Central environmental protection inspector and stock price crash risk—evidence from polluting industries firms in China

In recent years, under the background of vigorously promoting environmental governance, the implementation effect of the central environmental protection inspection is an issue of great concern to the government and the public. This paper systematically investigates the impact of central environmental protection inspection on the risk of stock price crash using a sample of listed firms in polluting industries. The results show that compared with non-supervised areas, central environmental protection inspection can reduce the polluting industries’ firms’ stock price crash risk by reducing stock price bubbles. After a series of robustness tests, the results still held. The above transmission mechanism is more effective in the samples of private enterprises, low information transparency and disclosure quality enterprises, non-national civilized urban areas, and high promotion incentive areas. Furthermore, this paper found that there were differences in the effects of central environmental protection inspection in different batches. Among the effects of central environmental protection inspection in different batches, the effect of environmental regulation in the second, third, and fourth batches was better, and the effect of central environmental protection inspection in different batches gradually deepened. Finally, by analyzing the environmental governance of the central environmental protection inspection, it is found that the central environmental protection inspection has significant short-term and long-term control effect in air pollution governance, and it is still necessary to strengthen the law enforcement in water pollution governance.


Introduction
China's rapid economic development has also brought about many environmental problems, such as water, soil, and air pollution. The increasingly prominent ecological and environmental problems have attracted the attention of the central government, which has gradually formulated corresponding environmental governance acts. On April 24, 2014, the eighth meeting of the Standing Committee of the 12th National People's Congress amended the Environmental Protection Law of the People's Republic of China, which clearly established environmental protection as a basic state policy. The implementation effect of environmental policies has been a key concern of the relevant authorities, but due to the strength of local government environmental enforcement and the low cost of environmental violations by enterprises, the implementation effect of environmental policies has not been satisfactory.
Compared with previous environmental policies, the Central Environmental Protection Inspector (CEPI) has made the results of the inspectors an important basis for the assessment and appointment of leading cadres, increased the assessment of local governments' environmental performance, enhanced local governments' enforcement efforts, and improved the problems of poor policy implementation and difficult accountability of local governments in the process of environmental governance. Scholars have studied the implementation effect of CEPI in terms of environmental quality improvement (Wu and Hu 2019 b; Jia and Chen 2019; He and Geng 2020; Zhang et al. 2018), corporate stock value (Tian et al. 2019), and capital market response  (Zeng et al. 2021), respectively. Then, in a market environment where China's environmental regulatory system is not yet perfect, can CEPI have a good effect on environmental information disclosure, reduce the hoarding of negative corporate information, and reduce the risk of corporate stock price crash risk? Are there differences in the effects of CEPI between batches? The answers to the above two questions help us further assess the effectiveness of the implementation of the CEPI policy. The existing literature examining the causes of stock price crash risk focuses on both internal corporate governance factors (Hutton et al. 2009;Xu et al. 2014) and external environmental factors (Hong and Stein 2003;Callen and Fang 2015) and concludes that information opacity from the cover-up of negative news is an important cause of crash risk. Few scholars have analyzed the formation of stock price crash risk from the perspective of environmental information disclosure. Our study is a strong addition to the relevant literature.
We find that CEPI reduces the risk of stock price collapse for firms in polluting industries in the inspector areas compared to non-inspector areas. We also find that the effect of CEPI on firms' stock price crash risk compared to non-environmental inspector regions is mainly found in the sample of private firms and those with lower information transparency and in the sample of non-national civilized cities and regions with weaker incentives for officials' promotion. Further analysis finds that CEPI reduces the risk of stock price crash for firms in polluting industries by reducing stock price bubbles. Finally, the action of the batches of CEPI, mainly the second, third, and fourth batches, reduces the stock price crash risk of polluting industry firms in the region compared to non-environmental inspector regions.
Compared with the existing literature, the contribution of this paper is mainly in the following three aspects: First, the market effects of CEPI are identified from the perspective of individual stock price crash risk, and the intrinsic impact mechanism is analyzed at the stock price bubble level. This paper provides new empirical evidence for a deep understanding of the economic consequences of CEPI and extends CEPI-related research. The most relevant literature to this paper is Zhang et al. (2021); we differ from Zhang et al. (2021) in three ways. The first is the difference in data structure. Zhang et al. (2021) use firm-year level panel data. The time span between batches of CEPI is known from the policy compendium section to be less than 1 year. Using yearly data may not clearly identify the implementation effects of each batch of CEPI. Therefore, we use firm-quarter level panel data for regression analysis. The second is the mechanism of the impact of the CEPI. We analyze the potential mechanisms by which CEPI affects the firm's stock price crash risk from the stock price bubble perspective. The third is the environmental governance effect of the CEPI. We systematically analyze the short-run and long-run environmental governance effects of the CEPI. Second, most of the studies related to environmental disclosure have endogeneity problems. This is because relying on methods such as environmental reports (Gray et al. 2001), annual and independent reports (Guthrie et al. 2008), social responsibility reports (Lu and Abeysekera 2014), press conferences and corporate meetings (Zeghal and Ahmed 1990) to measure environmental information disclosure quality inevitably gives rise to endogeneity problems such as mutual causality and sample selfselection. In this paper, we use the exogenous shock event of CEPI to conduct the analysis, which greatly alleviates the endogeneity problem and more accurately reveals the market effect of environmental information disclosure. Third, this paper also assesses the environmental governance effects of CEPI. It is found that in terms of air pollution treatment, the CEPI increases the long-term treatment investment and improves the short-term emission reduction effect of enterprises. In water pollution treatment, the effect of CEPI on long-term treatment investment and short-term emission reduction effect is weaker. The research in this paper contributes to an indepth understanding of the mechanism of CEPI and provides a reference for the government to use existing tools to improve environmental governance.

