
ERD 0.3366**0.2328**
(0.1422)(0.0963)
Control variablesYESYESYES
_cons3.6005***−3.8766**−2.3059*
(0.5016)(1.7437)(1.1833)
Firm FEsYESYESYES
Year FEsYESYESYES
R-squared0.50990.00920.0147

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Additional tests – Industry heterogeneity

Heavily-pollutedLowly-polluted
(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0342**0.0227**0.0203*0.0123*
(0.0152)(0.0101)(0.0111)(0.0073)
Control variablesYESYESYESYES
_cons−1.8099**−0.7186−0.2210.6068*
(0.7838)(0.5231)(0.4826)(0.3192)
Firm FEsYESYESYESYES
Year FEsYESYESYESYES
R-squared0.26780.27250.2890.2929

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Additional tests – ownership heterogeneity

State-ownedPrivately-owned
(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0368**0.0249***0.01880.0106
(0.0145)(0.0094)(0.0115)(0.0077)
Control variablesYESYESYESYES
_cons0.87591.2187***−2.0818***−0.8533**
(0.6839)(0.4457)(0.5197)(0.348)
Firm FEsYESYESYESYES
Year FEsYESYESYESYES
R-squared0.27030.28240.29510.2949

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

Additional tests – coverage of star analysts

Star analystsNon-star analysts
(1)(2)(3)(4)
NCSKEWDUVOLNCSKEWDUVOL
ERD0.0311***0.0173**0.00620.009
(0.0104)(0.007)(0.0233)(0.0152)
Control variablesYESYESYESYES
_cons−0.9355**0.08920.35790.6883
(0.4688)(0.3137)(1.0181)(0.6636)
Firm FEsYESYESYESYES
Year FEsYESYESYESYES
R-squared0.31390.31660.44330.447

Note(s): Standard errors are in parentheses

***p < 0.01, **p < 0.05, *p < 0.1

Source(s): This table is created by the author

References

Avramov, D., Cheng, S., Lioui, A. and Tarelli, A. (2022), “Sustainable investing with ESG rating uncertainty”, Journal of Financial Economics, Vol. 145 No. 2, pp. 642-664, doi: 10.1016/j.jfineco.2021.09.009.

Bei, Z., Wei, W. and Wang, S. (2011), “Effects of stock investment on property income growth”, Technoeconomics and Management Research, Vol. 04, pp. 18-21.

Billio, M., Costola, M., Hristova, I., Latino, C. and Pelizzon, L. (2021), “Inside the ESG ratings: (Dis) agreement and performance”, Corporate Social Responsibility and Environmental Management, Vol. 28 No. 5, pp. 1426-1445, doi: 10.1002/csr.2177.

Cen, T. (2023), “Green finance reform and stock price crash risk: evidence from Chinese heavily polluting companies”, Finance Research Letters, Vol. 56 C, 104133, doi: 10.1016/j.frl.2023.104133, available at: https://ideas.repec.org//a/eee/finlet/v56y2023ics1544612323005056.html

Chen, H., Chen, J.Z., Lobo, G.J. and Wang, Y. (2011), “Effects of audit quality on earnings management and cost of equity capital: evidence from China”, Contemporary Accounting Research, Vol. 28 No. 3, pp. 892-925, doi: 10.1111/j.1911-3846.2011.01088.x.

Chen, J., Harrison, H. and Jeremy, C. (2001), “Forecasting crashes: trading volume, past returns and conditional skewness in stock prices”, Journal of Financial Economics, Vol. 61, pp. 345-391.

Chen, G., Han, J. and Yuan, H. (2022), “Urban digital economy development, enterprise innovation, and ESG performance in China”, Frontiers in Environmental Science, Vol. 10, 955055, doi: 10.3389/fenvs.2022.955055.

Chen, Y., Li, T., Zeng, Q. and Zhu, B. (2023), “Effect of ESG performance on the cost of equity capital: evidence from China”, International Review of Economics and Finance, Vol. 83 C, pp. 348-364, doi: 10.1016/j.iref.2022.09.001.

