Professional Documents
Culture Documents
(外文) ESG的双重性:评级与分歧对股价崩盘风险的影响
(外文) ESG的双重性:评级与分歧对股价崩盘风险的影响
JEL classification: This study scrutinizes the effects of Environmental, Social, and Governance (ESG) and ESG
G12 disagreement on the stock price crash risk among Chinese listed firms. Shifting the focus from
G14 mere ESG ratings, the research emphasizes the effect of ESG disagreement across various rating
G34
agencies. Our results confirm that higher ESG ratings contribute to a diminished likelihood of
Keywords: stock price crash risk, while the salutary impact is diminished by the prevalent ESG disagree-
ESG ment among rating agencies. Additionally, subgroup analyses reveal heterogeneity, with the
ESG disagreement
moderating effects of ESG disagreement notably pronounced in State-Owned Enterprises (SOEs)
Stock crash risk
than the non-SOEs. These findings accentuate the imperative for rating agencies to harmonize
Sustainable investment
their ESG evaluating methodologies and call for normalization in firms’ ESG disclosure practices,
thereby minimizing stock price crash risk and fortifying financial market stability.
1. Introduction
ESG rating signifies a firm’s environmental, social responsibility, and governance performance. Over recent years, the significance
of these ratings in the financial market has intensified. Firms with strong ESG performance tend to attract long-term investments
and exude higher managerial ethics and information transparency (Lashkaripour, 2023; Feng et al., 2022; He et al., 2023). This
strength often correlates with a decreased stock price crash risk (Zhou et al., 2021; Feng et al., 2022; Yu et al., 2023; Kim et al.,
2014; Wu and Hu, 2019).
Yet, the emerging nature of ESG criteria means rating divergences are common, complicating investment decisions and muddying
performance evaluations (Christensen et al., 2022). Some researchers attribute these disagreements as potential contributors to
inconclusive results in ESG-stock return studies (Berg et al., 2021). While the existing literature largely examines individual effects
of ESG performance or rating disagreements, few explore their combined impact on stock crash risks. Our research bridges this gap
by analyzing the interplay of ESG ratings and their disagreement in the Chinese context.
We focus on the case of China for a few reasons. Firstly, we expect that such ESG disagreement might be pronounced in a young
and less-developed stock market due to inadequate market supervision standards and information disclosure guidelines. As shown in
Table 1, the ESG ratings between nine existing ESG rating agencies are highly diverged in the Chinese financial market.1 Secondly,
✩ Jingzhou Yan gratefully acknowledges the financial support from National Natural Science Foundation of China (72371178), Natural Science Foundation
of Sichuan Province (2023NSFSC1032), the Fundamental Research Funds for the Central Universities, Postdoctoral Research Foundation of Sichuan University
(skbsh2202-18).
∗ Corresponding author.
E-mail addresses: [email protected] (D. Luo), [email protected] (J. Yan), [email protected] (Q. Yan).
1 The ratings are standardized to be comparable, as described in Section 3.
https://doi.org/10.1016/j.frl.2023.104479
Received 6 July 2023; Received in revised form 21 August 2023; Accepted 14 September 2023
Available online 19 September 2023
1544-6123/© 2023 Elsevier Inc. All rights reserved.
D. Luo et al. Finance Research Letters 58 (2023) 104479
Table 1
Correlation coefficients of the ratings across different rating agencies.
Agencies Sino SynTao Rankins CASVI Susall WIND Bloomberg MSCI Russell
Sino 1
SynTao 0.295 1
Rankins 0.354 0.587 1
CASVI 0.335 0.296 0.322 1
Susall 0.409 0.501 0.558 0.998 1
WIND 0.360 0.530 0.472 0.321 0.457 1
Bloomberg 0.303 0.564 0.650 0.383 0.663 0.470 1
MSCI 0.183 0.479 0.578 0.220 0.466 0.371 0.606 1
Russell 0.199 0.606 0.637 0.261 0.447 0.439 0.625 0.583 1
Note: This table presents the Pearson correlation coefficients of the standardized ESG ratings assigned to Chinese listed firms
from nine rating agencies. The correlation coefficients are all significant at 99% confidence level.
ESG investment in China is still a relatively new concept, only gaining significant attention in recent years. Most domestic ESG
rating agencies began publishing ESG ratings in 2010, and some even later in 2016. Furthermore, it was not until 2016 that the
Ministry of Finance first encouraged listed firms to disclose ESG performances on a voluntary basis. As suggested by Christensen
et al. (2022), a wide range of different evaluations are likely to occur in emerging fields where evaluation standards and norms are
not fully established. Last but not least, with a preponderance of retail investors in the Chinese stock market, the repercussions of
stock price plunges could be particularly severe in China.