Policy background
In July 2015, the Environmental Protection Inspection Program (Pilot) was introduced, and the Chinese central government established an environmental protection inspection mechanism (Wang 2021;Wang et al. 2022). The first round of CEPI was then carried out in batches (pilot, first batch, second batch, third batch and fourth batch) in 31 provinces, cities, and regions across the country. Unlike previous environmental policies, the CEPI has the characteristic of monitoring both "enterprises" and "local governments." This means that the CEPI is no longer looking for a short-lived "political blue sky," but rather sustainable development. CEPI provides feedback to local governments on environmental pollution through mass monitoring, such as calls and letters. They then urge the local governments to rectify the situation by a deadline and make the local governments disclose the rectification plans and implementation status to the society. It can be said that the CEPI are a major innovation in China's environmental regulatory system. The CEPI inspected 31 provinces, cities, and regions from January 4, 2016 to September 15, 2017. From January 4, 2016 to February 4, 2016, a pilot CEPI was launched in Hebei Province. Since then, the environmental protection inspectors have taken 21 months to achieve full coverage of the country's 31 provinces, cities, and regions, with each batch geared toward the eastern, central, and western regions. Each batch lasts about a month and involves 7 or 8 provinces, cities, and regions. The whole inspection process is divided into three stages; each stage lasts about 10 days. In the first stage, the inspection team talks with the provincial party committees, provincial governments, and leaders of relevant departments and receives complaints from the public. The second stage investigates and verifies environmental problems by checking official documents and investigating pollution on the ground and comes up with a list of environmental problems. The third stage speaks about the list of problems transferred to the local government, requiring it to rectify and feedback by a deadline. Pilot areas and four batches of environmental protection inspectors received a total of 99,783 cases reported by the public, shut down and rectified 82,081 enterprises, and interviewed 17,601 relevant personnel; 17,707 people were held accountable, detained 1543 people, and fined 1.306 billion yuan. Among them, the fines are not less than 100 million yuan in the following regions: Shandong Province, Jiangsu Province, Zhejiang Province, Fujian Province, and Guangdong Province. The region with the most fines is Zhejiang Province.