Chen, X., Safdar Sial, M., Tran, D.K., Alhaddad, W., Hwang, J. and Thu, P.A. (2020), “Are socially responsible companies really ethical? The moderating role of state-owned enterprises: evidence from China”, Sustainability, Vol. 12 No. 7, p. 2858, doi: 10.3390/su12072858.

Christensen, D.M., Serafeim, G. and Sikochi, A. (2022), “Why is corporate virtue in the eye of the beholder? The case of ESG ratings”, The Accounting Review, Vol. 97 No. 1, pp. 147-175, doi: 10.2308/tar-2019-0506.

Custom Market Insights (2023), “Global environmental social and governance (ESG) investing market 2024-2033”, [Consumer Goods], available at: https://www.custommarketinsights.com/report/environmental-social-governance-investing-market/?trk=article-ssr-frontend-pulse_little-text-block

Desai, H., Liang, B. and Singh, A.K. (2000), “Do all-stars shine? Evaluation of analyst recommendations”, Financial Analysts Journal, Vol. 56 No. 3, pp. 20-29, doi: 10.2469/faj.v56.n3.2357.

Dong, M., Li, M., Wang, H. and Pang, Y. (2024), “ESG disagreement and stock price crash risk: evidence from China”, Asia-Pacific Financial Markets, pp. 1-33, doi: 10.1007/s10690-024-09453-y.

Fang, J. (2023), “Corporate social responsibility and stock-price crash risk: test of regulation mechanism based on institutional ownership and internal control”, Journal of Industrial and Production Engineering, Vol. 40 No. 8, pp. 624-637, doi: 10.1080/21681015.2023.2253817.

Fu, T. and Li, J. (2023), “An empirical analysis of the impact of ESG on financial performance: the moderating role of digital transformation”, Frontiers in Environmental Science, Vol. 11, doi: 10.3389/fenvs.2023.1256052, available at:

Gang, S.U.N. and Qi, Z. (2021), “High-tech enterprises recognition and stock price crash risk of Chinese listed firms——micro-evidence from implementation of industry policy”, Collected Essays on Finance and Economics, Vol. 273 No. 6, p. 78.

Gao, J., Li, J. and Luo, Y. (2022), “Does the enterprise's choice of intangible capital help to promote its ESG score?”, Frontiers in Environmental Science, Vol. 10, doi: 10.3389/fenvs.2022.978955.

Gibson Brandon, R., Krueger, P. and Schmidt, P.S. (2021), “ESG rating disagreement and stock returns”, Financial Analysts Journal, Vol. 77 No. 4, pp. 104-127, doi: 10.1080/0015198x.2021.1963186.

Gu, Z., Tang, S. and Wu, D. (2019), “The political economy of labor-employment decisions: evidence from China”, (SSRN Scholarly Paper 2786533), doi: 10.2139/ssrn.2786533.

Guan, S.S. (2022), “Meme investors and retail risk”, Boston College Law Review, Vol. 63, p. 2051.

Hsu, B.-X. and Chen, Y.-M. (2023), “The relationship between corporate social responsibility, external orientation, and environmental performance”, Technological Forecasting and Social Change, Vol. 188, 122278, doi: 10.1016/j.techfore.2022.122278.

Hu, X., Hua, R., Liu, Q. and Wang, C. (2023), “The green fog: environmental rating disagreement and corporate greenwashing”, Pacific-Basin Finance Journal, Vol. 78, 101952, doi: 10.1016/j.pacfin.2023.101952.

Huang, W., Luo, Y., Wang, X. and Xiao, L. (2022), “Controlling shareholder pledging and corporate ESG behavior”, Research in International Business and Finance, Vol. 61, 101655, doi: 10.1016/j.ribaf.2022.101655.