Our study intersects with two distinct areas of literature. The first centers on the relationship between ESG and stock price crash
risk. A growing body of studies examines how a firm’s ESG performance affects its stock return (Pástor et al., 2021; Shanaev and
Ghimire, 2022; Serafeim and Yoon, 2022). As ESG becomes an essential consideration for investors, it is recognized as a significant
factor in asset pricing. Additionally, many findings highlight the impact of a firm’s ESG performance on its financial stability and
risk profile (Capelli et al., 2023; Li et al., 2022; He et al., 2022). Recent studies suggest that strong ESG performance reduces the
likelihood of stock price crashes (Zhou et al., 2021; Feng et al., 2022; Yu et al., 2023; Kim et al., 2014; Wu and Hu, 2019). This
is often attributed to ESG’s connection with better management ethics, stronger stakeholder relationships, and clearer information,
reducing risks for investors and lowering financing costs for firms. While many confirm the benefits of good ESG performance in
reducing crash risks, the effects of ESG disagreements on risk are less explored.
Our work also contributes to discussions on ESG disagreement. An increasing number of studies are delving into the causes
and implications of variations in ESG ratings (Berg et al., 2021; Christensen et al., 2022; Avramov et al., 2022; Luo et al.,
2023). Gibson Brandon et al. (2021) suggest that such discrepancies in ratings have a marked impact on excess returns, particularly
influencing the returns of ‘‘green’’ assets. Additionally, Berg et al. (2021) posit that disagreements in ESG ratings can be perceived
as an added risk by investors, thereby influencing their investment decisions. Such variations also shape the way investors and
managers perceive the credibility of ESG scores. Our study is distinctive in its focus on instances of negative jumps in stock prices,
offering a more profound insight into the related risks.
The remainder of this paper is structured as follows. Section 2 proposes theoretical analysis and hypothesis development.
Section 3 explains the data, variables, and models. Section 4 reports the empirical results. 5 illustrates several robustness tests
that we have performed to validate our findings. Section 6 presents a heterogeneity analysis. Finally, Section 7 summarizes our
findings.
The effects of ESG and ESG disagreement are multifaceted. From a stakeholder theory perspective, a superior ESG rating signifies
the firm’s focus on long-term value creation, investor relations and minimized risk-taking (Knox et al., 2005; Deng et al., 2013), all
of which are associated with improved firm performance and a reduction in stock price crash risk (Choi and Wang, 2009; Callen and
Fang, 2015; Cao et al., 2016). From an information theory perspective, while better ESG performance of a firm is associated with
better information transparency (He et al., 2023), ESG disagreement can compromise the quality and consistency of information
accessible to market participants. It is commonly accepted that information asymmetry serves as a primary catalyst for stock market
crashes (Jin and Myers, 2006). Consequently, this can augment stock crash risk, as investors may encounter difficulties in accurately
discerning a firm’s risk profile.
For managers within a firm, if they perceive rating agencies as having divergent or ambiguous ESG assessment criteria, they
may hardly identify a clear pathway to improving the firm’s ESG performance and lack the incentive to allocate resources toward
enhancing it. For external investors, disagreement signifies the divergence in opinions and evaluations among rating agencies
regarding a firm’s ESG performance. This divergence can undermine investor confidence in a company’s ESG performance and
even diminish trust in the market, consequently increasing the stock price crash risk. As a result, while robust ESG performance can
mitigate stock crash risk, disagreement in ESG ratings may temper its beneficial impact.
Based on the aforementioned inferences, we propose and test the following hypotheses in subsequent sections:
H1: Better ESG performances of the firms reduce their stock crash risks without considering disagreement about ESG ratings.
H2: ESG disagreement across different rating agencies attenuates the decreasing effect of ESG on a firm’s stock crash risk.
2
D. Luo et al. Finance Research Letters 58 (2023) 104479
Independent variables. The independent variables in our model are the mean average ESG score (𝑚𝐸𝑆𝐺) and the ESG disagreement
(𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒) across various rating agencies. Both variables are constructed using ESG ratings for firms listed on the Chinese
A-Share market. ESG ratings are collected from 9 rating agencies. These include 3 international rating agencies: MSCI, Bloomberg
and FTSE Russell, and 6 domestic rating agencies: Sino Security, SynTao Green Finance, CASVI, Wind, Rankins, and Susall Wave.
Some agencies assign continuous numerical ESG ratings from 1 to 10, 1 to 5, or 1 to 100 to the firms (e.g. Bloomberg, Russel, and
Wind, etc.), while others use alphabetic rating scales from AAA to CCC or D (e.g. MSCI, CASVI, and Susall Wave, etc.). To make
these ratings comparable, we standardize all collected ratings to numerical scores from 1 to 10. Then, 𝑚𝐸𝑆𝐺𝑖,𝑡 is given by the mean
of all standardized ESG ratings of firm 𝑖 in year 𝑡.
The ESG rating disagreement index is constructed following Avramov et al. (2022). For the 9 ratings from different agencies, we
can match them into at most 𝐶92 = 36 pairs2 and calculate the pairwise rating uncertainties for each pair in each year by:
√
√( )2 ( )2
√ 𝑖,𝑡 𝑖,𝑡
𝑔 𝑖,𝑡 +𝑔 𝑖,𝑡
√ 𝑖,𝑡 𝑔𝑎 +𝑔𝑏 | 𝑖,𝑡 |
√ 𝑔𝑎 − 2 + 𝑔𝑏𝑖 − 𝑎 2 𝑏 |𝑔𝑎 − 𝑔𝑏𝑖,𝑡 |
√
𝑖,𝑡
𝑃 𝑈𝑎,𝑏 = = | √ |, (1)
2−1 2
𝑖,𝑡
where 𝑃 𝑈𝑎,𝑏 is the pairwise uncertainty between rating agency 𝑎 and 𝑏 of stock 𝑖 in year 𝑡, 𝑔𝑎𝑖,𝑡 and 𝑔𝑏𝑖,𝑡 are the standardized ESG
ratings given to stock 𝑖 in year 𝑡 by agency 𝑎 and 𝑏, respectively. We then obtain the ESG rating disagreement 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒𝑖,𝑡 by
the mean of pairwise uncertainty across all rater pairs.