Literature review
There is an incentive for corporate management to hide negative news for compensation incentives and to reduce litigation. The concentrated release of negative information after hoarding can exacerbate the risk of stock price crash of a company. Based on the above perceptions, scholars have conducted numerous studies. The factors influencing the risk of stock price crash mainly include internal governance mechanisms and external governance mechanisms. Internal governance mechanisms include information transparency (Hutton et al. 2009;Defond et al. 2015), shareholder characteristics (Boubaker et al. 2014;Zhou et al. 2021), director characteristics (Jebran et al. 2020;Jin et al. 2022), management overconfidence (Kim and Zhang 2016), and corporate social responsibility (Thuy et al. 2021(Thuy et al. , 2022. External governance mechanisms include tax avoidance (Kim et al. 2011), investor protection (Zhang et al. 2017), media coverage , economic policy uncertainty (Jin et al. 2019), and religious beliefs (Callen and Fang 2015).
In addition, it appears that investor sentiment affects the risk of stock price crashes in terms of capital market reactions (Hong and Stein 2003). With a high proportion of retail investors in the Chinese stock market, market investors are not fully rational, and investor behavior is vulnerable to market signals. Solomon (2012) found that companies' selective disclosure of positive news and concealment of negative news can lead to depressed investor sentiment and significant stock price declines afterwards. Corporate social responsibility can enhance corporate value . Therefore, symbolic green behavior or concealment of negative news by companies can lead to optimistic investor sentiment accompanied by an irrational increase in stock prices. However, when negative corporate information is disclosed, investors become depressed and dump their shares, stock prices fall, and the risk of stock price crash increases.
In summary, although the causes of stock price crash risk have been analyzed from the perspective of various types of information disclosure, there is little literature on the impact of environmental information disclosure on stock price crash risk. In the studies related to environmental information disclosure, the way of corporate environmental information disclosure is mainly described qualitatively. For enterprises, the above-mentioned environmental information disclosure methods are still somewhat autonomous and selective. The CEPI provides a good external environment for testing the effectiveness of mandatory environmental information disclosure.
According to the above theory, there is a dual effect of environmental regulation on the risk of stock price crash: One is the information effect. According to principal-agent theory, the existence of corporate agency costs ) leads management to have an incentive to hide negative information (Wu and Hu 2019a ). When firms hoard negative information to a certain extent, the stock price bubble bursts and the stock price falls sharply, causing the risk of stock price crash (Jin and Myers 2006;Kothari et al. 2009). Environmental regulation can enhance the transparency of corporate information, improve the quality of environmental information disclosure, reduce information asymmetry, and improve the accuracy of surplus forecasts, thus reducing the risk of stock price crash (Hutton et al. 2009;Li et al. 2017). Second is investor sentiment effect. Investor sentiment affects investor behavior, and negative sentiment may lead investors to sell their stocks sharply, resulting in a sharp decline in stock prices and causing a stock price crash (Baker and Wurgler 2006). The environmental information that companies are forced to disclose may be negative information concealed by the management, and the concentration of such information disclosure may trigger investors' doubts about the company and increase investors' concerns about the company's environmental risks (Xu et al. 2021). Thus, it will have a shock on the company's stock price and increase the risk of stock price crash.
Our research can provide policy reference for developing countries to improve the environmental regulation system and explore the scope and content of environmental regulation subjects.