Hutton, A.P., Marcus, A.J. and Tehranian, H. (2009), “Opaque financial reports, R2, and crash risk”, Journal of Financial Economics, Vol. 94 No. 1, pp. 67-86, doi: 10.1016/j.jfineco.2008.10.003.

Jiang, D., Peng, X. and Zhu, H. (2023), “Shared analyst coverage and connected-firm momentum spillover in China”, Systems Engineering Theory and Practice, Vol. 43 No. 7, pp. 1891-1909.

Jin, L. and Myers, S.C. (2006), “R2 around the world: new theory and new tests”, Journal of Financial Economics, Vol. 79 No. 2, pp. 257-292, doi: 10.1016/j.jfineco.2004.11.003.

Jin, X. and Wu, Y. (2023), “The impact of digital investment on corporate ESG performance——regulatory effect based on the market environment”, Digitalization and Management Innovation II: Proceedings of DMI 2023, Vol. 376, p. 400.

Junyan, Z., Yucheng, J. and Cheng, Z. (2023), “Digital economy of the host country and location choice of China's OFDI: a study based on the data of listed companies”, Journal of Shanghai University of Finance and Economics, Vol. 25 No. 04, doi: 10.16538/j.cnki.jsufe.2023.04.004.

Kempf, E., Manconi, A. and Spalt, O. (2017), “Distracted shareholders and corporate actions”, Review of Financial Studies, Vol. 30 No. 5, pp. 1660-1695, doi: 10.1093/rfs/hhw082.

Kim, S.E. (2023), “The duality of variance among ESG assessments”, Missouri Law Review, Vol. 88, p. 409.

Kim, J., Kim, M., Im, S. and Choi, D. (2021), “Competitiveness of E Commerce firms through ESG logistics”, Sustainability, Vol. 13 No. 20, 11548, doi: 10.3390/su132011548.

Kim, Y., Li, H. and Li, S. (2014), “Corporate social responsibility and stock price crash risk”, Journal of Banking and Finance, Vol. 43, pp. 1-13, doi: 10.1016/j.jbankfin.2014.02.013.

Kotsantonis, S., Pinney, C. and Serafeim, G. (2016), “ESG integration in investment management: myths and realities”, The Journal of Applied Corporate Finance, Vol. 28 No. 2, pp. 10-16, doi: 10.1111/jacf.12169.

Leliaert, P.J., Candries, W. and Tilmans, R. (2003), “Identifying and managing IC: a new classification”, Journal of Intellectual Capital, Vol. 4 No. 2, pp. 202-214, doi: 10.1108/14691930310472820.

Li, J., Fu, T., Han, S. and Liang, R. (2023a), “Exploring the impact of corporate social responsibility on financial performance: the moderating role of media attention”, Sustainability, Vol. 15 No. 6, 5023, doi: 10.3390/su15065023.

Li, J., Yang, Z., Chen, J. and Cui, W. (2021), “Study on the mechanism of ESG promoting corporate performance: based on the perspective of corporate innovation”, Science of Science and Management of S. and T., Vol. 42 No. 9, pp. 71-89, doi: CNKI:SUN:KXXG.0.2021-09-005.

Liu, G. and Wang, S. (2023), “Digital transformation and trade credit provision: evidence from China”, Research in International Business and Finance, Vol. 64, 101805, doi: 10.1016/j.ribaf.2022.101805.

Liu, X., Dai, J., Dong, X. and Liu, J. (2024), “ESG rating disagreement and analyst forecast quality”, International Review of Financial Analysis, Vol. 95, 103446, doi: 10.1016/j.irfa.2024.103446.

Loh, R.K. and Mian, G.M. (2006), “Do accurate earnings forecasts facilitate superior investment recommendations?”, Journal of Financial Economics, Vol. 80 No. 2, pp. 455-483, doi: 10.1016/j.jfineco.2005.03.009.