Dependent variable. The dependent variable in our model is the stock price crash risk, typically referred to as the probability of
extreme negative return (Jin and Myers, 2006; Kim et al., 2011a,b). Hence, we measure stock crash risk by Negative Skewness
(𝑁𝐶𝑆𝐾𝐸𝑊 ) and Down-to-Up Volatility (𝐷𝑈 𝑉 𝑂𝐿) following Hutton et al. (2009), Kim et al. (2011a) and Kim et al. (2011b). To
construct these two variables, we first obtain the residual 𝜀𝑖,𝑠 of stock 𝑖 in week 𝑠 from the following regression using a rolling
window of 180 trading days with return data from the CSMAR database:
∗ ∗ ∗ ∗ ∗
𝑟𝑖,𝑠 = 𝛼 + 𝛽1,𝑖 𝑟𝑚,𝑠−2 + 𝛽2,𝑖 𝑟𝑚,𝑠−1 + 𝛽3,𝑖 𝑟𝑚,𝑠 + 𝛽4,𝑖 𝑟𝑚,𝑠+1 + 𝛽5,𝑖 𝑟𝑚,𝑠+2 + 𝜀𝑖,𝑠 , (2)
where 𝑟𝑖,𝑠 is the return of stock 𝑖 in week 𝑠, 𝑟𝑚,𝑠 is the market return in week 𝑠, we also include lagged and leading market returns
to reduce the bias from non-synchronous trading (Dimson, 1979). We can then obtain the idiosyncratic return of each firm 𝑅𝑖,𝑠 by:
( )
𝑅𝑖,𝑠 = ln 1 + 𝜀𝑖,𝑠 . (3)
Then, the stock price crash risk of a firm, measured by 𝑁𝐶𝑆𝐾𝐸𝑊 and 𝐷𝑈 𝑉 𝑂𝐿 can be measured as:
∑
𝑛(𝑛 − 1)3∕2 𝑅3𝑖,𝑠
𝑁𝐶𝑆𝐾𝐸𝑊 𝑖,𝑡 = − ∑ 2 3∕2 , (4)
(𝑛 − 1)(𝑛 − 2)( 𝑅𝑖,𝑠 )
∑
(𝑛𝑢 − 1) DOWN 𝑅2𝑖,𝑠
𝐷𝑈 𝑉 𝑂𝐿𝑖,𝑡 = ln ∑ , (5)
(𝑛𝑑 − 1) UP 𝑅2𝑖,𝑠
where 𝑛 is the number of trading weeks of stock 𝑖 in year 𝑡. 𝑛𝑑 (𝑛𝑢 ) is the number of weeks when 𝑅𝑖,𝑠 is above (below) the average
idiosyncratic return of stock 𝑖 in year 𝑡. Negative Skewness 𝑁𝐶𝑆𝐾𝐸𝑊 measures the probability of negatively deviating stock returns,
and Down-to-Up Volatility 𝐷𝑈 𝑉 𝑂𝐿 defines the ratio of downward fluctuations to upward fluctuations in returns. Consequently,
higher values of 𝑁𝐶𝑆𝐾𝐸𝑊 or 𝐷𝑈 𝑉 𝑂𝐿 indicate a higher stock price crash risk of the asset.
Control variables. We incorporate a list of control variables in our models in line with existing literature (Chen et al., 2001;
Hutton et al., 2009). These control variables are collected from the CSMAR database and consist of firm size (𝑙𝑛𝐴𝑆𝑆𝐸𝑇 ), leverage
ratio(𝐿𝐸𝑉 ), market-to-book ratio (𝑀𝐵), return on equity (𝑅𝑂𝐸), the natural logarithm of turn-over rates (𝑙𝑛𝑇 𝑂𝑉 𝐸𝑅), the standard
deviation of idiosyncratic returns (𝑆𝐼𝐺𝑀𝐴), age (𝐴𝐺𝐸), and 𝑂𝑃 𝐴𝑄𝑈 𝐸 which is the annual discretionary accruals of the modified
Jones model following Jones (1991), Dechow et al. (1995), and Hutton et al. (2009). All control variables are winsorzied on 1%
and 99% percentiles in order to reduce the influence of outliers.
The summary statistics for the independent, dependent, and control variables are presented in Table 2. After excluding firms
rated by fewer than two rating agencies and firms with insufficient data, our dataset comprises 16,437 observations, encompassing
3459 firms from 2010 to 2021 in an unbalanced panel. The explanatory variable 𝑚𝐸𝑆𝐺 ranges from 1 to 10, while 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒
has a mean value of 1.868 and a maximum of 6.364, indicating its non-negligible existence.