Hypothesis formulation
We believe that CEPI affects the risk of corporate stock price crash in two ways: First is the information effect. For one, multiple and continuous CEPIs are conducive to reducing the degree of information asymmetry between the central government and local governments. CEPI with random sampling nature is conducive to eliminating local governments' sheltering behavior for local polluters. It increases the possibility of environmental violations being exposed and reduces the hoarding of negative information about enterprises, which in turn reduces the risk of stock price crash. Next is in terms of public participation in improving air pollution governance research. Public participation can improve environmental performance and environmental governance by enhancing the effectiveness of regulators' resource allocation (Dong et al. 2011) and the efficiency of government decision-making (North et al. 2014), which in turn improves environmental performance and environmental governance (Wu et al. 2018;Li et al. 2018). The public participation feature of CEPI is obvious, with online collection and verification of public complaint information and public disclosure of inspector feedback results, substantially enhancing public participation in environmental governance. This feature of CEPI will further increase the disclosure of negative corporate information, which in turn reduces the risk of stock price crash. Second is the investor sentiment effect. By disclosing the environmental violations of enterprises, as well as the lack of rectification and negative response, CEPI arouses investors' doubts about the enterprises and reduces their trust in them. Investor sentiment goes into a slump, selling stocks, and stock prices fall, exacerbating the risk of corporate stock price crash. In addition, the continued plunge in stock prices will also make investors lose expectations for the future of the stock market. In turn, investors will develop a more pessimistic market sentiment. This could cause a more serious stock market crash or even a recession in the real economy.
Accordingly, we propose the following hypotheses.
Hypothesis 1: CEPI reduces the risk of stock price crash for firms in polluting industries in inspector areas compared to non-inspector areas.
Hypothesis 2: CEPI increases the risk of stock price crash for firms in polluting industries in inspector areas compared to non-inspector areas.

Measuring the stock price crash risk
Referring to Chen et al. (2001) and Hutton et al. (2009), we use NCSKEW and DUVOL to measure the firms' stock price crash risk. The specific calculations are as follows: First, calculate the daily return at the firm level where n is the number of trading days in a quarter. A larger NCSKEW indicates a higher stock price crash risk.
Finally, calculate DUVOL: where n up (n down ) . A is the number of trading days when the daily return of the stock is above (below) the mean of the current quarter's return. The sample is divided into a rising stock price group (up) and a falling stock price group (down) based on whether the stock's daily return is above the average return of the current quarter. A larger value of DUVOL indicates a higher stock price crash risk.

Research design
For the selection of the study interval, we made the following adjustments. First, the CEPI first started on January 4, 2016 and ended on September 15, 2017. Although the environmental protection inspector look-back operation started in early 2018, each region started the inspector look-back only 1 year after the CEPI. Therefore, the inspector look-back policy will basically have no impact on our results. Accordingly, we selected a cut-off date of the end of 2019. Second, although the earliest CEPI began on January 4, 2016, the CEPI team was already stationed in Hebei Province at the end of 2015. We therefore set the start of the control group sample in 2014. Accordingly, we use the data of listed companies in polluting industry firms from 2014 to 2019 and apply the multi-period DID method (1) (3) to test the impact of the CEPI on the stock price crash risk of polluting industry firms: which y i,p,t denotes the risk of stock price crash for firm i in region p and period t. DT p,t denotes a policy dummy variable that takes the value of 1 when an area is launched as an inspector for the quarter and thereafter, and 0 for the others. The definition of polluting enterprises is based on the "List of Listed Enterprises for Environmental Verification Industry Classification and Management" published by the Ministry of Environmental Protection on June 24, 2008. Following Chen et al. (2001) and Kim et al. (2011), we further control for lagged one-period indicators ( C i,p,t−1 ) that may affect the firms' stock price crash risk, including Lev, Size, Roe, Sigma, Ret, Turnover, Top10, and lag NCSKEW.
Also, since district characteristics may affect whether the district is inspected or not, to control for the endogeneity problem caused by omitting district characteristics, we control for a one-period lag of the district-level characteristic variables ( Disc p,t−1 ), including regional economic development indicators LnGDP and regional population indicators LnPop. We further control for firm fixed effects and time fixed effects.

Descriptive statistics
We select data on listed companies in China's polluting industry firms from the first quarter of 2014 to the fourth quarter of 2019 to examine the impact of CEPI on the stock price crash risk of polluting industry firms. Among them, firm-level indicators are from the CSMAR database and regional-level data are from the EPS database. We screen the initial sample as follows: (1) in estimating the stock price crash risk indicators, exclude samples with less than 30 trading days per quarter; (2) exclude samples of financial industry enterprises; (3) exclude "ST" and "*ST" enterprises. The continuous variables were subjected to a 1% tailing process.
Descriptive statistics are shown in Table 1, where the mean values of NCSKEW and DUVOL are − 0.4214 and − 0.3268, respectively, and the median values are − 0.4898 and − 0.3805, respectively. This indicates that the stock price crash risk is at a low level. The mean value of the gearing ratio is 0.4073, with a minimum value of 0.0589 and a maximum value of 0.8788. This indicates that the leverage ratio of listed companies varies widely and is at a low level.