Luo, D., Yan, J. and Yan, Q. (2023), “The duality of ESG: impact of ratings and disagreement on stock crash risk in China”, Finance Research Letters, Vol. 58 PB, 104479, doi: 10.1016/j.frl.2023.104479, available at: https://ideas.repec.org//a/eee/finlet/v58y2023ipbs1544612323008516.html

Luo, J., Yin, H. and Zhang, H. (2020), “The disincentive effect of stars: evidence from analyst coverage”, Journal of Accounting, Auditing and Finance, Vol. 35 No. 4, pp. 803-828, doi: 10.1177/0148558x19832096.

Manita, R., Elommal, N., Baudier, P. and Hikkerova, L. (2020), “The digital transformation of external audit and its impact on corporate governance”, Technological Forecasting and Social Change, Vol. 150, 119751, doi: 10.1016/j.techfore.2019.119751.

Ming-quan, S., Lu, Y.U. and Wen-bing, W. (2023), “ESG disclosure and stock price crash risk”, Journal of Guizhou University of Finance and Economics, Vol. 41 No. 02, p. 32.

Nenavani, J., Prasuna, A., Siva Kumar, S.N.V. and Kasturi, A. (2024), “ESG measures and financial performance of logistics companies”, Letters in Spatial and Resource Sciences, Vol. 17 No. 1, p. 5, doi: 10.1007/s12076-023-00358-4.

Nguyen, H., Calantone, R. and Krishnan, R. (2020), “Influence of social media emotional word of mouth on institutional investors' decisions and firm value”, Management Science, Vol. 66 No. 2, pp. 887-910, doi: 10.1287/mnsc.2018.3226.

Polat, M.F. (2020), “Are individual investors informed traders? Evidence from their mimicking behavior”, (Doctoral dissertation, University of Maryland, College Park).

Ren, Y., Tian, Z. and Liu, C. (2023), “Impact of the digital economy on urban green growth: empirical evidence from China”, PLoS One, Vol. 18 No. 12, e0296072, doi: 10.1371/journal.pone.0296072.

Rodov, I. and Leliaert, P. (2002), “FiMIAM: financial method of intangible assets measurement”, Journal of Intellectual Capital, Vol. 3 No. 3, pp. 323-336, doi: 10.1108/14691930210435642.

Schmidt, D. (2019), “Distracted institutional investors”, Journal of Financial and Quantitative Analysis, Vol. 54 No. 6, pp. 2453-2491, doi: 10.1017/S0022109018001242.

Shangkun, D., Yizhou, S. and Akito, S. (2012), “Robustness test of genetic algorithm on generating rules for currency trading”, Procedia Computer Science, Vol. 13, pp. 86-98, doi: 10.1016/j.procs.2012.09.117.

Shi, S. (2021), “Investor attention, analyst optimism, and stock price crash risk”, Proceedings of Business and Economic Studies, Vol. 4 No. 3, pp. 63-72, doi: 10.26689/pbes.v4i3.2200.

Su, K. (2015), “Management equity incentive, risk taking and efficiency of capital allocation”, Journal of Management Science, Vol. 28 No. 3, pp. 14-25.

Sun, K.P. and Xiao, X. (2018), “Influence mechanism of internet social media on investor sentiment contagion and stock price collapse risk”, Journal of Technology Economics, Vol. 37 No. 06, pp. 93-102.

Tusheng, X., Shunlin, S. and Lu, L. (2017), “Information disclosure quality and stock price crash risk: the mediating role of analyst forecast”, Journal of Finance and Economics, Vol. 43 No. 02, doi: 10.16538/j.cnkij.fe.2017.02.008.

Wang, J., Wang, S., Dong, M. and Wang, H. (2024), “ESG rating disagreement and stock returns: evidence from China”, International Review of Financial Analysis, Vol. 91, 103043, doi: 10.1016/j.irfa.2023.103043.

Wu, W., Zhao, Y. and Su, Z. (2022), “Corporate IT construction and audit fees: new evidence in the era of digitalization”, Auditory Research, Vol. 1, pp. 106-117.