2 Ratings from Wind and Rankins start from 2019 and CASVI starts from 2016.
3
D. Luo et al. Finance Research Letters 58 (2023) 104479
Table 2
Summary statistics of the variables.
Variables N Mean Std. Dev. Min Max
NCSKEW 16,437 −0.439 0.751 −2.597 1.685
𝐷𝑈 𝑉 𝑂𝐿 16,437 −0.301 0.482 −1.458 1.009
mESG 16,437 5.325 2.222 1 10
ESGdisagree 16,437 1.868 1.283 0 6.364
LEV 16,437 0.475 0.209 0.0721 0.937
lnASSET 16,437 22.85 1.479 20.10 26.84
MB 16,437 0.318 0.158 −0.0397 0.779
SIGMA 16,437 0.121 0.0559 0.0413 0.346
lnTOVER 16,437 5.886 0.823 3.573 7.588
ROE 16,437 0.0579 0.156 −0.667 0.408
AGE 16,437 13.05 7.167 2 27
OPAQUE 16,437 0.0877 0.422 −2.016 2.285
CRASH 16,399 9.770 2.519 2 17
ARsq 16,399 0.295 0.147 0.0126 0.748
IDIOSYN 16,380 0.651 0.821 −2.169 2.694
Note: This table reports summary statistics of the dependent, independent and control variables.
All control variables are winsorzied on 1% and 99% percentiles. The last three variables
‘‘𝐶𝑅𝐴𝑆𝐻’’, ‘‘𝐴𝑅𝑠𝑞’’ and ‘‘𝐼𝐷𝐼𝑂𝑆𝑌 𝑁’’ will be discussed in detail in Section 5.
Model specification. We first utilize the following specification to investigate the effects of ESG rating on stock price crash risk among
Chinese listed firms:
where 𝐶𝑟𝑎𝑠ℎ𝑅𝑖𝑠𝑘𝑖,𝑡 represents either 𝑁𝐶𝑆𝐾𝐸𝑊 or 𝐷𝑈 𝑉 𝑂𝐿 of firm 𝑖 on year 𝑡. The mean of the standardized ESG ratings from
the prior year (𝑚𝐸𝑆𝐺𝑖,𝑡−1 ) serves as an explanatory variable. To address potential reverse causality concerns, both the independent
variable 𝑚𝐸𝑆𝐺 and the control variables 𝐶𝑜𝑛𝑡𝑟𝑜𝑙𝑠 are lagged by one year. Additionally, we include firm-fixed effects and year-fixed
effects in the model.
Next, to account for the influence of ESG rating disagreement, we further integrate the variable 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and its interaction
term with 𝑚𝐸𝑆𝐺 into the baseline model:
4. Results
The regression outcomes of model (6) and (7) are presented in Table 3. The first two columns report the estimation results of
model (6), employing 𝑁𝐶𝑆𝐾𝐸𝑊 and 𝐷𝑈 𝑉 𝑂𝐿 as dependent variables, respectively. The coefficients of 𝑚𝐸𝑆𝐺 are −0.0110 and
−0.00661, both significant at the 99% confidence level, implying that 1 unit of increase in the mean of ESG is correlated with a
−0.0110 unit of 𝑁𝐶𝑆𝐾𝐸𝑊 or −0.00661 unit of 𝐷𝑈 𝑉 𝑂𝐿 after controlling other factors. In essence, our results are in line with Zhou
et al. (2021) and Feng et al. (2022), and suggest that firms with higher average ESG ratings are less likely to experience a stock
crash compared to those with lower mean ESG ratings, ceteris paribus.
The third and fourth columns display the estimation results of model (7), which incorporates 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and the interaction
term 𝑚𝐸𝑆𝐺 × 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 into the model. According to model (7), the marginal effect of the ESG rating on stock crash risk is
estimated by 𝛽1 + 𝛽3 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒. As depicted in the table, the coefficients of both 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and 𝑚𝐸𝑆𝐺 × 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 are
significant at the 95% level.
The opposite signs of 𝛽1 and 𝛽3 suggest a mitigating effect of ESG disagreement. In other words, when the mean ESG rating is
high but there is a significant disagreement among the ratings, the beneficial effect of higher ESG ratings on crash risk is diminished.
This result proposes that the advantage of a higher ESG rating in decreasing stock crash risk is only accentuated for firms with lower
ESG disagreement since these firms display a stronger congruence of ESG-related values and practices, as perceived by both internal
and external parties. Consequently, we accept H1 and H2.
5. Robustness tests
We conduct several robustness tests to ensure the validity of our findings. We first exclude the stocks marked as ST or PT3 to
avoid the outlier effects. Moreover, we replace the dependent variable with three other proxies of stock crash risk. For one, we
3 In the Chinese stock market, ‘‘ST’’ stands for ‘‘Special Treatment’’ and refers to companies that have been flagged for financial difficulties or potential
bankruptcy by the China Securities Regulatory Commission (CSRC). ‘‘PT’’ stands for ‘‘Potential Treatment’’ and refers to companies deemed to be at risk of being
flagged as ST stocks in the future.