Analysis of basic results
We use a multi-period DID approach to test the impact of CEPI on the firms' stock price crash in polluting industries, and the regression results are shown in Table 2. Controlling for time fixed effects and firm fixed effects, the coefficients of DT are significantly negative after gradually adding firmand region-level control variables. It indicates that the CEPI reduces the firms' stock price crash in polluting industries in the inspector areas compared to non-inspector areas. The results in columns (5)-(6) show that after the CEPI, the firms' stock price crash in the inspector areas is reduced by 0.0538 (NCSKEW) and 0.0454 (DUVOL) compared to the non-inspector areas, which is equivalent to 12.77% and 13.89% of the mean values of stock price crash risk indicators NCSKEW and DUVOL.
A prerequisite for the unbiased results of the difference in difference estimation is that the parallel trend hypothesis is satisfied between the control and experimental groups, i.e., the control and control groups share a common trend of change prior to the event. To test whether the baseline regression satisfies the common trend test, we include the interaction term between the dummy variable and the policy variable at each time point in the regression. And if the coefficient of the interaction term before the inspector is not significant, it indicates that the parallel trend holds. The regression model is as follows: where DT is a dummy variable, takes 1 if province p has an environmental inspector at time t-τ, and 0 otherwise. β 0 is the effect of the environmental protection inspector in the current period, β −1 to β −6 is the effect of the environmental inspector before 1-6 periods, β 1 to β 12 is the effect of the environmental protection inspector after 1-12 periods, and β 1 to β 12 is the dynamic effect of the inspector over time.
If the coefficients of β −1 to β −6 are close to 0 and the coefficients of β 1 to β 12 are different from 0, it indicates that the parallel trend test hypothesis is valid. The results of the parallel trend test in Fig. 1 show that the coefficients of β −2 to β −6 are not statistically significantly different from 0, and the coefficients of β 1 to β 12 are negative and statistically significantly smaller than 0. The parallel trend hypothesis is valid.

Excluding special samples
First, the CEPI may exert unprecedented environmental regulatory pressure on enterprises in the inspected areas. Enterprises may have relocated due to the inspector pressure. Accordingly, we exclude the sample of enterprises that changed their business address between January 1, 2016 and December 31, 2017 from the regression, where the data on business address change are from the WIND database.
Second, the presence of estimated disturbances from other important environmental policies is tested. Since the de-capacity policy in 2016 and 2017 may have a significant impact on the relevant enterprises, the regression is conducted after excluding the sample of enterprises within the sample that belong to the de-capacity focus industries. The de-capacity focus industries released in 2016 and 2017 include six industries: steel, coal, cement, shipbuilding, electrolytic aluminum, and glass, and we adopt the CITIC Securities industry classification criteria to exclude these six-industry enterprise sample. The regression results of the above robustness test are shown in Table 3. Columns (1) and (2) are the regression results excluding the sample of corporate address changes. Columns (3) and (4) are the results of the regressions excluding the de-capacity samples. The coefficient of DT is significantly negative, which is consistent with the regression results in the main results.