Xu, N., Chan, K.C., Jiang, X. and Yi, Z. (2013), “Do star analysts know more firm-specific information? Evidence from China”, Journal of Banking and Finance, Vol. 37 No. 1, pp. 89-102, doi: 10.1016/j.jbankfin.2012.08.014.

Yan, Z. (2023), “What makes a better company? ESG and ESG ratings”, pp. 92-98, doi: 10.2991/978-94-6463-098-5_12.

Yin, F., Xiao, Y., Cao, R. and Zhang, J. (2023), “Impacts of ESG disclosure on corporate carbon performance: empirical evidence from listed companies in heavy pollution industries”, Sustainability, Vol. 15 No. 21, 15296, doi: 10.3390/su152115296.

Zhao, R., Fan, R., Xiong, X., Wang, J. and Hilliard, J. (2023a), “Media tone and stock price crash risk: evidence from China”, Mathematics, Vol. 11 No. 17, 3675, doi: 10.3390/math11173675.

Zhao, X., Li, X. and Ren, C. (2023b), “Can digital transformation reduce corporate stock price crashes?”, PLoS One, Vol. 18 No. 12, e0295793, doi: 10.1371/journal.pone.0295793.

Zhu, M. and Xu, Y. (2022), “Exchange rate risk hedging and M&As: a free cash flow perspective”, Financial Research, Vol. 510 No. 12, p. 36.

Further reading

Cho, S.J., Chung, C.Y. and Young, J. (2019), “Study on the relationship between CSR and financial performance”, Sustainability, Vol. 11 No. 2, 343, doi: 10.3390/su11020343.

Galant, A. and Cadez, S. (2017), “Corporate social responsibility and financial performance relationship: a review of measurement approaches”, Economic Research-Ekonomska Istraživanja, Vol. 30 No. 1, pp. 676-693, doi: 10.1080/1331677x.2017.1313122.

Li, Z., Li, S., Huo, Z., Liu, Y. and Zhang, H. (2023b), “Does CSR information disclosure improve investment efficiency? The moderating role of analyst attention”, Sustainability, Vol. 15 No. 16, 12310, doi: 10.3390/su151612310.

Liu, X.K., Wright, A.M. and Wu, Y.-J. (2015), “Managers' unethical fraudulent financial reporting: the effect of control strength and control framing”, Journal of Business Ethics, Vol. 129 No. 2, pp. 295-310, doi: 10.1007/s10551-014-2156-1.

Naqvi, S.K., Shahzad, F., Rehman, I.U., Qureshi, F. and Laique, U. (2021), “Corporate social responsibility performance and information asymmetry: the moderating role of analyst coverage”, Corporate Social Responsibility and Environmental Management, Vol. 28 No. 6, pp. 1549-1563, doi: 10.1002/csr.2114.

Ricci, F., Scafarto, V., Ferri, S. and Tron, A. (2020), “Value relevance of digitalization: the moderating role of corporate sustainability. An empirical study of Italian listed companies”, Journal of Cleaner Production, Vol. 276, 123282, doi: 10.1016/j.jclepro.2020.123282.

Ruirui, L. (2019), “An empirical study on inter-temporal interaction between corporate social responsibility and financial performance”, 2019 International Conference on Management, Finance and Social Science Research (MFSSR 2019), pp. 615-619, doi: 10.25236/mfssr.2019.124.

Wang, Z. (2023), “A study of the impact of corporate ESG performance on the cost of debt”, Frontiers of International Accounting, Vol. 12 No. 01, pp. 59-70, doi: 10.12677/FIA.2023.121008.

Zumente, I. and Lāce, N. (2021), “ESG rating—necessity for the investor or the company?”, Sustainability, Vol. 13 No. 16, 8940, doi: 10.3390/su13168940.

Corresponding author

Yiyuan Li can be contacted at: Yiyuan.Li@anu.edu.au

Related articles