4
D. Luo et al. Finance Research Letters 58 (2023) 104479
Table 3
Results of the baseline model.
Variables (1) (2) (3) (4)
NCSKEW DUVOL NCSKEW DUVOL
mESG −0.0110∗∗∗ −0.00661∗∗∗ −0.0192∗∗∗ −0.0109∗∗∗
(0.00298) (0.00182) (0.00436) (0.00271)
ESGdisagree −0.0373∗∗ −0.0190∗∗
(0.0153) (0.00962)
mESG × ESGdisagree 0.00684∗∗ 0.00363∗∗
(0.00268) (0.00169)
LEV −0.210∗∗∗ −0.0972∗∗∗ −0.215∗∗∗ −0.0993∗∗∗
(0.0492) (0.0301) (0.0492) (0.0301)
lnASSET 0.0154∗∗ 0.00336 0.0166∗∗ 0.00397
(0.00703) (0.00427) (0.00703) (0.00428)
MB −0.428∗∗∗ −0.182∗∗∗ −0.428∗∗∗ −0.182∗∗∗
(0.0533) (0.0327) (0.0532) (0.0327)
SIGMA −2.555∗∗∗ −1.673∗∗∗ −2.556∗∗∗ −1.673∗∗∗
(0.132) (0.0835) (0.132) (0.0835)
lnTOVER −0.0277∗∗∗ −0.0242∗∗∗ −0.0272∗∗∗ −0.0240∗∗∗
(0.0103) (0.00642) (0.0103) (0.00642)
ROE 0.000347 −0.000116 0.000346 −0.000118
(0.000416) (0.000262) (0.000416) (0.000262)
AGE −0.00623∗∗∗ −0.00364∗∗∗ −0.00630∗∗∗ −0.00366∗∗∗
(0.000994) (0.000598) (0.000996) (0.000600)
OPAQUE 0.0102 0.0145 0.00841 0.0135
(0.0228) (0.0138) (0.0228) (0.0138)
Constants 0.0678 0.317∗∗ 0.0789 0.322∗∗
(0.215) (0.134) (0.215) (0.134)
Year-fixed Effects Yes Yes Yes Yes
Firm-fixed Effects Yes Yes Yes Yes
N 16437 16437 16437 16437
Adj.R2 0.0706 0.0873 0.0710 0.0875
Note: This table reports the results of OLS regression of model (6) and model (7). Standard errors
are in parentheses. ∗ 𝑝 < 0.1, ∗∗ 𝑝 < 0.05, ∗∗∗ 𝑝 < 0.01.
follow Hutton et al. (2009) to capture the incidence of extreme negative price jump by the occurrences of stock price below the
1% of the distribution of the firm’s idiosyncratic returns in the given year. Like 𝑁𝐶𝑆𝐾𝐸𝑊 and 𝐷𝑈 𝑉 𝑂𝐿, higher ‘‘𝐶𝑅𝐴𝑆𝐻’’ is
corresponding to greater stock price crash risk. The second one is the adjusted 𝑅2 from Eq. (2) following Morck et al. (2000),
denoted by ‘‘𝐴𝑅𝑠𝑞’’. Since the adjusted 𝑅2 of this model captures the market synchronicity of the firm, it is negatively related to the
company-specific risk on its earning management (Vo and Phan, 2019). For the last replacement variable, we obtain the firm-specific
idiosyncratic risk as the log-transformation of the 𝑅2 from Eq. (2):
( )
1 − 𝑅2
𝐼𝐷𝐼𝑂𝑆𝑌 𝑁 = ln . (8)
𝑅2
As discussed in Section 1, we expect firms with higher (average) ESG ratings to be less prone to stock crash risk and to have stock
returns that are less volatile. As a result, we expect the marginal effect of 𝑚𝐸𝑆𝐺 to be positive on ‘‘𝐶𝑅𝐴𝑆𝐻’’ and ‘‘𝐼𝐷𝐼𝑂𝑆𝑌 𝑁’’,
and negative on ‘‘𝐴𝑅𝑠𝑞’’.
The results of the robustness tests are shown in Table 4. The first two columns indicate that after excluding samples marked ‘‘ST’’
or ‘‘PT’’, our findings remain consistent with the baseline regression. In columns (3), (4), and (5), the directions and significance of
the estimated coefficients of 𝑚𝐸𝑆𝐺, 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒, and their interactions are as expected when switching the dependent variable
to 𝐶𝑅𝐴𝑆𝐻, 𝐴𝑅𝑠𝑞, and 𝐼𝐷𝐼𝑂𝑆𝑌 𝑁. These results collectively indicate that firms’ ESG performances are negatively associated with
stock crash risk, while ESG disagreement across different rating agencies attenuates the effect of ESG.
6. Heterogeneity test
In line with the broader themes discussed in our literature review and building upon established theoretical frameworks, our
heterogeneity analysis delves deeper into the differentiated impact of ESG ratings and their disagreements on stock price crash risks
across different types of firms. A critical differentiator in this context is the state ownership dimension. State-Owned Enterprises
(SOEs) operate under distinct conditions compared to their non-SOE counterparts. The implications of ESG discrepancies for SOEs
are particularly noteworthy for several interconnected reasons.