Exclusion of spillover effects
The CEPI divides the treatment and control groups based on geographical boundaries. Due to policy externalities, spillover effects may occur between treatment and control groups. That is, CEPI may affect the neighboring control group provinces, which violates the independence between treatment and control groups. Accordingly, we refer to Clarke (2017) to test the policy effects of CEPI after excluding spillover effects. The Spillover-Robust DID method is constructed, and the model is set as follows: where Close p,t denotes the nearest neighbor treatment group effect. If region p has CEPI in quarter t, its neighboring regions are set to 1 in quarter t and thereafter, and the rest are set to 0. The remaining variables are defined in the same way as in Eq. (4). Since CEPI is a one-time inspector mechanism, there is no spillover effect for any of the regions that have been inspected, so none of the inspected regions are set as adjacent regions. The test results of the spillover effect of the CEPI are shown in Table 4, the DT coefficient is significantly negative, and the Close coefficient is not significant, indicating that there is no spillover effect of the CEPI on neighboring areas.

Substitution of stock price crash risk indicators
We use NCSKEW and DUVOL as explanatory variables in the main text. Referring to Callen and Fang (2015), we measured the stock price crash risk by whether the stock return was down or up (Crash), to avoid the estimation bias caused by the selection of explanatory variables. The regression results are shown in column (1) of Table 5, and after replacing the indicators, the regression results are still consistent with the main results.

Placebo test
To test the randomness of the timing of the implementation of the CEPI, we conduct a placebo test for the time of the occurrence of the policy antecedently. If the coefficient of DT remains significantly negative in the results for one period ahead and two periods ahead, it indicates that there are non-CEPI factors driving the change in stock price crash (6) y i,p,t = + 1 DT p,t + 2 Close p,t + C i,p,t−1 + Disc p,t−1 + i + t + i,p,t Table 3 Excluding special samples ***Significant at 1% level; **significant at 5% level; *significant at 10% level  (2)-(5) of Table 5; the coefficient of DT is not significant, ruling out any interference in our results from some of the non-CEPI factors or potentially unpredictable factors.

Corporate heterogeneity test
Compared with state-owned enterprises, private enterprises have fewer agency problems among themselves, and managers' interests are relatively consistent with those of the company. Therefore, when negative information appears in a company, managers will also hide negative information out of personal interest. Negative environmental information disclosure can reduce managers' opportunistic behavior, reduce negative information hiding, and thus reduce the risk of stock price crash. Accordingly, we empirically analyze whether the dampening effect of CEPI on stock price crash risk is more significant among private firms. The regression results of the firm heterogeneity test are shown in Table 6. The coefficient of DT is significantly negative in the private enterprise sample in columns (1)-(4) and insignificant in the state-owned enterprise sample. This indicates that the CEPI reduces the stock price crash risk by increasing the disclosure of negative environmental information of private firms in the inspected areas compared to non-environmental inspector areas, and this effect is not significant in the stateowned enterprise (SOE) sample, reflecting to some extent the sheltering of SOEs by the local government.
In addition to this, to further test the differences between groups. We used Bootstrap method to calculate empirical p-values for the differences between groups in the sample by firm nature. A random sample of 300 times was set. The results of the test are shown in Table 6, and the coefficients between groups show significant differences.
2. Information transparency and disclosure quality heterogeneity test Kim and Zhang (2014) show that the lower the transparency of the firm, the stronger the manager's incentive to hide bad news. Based on this logic, for firms with low transparency of financial information and poor quality of information disclosure, the more managers hide bad news, the higher stock price crash risk of the firm. The CEPI has increased the disclosure of negative environmental information of firms in the inspected areas by increasing environmental supervision. For firms with low transparency, CEPI can disclose more environmental information. The theoretical effect of CEPI on stock price crash risk reduction is more effective in the sample with lower information transparency and poorer disclosure quality.
Based on the above analysis, we empirically test the mechanism of the role of information transparency and information disclosure quality in the stock price crash risk by CEPI. First, we refer to Hutton et al. (2009), which uses the sum of the absolute values of firms' surplus management in the previous 3 years to measure the information transparency of firms, and a larger value indicates a lower information transparency of firms. We divide the 2014 corporate information transparency indicators into a lower corporate transparency sample and a higher transparency sample based on the median of the 2014 corporate information transparency indicators. Second, we further examine the results of heterogeneity in the quality of corporate information disclosure. We use the Shenzhen Stock Exchange's rating of information disclosure quality of listed companies to measure the quality of corporate information disclosure, and its assessment results are divided into four levels: excellent, good, qualified, and unqualified. If the assessment result in 2014 is excellent or good, it is classified as a higher disclosure quality sample and takes the value of 1. If the assessment result in 2014 is qualified or unqualified, it is classified as a lower disclosure quality sample and takes the value of 0. The regression results of the sub-firm transparency heterogeneity test are shown in columns (1)-(4) of Table 7. The coefficient of DT is significantly negative in the low-transparency sample and insignificant in the high-transparency sample. It indicates that the CEPI mainly reduces the firms' stock price crash risk in low-transparency polluting industries in the region compared to non-inspector regions. The regression results of the heterogeneity test of information disclosure quality are shown in columns (5)-(8) of Table 7, where the coefficient of DT is significantly negative in the low disclosure quality sample and insignificant in the high disclosure quality sample. It indicates that the higher information asymmetry, the more significant inhibitory effect of the CEPI on the stock price crash risk. To further test the differences between groups, we calculated empirical p-values for the differences between sub-sample groups using the Bootstrap method, setting a random sample of 300 times. The results of the test are shown in Table 7, where the between-group coefficients of the core explanatory variables all show significant differences.

Officials' promotion incentives heterogeneity test
Although the central government in China maintains its political authority in environmental planning, the most basic implementation decisions and responsibilities have been assigned to local governments, and an environmental decentralization mechanism exists. Previous studies have shown that the official promotion of local officials is linked to the level of regional economic development (Li and Zhou 2005), but the goals of the central and local governments in environmental protection are often inconsistent. Local governments have weak incentives to strengthen local enforcement efforts (Van Rooij 2006) and sometimes provide false environmental data in response to central government environmental assessment requirements (Tian et al. 2020). Unlike previous environmental policies, CEPI is able to monitor local governments by mobilizing the public and includes the results in the assessment and appointment of government officials. Thus, CEPI has the characteristic of monitoring local governments. Zhang et al. (2018) show that direct central regulation can reduce information asymmetry between central and local governments. Xie and Yong (2020) also show that under the deterrent effect of CEPI, political connections are no longer an effective way for polluters to circumvent strict environmental regulation. We therefore expect the mitigating effect of CEPI on stock price crash risk to be more pronounced under higher official promotion incentives.
Accordingly, we test the heterogeneous effect of officials' promotion incentives in the CEPI. We use the following method to define promotion incentive indicators. First, define Remaining Political Age (RPA) as the difference between an official's age at the time of taking office and the legal retirement age. The retirement age is set to be 65 for men and 60 for women. Secondly, the future political promotion space of the officials is defined as the distance between the political level (LE) of their current position and the highest official level. Finally, Politics Promotion (PP) was calculated with PP = RPA/(5-LE). After that, the median of official promotion incentive index in 2014 divided them into a larger sample of official promotion incentive and a smaller sample; the results of official promotion heterogeneity test are shown in Table 8, the coefficients of DT for high official promotion incentive samples in columns (1)-(4) are basically significant negative, and the coefficients of DT for low official promotion incentive samples are not significant, indicating that the higher the official promotion incentive, the more significant effect of CEPI.
To further test the differences between groups, we calculated the empirical p-values of the differences between sub-sample groups using the Bootstrap method, setting a random sample of 300 times, and the test results are shown