Firstly, rooted in their ownership structure, SOEs are inherently exposed to elevated levels of public scrutiny and have intensified
expectations concerning ESG performance. This heightened scrutiny can lead to pronounced operational implications. Specifically,
5
D. Luo et al. Finance Research Letters 58 (2023) 104479
Table 4
Results of the robustness tests.
Variables (1) (2) (3) (4) (5)
NCSKEW DUVOL CRASH ARsq IDIOSYN
mESG −0.0176∗∗∗ −0.0104∗∗∗ −0.0548∗∗∗ 0.00567∗∗∗ −0.0288∗∗∗
(0.00471) (0.00282) (0.0146) (0.000607) (0.00384)
ESGdisagree −0.0340∗∗ −0.0181∗ −0.222∗∗∗ 0.00268∗ −0.0251∗
(0.0165) (0.0100) (0.0520) (0.00206) (0.0129)
mESG × ESGdisagree 0.00620∗∗ 0.00351∗∗ 0.0319∗∗∗ −0.000347 ∗
0.00370∗
(0.00289) (0.00176) (0.00912) (0.000361) (0.00226)
LEV −0.237∗∗∗ −0.116∗∗∗ −0.974∗∗∗ −0.00314 0.154∗∗∗
(0.0532) (0.0313) (0.162) (0.00714) (0.0457)
lnASSET 0.0211∗∗∗ 0.00617 0.130∗∗∗ 0.0159∗∗∗ −0.148∗∗∗
(0.00764) (0.00446) (0.0231) (0.00104) (0.00671)
MB −0.448∗∗∗ −0.192∗∗∗ −1.542∗∗∗ 0.135∗∗∗ −0.548∗∗∗
(0.0573) (0.0338) (0.176) (0.00764) (0.0487)
SIGMA −2.736∗∗∗ −1.770∗∗∗ −0.319 −0.332∗∗∗ 1.553∗∗∗
(0.142) (0.0868) (0.451) (0.0174) (0.109)
lnTOVER −0.0170 −0.0205∗∗∗ 0.106∗∗∗ −0.0129∗∗∗ 0.00733
(0.0112) (0.00671) (0.0346) (0.00143) (0.00902)
ROE 0.000726 0.000172 0.00890∗∗∗ −0.000235∗∗∗ −0.000636∗
(0.000459) (0.000279) (0.00141) (0.0000560) (0.000351)
AGE −0.00705∗∗∗ −0.00395∗∗∗ −0.00898∗∗∗ 0.000101 −0.00529∗∗∗
(0.00107) (0.000620) (0.00323) (0.000152) (0.000987)
OPAQUE 0.00989 0.0150 −0.00886 0.0148∗∗∗ −0.101∗∗∗
(0.0244) (0.0142) (0.0746) (0.00336) (0.0215)
Constants −0.0544 0.257∗ 8.064∗∗∗ 0.00880 3.885∗∗∗
(0.234) (0.140) (0.679) (0.0288) (0.183)
Year-fixed Effects Yes Yes Yes Yes Yes
Firm-fixed Effects Yes Yes Yes Yes Yes
N 16251 16251 16399 16399 16380
Adj.R2 0.0718 0.0914 0.0305 0.5741 0.5181
∗ ∗∗
Note: This table reports the results of robustness tests. Standard errors are in parentheses. 𝑝 < 0.1, 𝑝 < 0.05,
∗∗∗
𝑝 < 0.01
any lapses in ESG performance can be detrimental, leading to significant reputational hits and possible governmental sanctions,
both of which can elevate stock price crash risks. Recognizing these unique characteristics, the investment community often affords
special attention to the ESG metrics of SOEs, realizing its substantial bearing on anticipated profitability. Secondly, the literature has
highlighted the tendency of SOEs to exhibit reduced financial and governance transparency (Bushman et al., 2004). Such opacity
can act as a fertile ground for divergences in ESG ratings, increasing the propensity for unforeseen negative stock price movements.
Based on these considerations, we propose the following hypothesis:
H3: The effects of ESG performance and ESG disagreement are more pronounced in SOEs than in non-SOEs.
The results of the heterogeneity test are presented in Table 5. As illustrated in the first two columns, the results for SOEs align
with the baseline model. The coefficients of 𝑚𝐸𝑆𝐺 for the non-SOE subgroup, depicted in the third and fourth columns, are either
less significant or of lesser magnitude, and the coefficients of 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and 𝑚𝐸𝑆𝐺 × 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 are insignificant.4 Lastly, the
p-values from the SUEST tests are presented in the table. The SUEST tests assess the statistical differences in the coefficients of 𝑚𝐸𝑆𝐺,
𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and 𝑚𝐸𝑆𝐺 × 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 between SOEs and non-SOEs. With 𝑁𝐶𝑆𝐾𝐸𝑊 and 𝐷𝑈 𝑉 𝑂𝐿 as the dependent variables,
the p-values stand at 0.0081 and 0.0514, respectively. These findings allow us to reject the null hypothesis that the coefficients are
consistent across the studied subgroups, with the divergence being statistically significant at a 90% confidence interval or higher.
These results corroborate the null hypothesis H3 and collectively indicate that ESG concepts are less thoroughly ingrained in non-
SOEs compared with SOEs. The consequences of ESG disagreement, particularly under stringent regulation as exhibited by SOEs,
appear to be more severe.
7. Conclusion
This study examines the relationship between the highly diverged ESG ratings and stock price crash risk among Chinese listed
firms. While our findings corroborate that better ESG performance lower the stock price crash risk, ESG disagreement among rating
4 Although not reported in this table, the p-values of the coefficients of 𝑚𝐸𝑆𝐺 and 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 in the third and fourth columns of Table 5 are 0.203,
0.111, 0.536, and 0.298, respectively, all falling below the 90% confidence level threshold.
6
D. Luo et al. Finance Research Letters 58 (2023) 104479
Table 5
Results of heterogeneity tests.
Variables (1) (2) (3) (4)
𝑁𝐶𝑆𝐾𝐸𝑊 𝐷𝑈 𝑉 𝑂𝐿
𝑆𝑂𝐸 𝑁𝑜𝑛 − 𝑆𝑂𝐸 𝑆𝑂𝐸 𝑁𝑜𝑛 − 𝑆𝑂𝐸
mESG −0.0199∗∗∗ −0.0163∗∗∗ −0.0125∗∗∗ −0.00867∗∗
(0.00674) (0.00580) (0.00419) (0.00360)
ESGdisagree −0.0588∗∗ −0.0246 −0.0398∗∗∗ −0.00681
(0.0243) (0.0197) (0.0153) (0.0124)
mESG × ESGdisagree 0.00939∗∗ 0.00553 0.00633∗∗ 0.00213
(0.00416) (0.00352) (0.00262) (0.00221)
LEV −0.146∗ −0.236∗∗∗ −0.0406 −0.120∗∗∗
(0.0800) (0.0625) (0.0485) (0.0384)
lnASSET 0.00715 0.0298∗∗∗ −0.000554 0.0103∗
(0.0109) (0.00946) (0.00658) (0.00577)
MB −0.324∗∗∗ −0.481∗∗∗ −0.0852 −0.237∗∗∗
(0.0868) (0.0684) (0.0531) (0.0421)
SIGMA −3.145∗∗∗ −2.128∗∗∗ −2.009∗∗∗ −1.418∗∗∗
(0.211) (0.169) (0.134) (0.107)
lnTOVER −0.00968 −0.0491∗∗∗ −0.0152 −0.0347∗∗∗
(0.0163) (0.0136) (0.0102) (0.00844)
ROE 0.00206∗∗∗ −0.000604 0.00101∗∗ −0.000746∗∗
(0.000718) (0.000517) (0.000454) (0.000325)
AGE −0.00621∗∗∗ −0.00517∗∗∗ −0.00444∗∗∗ −0.00255∗∗∗
(0.00165) (0.00141) (0.000992) (0.000848)
OPAQUE 0.00864 0.0528 0.0107 0.0413∗
(0.0292) (0.0373) (0.0176) (0.0228)
Constant 0.200 −0.0703 0.353∗ 0.299
(0.316) (0.315) (0.195) (0.197)
p-value of SUEST test 0.0081 0.0541
Year-fixed Effects Yes Yes Yes Yes
Firm-fixed Effects Yes Yes Yes Yes
N 6981 9456 6981 9456
Adj.R2 0.0993 0.0610 0.1256 0.0721
Note: This table reports the results of subgroup analyses, which divide the sample into two subgroups: the
SOEs and the non-SOEs. Standard errors are in parentheses. The SUEST test is used to compare coefficients
between the SOE and non-SOE subgroups, specifically evaluating if the coefficients of 𝑚𝐸𝑆𝐺, 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 and
𝑚𝐸𝑆𝐺 × 𝐸𝑆𝐺𝑑𝑖𝑠𝑎𝑔𝑟𝑒𝑒 differ statistically across them. ∗ 𝑝 < 0.1, ∗∗ 𝑝 < 0.05, ∗∗∗ 𝑝 < 0.01.
agencies can temper this beneficial impact. Specifically, the benefit of a higher ESG rating in mitigating stock price crash risk is only
pronounced for firms with lower ESG disagreement. The robustness tests also confirm the results. Our subgroup analyses further
reveal that these effects exhibit group-specific heterogeneity, being more statistically significant in SOEs. This highlights the unique
challenges and expectations faced by state-owned entities regarding their ESG performance and its impacts on their stock price crash
risk.
The authors declare that they have no known competing financial interests or personal relationships that could have appeared
to influence the work reported in this paper.
Data availability
References
Avramov, D., Cheng, S., Lioui, A., Tarelli, A., 2022. Sustainable investing with ESG rating uncertainty. J. Financ. Econ. 145 (2), 642–664.
Berg, F., Koelbel, J., Pavlova, A., Rigobon, R., 2021. ESG confusion and stock returns: Tackling the problem of noise. Soc. Sci. Electron. Publ..
Bushman, R.M., Piotroski, J.D., Smith, A.J., 2004. What determines corporate transparency? J. Account. Res. 42 (2), 207–252.
Callen, J.L., Fang, X., 2015. Short interest and stock price crash risk. J. Bank. Financ. 60, 181–194.
Cao, C., Xia, C., Chan, K.C., 2016. Social trust and stock price crash risk: Evidence from China. Int. Rev. Econ. Finance 46, 148–165.
Capelli, P., Ielasi, F., Russo, A., 2023. Integrating ESG risks into value-at-risk. Finance Res. Lett. 103875.
7
D. Luo et al. Finance Research Letters 58 (2023) 104479
Chen, J., Hong, H., Stein, J.C., 2001. Forecasting crashes: Trading volume, past returns, and conditional skewness in stock prices. J. Financ. Econ. 61 (3),
345–381.
Choi, J., Wang, H., 2009. Stakeholder relations and the persistence of corporate financial performance. Strateg. Manage. J. 30 (8), 895–907.
Christensen, D.M., Serafeim, G., Sikochi, A., 2022. Why is corporate virtue in the eye of the beholder? The case of ESG ratings. Account. Rev. 97 (1), 147–175.
Dechow, P.M., Sloan, R.G., Sweeney, A.P., 1995. Detecting earnings management. Account. Rev. 193–225.
Deng, X., Kang, J., Low, B.S., 2013. Corporate social responsibility and stakeholder value maximization: Evidence from mergers. J. Financ. Econ. 110 (1), 87–109.
Dimson, E., 1979. Risk measurement when shares are subject to infrequent trading. J. Financ. Econ. 7 (2), 197–226.
Feng, J., Goodell, J.W., Shen, D., 2022. ESG rating and stock price crash risk: Evidence from China. Finance Res. Lett. 46, 102476.
Gibson Brandon, R., Krueger, P., Schmidt, P.S., 2021. ESG rating disagreement and stock returns. Financ. Anal. J. 77 (4), 104–127.
He, F., Ding, C., Yue, W., Liu, G., 2023. ESG performance and corporate risk-taking: Evidence from China. Int. Rev. Financ. Anal. 87, 102550.
He, F., Qin, S., Liu, Y., Wu, J.G., 2022. CSR and idiosyncratic risk: Evidence from ESG information disclosure. Finance Res. Lett. 49, 102936.
Hutton, A.P., Marcus, A.J., Tehranian, H., 2009. Opaque financial reports, R2, and crash risk. J. Financ. Econ. 94 (1), 67–86.
Jin, L., Myers, S.C., 2006. R2 around the world: New theory and new tests. J. Financ. Econ. 79 (2), 257–292.
Jones, J.J., 1991. Earnings management during import relief investigations. J. Account. Res. 29 (2), 193–228.
Kim, Y., Li, H., Li, S., 2014. Corporate social responsibility and stock price crash risk. J. Bank. Financ. 43, 1–13.
Kim, J.-B., Li, Y., Zhang, L., 2011a. CFOs versus CEOs: Equity incentives and crashes. J. Financ. Econ. 101 (3), 713–730.
Kim, J.-B., Li, Y., Zhang, L., 2011b. Corporate tax avoidance and stock price crash risk: Firm-level analysis. J. Financ. Econ. 100 (3), 639–662.
Knox, S., Maklan, S., French, P., 2005. Corporate social responsibility: Exploring stakeholder relationships and programme reporting across leading FTSE companies.
J. Bus. Ethics 61, 7–28.
Lashkaripour, M., 2023. ESG tail risk: The Covid-19 market crash analysis. Finance Res. Lett. 53, 103598.
Li, H., Zhang, X., Zhao, Y., 2022. ESG and firm’s default risk. Finance Res. Lett. 47, 102713.
Luo, D., Shan, X., Yan, J., Yan, Q., 2023. Sustainable investment under ESG volatility and ambiguity. Econ. Model. 106471.
Morck, R., Yeung, B., Yu, W., 2000. The information content of stock markets: why do emerging markets have synchronous stock price movements? J. Financ.
Econ. 58 (1), 215–260, Special Issue on International Corporate Governance.
Pástor, L., Stambaugh, R.F., Taylor, L.A., 2021. Sustainable investing in equilibrium. J. Financ. Econ. 142 (2), 550–571.
Serafeim, G., Yoon, A., 2022. Stock price reactions to ESG news: The role of ESG ratings and disagreement. Rev. Account. Stud. 1–31.
Shanaev, S., Ghimire, B., 2022. When ESG meets AAA: The effect of ESG rating changes on stock returns. Finance Res. Lett. 46, 102302.
Vo, X.V., Phan, D.B.A., 2019. Herd behavior and idiosyncratic volatility in a frontier market. Pac.-Basin Finance J. 53, 321–330.
Wu, C.-M., Hu, J.-L., 2019. Can CSR reduce stock price crash risk? Evidence from China’s energy industry. Energy Policy 128, 505–518.
Yu, H., Liang, C., Liu, Z., Wang, H., 2023. News-based ESG sentiment and stock price crash risk. Int. Rev. Financ. Anal. 88, 102646.
Zhou, F., Zhu, J., Qi, Y., Yang, J., An, Y., 2021. Multi-dimensional corporate social responsibilities and stock price crash risk: Evidence from China. Int. Rev.
Financ. Anal. 78, 101928.