Mechanism analysis
We test whether CEPI reduces the stock price crash risk by reducing stock price bubbles. Refer to Dass et al. (2008) to define the stock price bubble indicator. First, we use the price-to-sales ratio (PS) to indicate the size of the stock price bubble of listed companies. The larger the value of PS, the larger the stock price bubble. Second, the sample is sorted by PS from smallest to largest and divided equally into four groups, and the fourth group of stocks is defined as bubble stocks. The dummy variable PSDummy is defined, and the bubble stocks take the value of 1; the rest take the value of 0. Then, we refer to Baron and Kenny (1986) for the mediating effect test to test whether CEPI can reduce the stock price crash risk by reducing the stock price bubble. The regression results are shown in Table 9. In columns (1) and (2), the coefficient of DT is significantly negative, indicating that the CEPI reduces the stock price crash risk. In columns (3) and (6), the coefficient of DT is significantly negative, indicating that the CEPI reduces the corporate stock price bubble. In columns (4) and (5), the coefficients of DT and PS are negative, and the coefficients of DT are smaller than columns (1) and (2). In columns (7) and (8), the coefficient of DT and PSDummy is negative and significant, and the coefficient of DT is smaller than columns (1) and (2). Finally, most of the p statistical value for Sobel test is smaller than 0.05. The regression results in Table 9 show that the mediation effect is valid. That is, CEPI can reduce the stock price crash risk by reducing stock price bubbles.

The batches of CEPI effect test
Considering that the CEPI is environmental policies that gradually expand in scope in batches, the environmental protection inspectors of the previous batch may have certain warning and learning effects on the next batch. Theoretically, CEPI should deepen continuously in each batch. Accordingly, we test the stock price crash risk effect of CEPI in batches, and the results of the batch wise regressions are shown in Table 10. The results in columns (5) and (6) show that the coefficients of DT*Batch 2, DT*Batch 3, and DT*Batch 4 are significantly negative, and the coefficients of DT*Pilot and DT*Batch 1 are not significant, indicating that among the batches of environmental protection inspectors, the effect of pilot areas and the first batch of CEPI is poorer, and the second, third, and fourth batches of CEPI have more significant effects. Moreover, by comparing the coefficients of DT for each batch in columns (5) and (6), it is found that the effect of CEPI is increasing from the pilot batch to the third batch as the policy of CEPI continues to progress, as shown by the regression coefficients of − 0.0418 > − 0.049 9 > − 0.0572 > − 0.0574 in column (5) and − 0.0199 > − 0.

The environmental governance effect of CEPI
The objectives of CEPI include restraining the environmental violations of non-compliant enterprises and increasing their environmental governance investment. Accordingly, we examine the impact of CEPI on long-term corporate governance investment and short-term emission reductions at the firm level. We examine the regional environmental governance effects of CEPI at the regional level. We test the governance effects of CEPI at the enterprise level. First, we test the impact of CEPI on the longterm governance investment of enterprises. We test the impact of CEPI on corporate emissions costs by collecting and collating data on corporate emissions costs from disclosure reports of listed companies and using the logged corporate emissions cost indicator Pay. We test the impact of CEPI on corporate water and air pollution treatment expenditures by using corporate air treatment input and water treatment input data from the CSMAR database. Among them, the indicators of corporate pollution control inputs are obtained from CSMAR's environmental research database of listed companies, and the logged water pollution input costs Water_Pay and logged air pollution input costs Gas_Pay indicators are used as corporate pollution control input indicators. The regression results in columns (1)-(3) of Table 11 show that CEPI significantly increases the expenditure of corporate emission fees and increases corporate air pollution treatment inputs, but not water pollution treatment inputs.
Second, we test the impact of CEPI on firms' short-term emission reduction. In particular, information on wastewater (Water) and exhaust gas (SO 2 and COD) emissions is obtained from the CSMAR database. The regression results in columns (4)-(6) of Table 11 show that the CEPI significantly reduces SO 2 and COD polluting gas emissions of enterprises in polluting industries and does not reduce wastewater emissions. Third, we examine the impact of CEPI on firms' green patent applications. Among them, green patent application data are obtained from the CNRDS database, including green invention patent application (GreenInv) and green utility model patent application (GreenNew) data. The regression results in columns (7)-(8) of Table 11 show that CEPI significantly increases the number of green invention patent and green utility model patent applications and enhances the R&D level of enterprises in polluting industries.
The regression results in Table 11 show that there is heterogeneity in the environmental governance effects of the CEPI. For water pollution control, the effect of CEPI on the long-term treatment investment and short-term emission reduction of enterprises is not significant. For air pollution treatment, the effect of CEPI on long-term treatment investment and short-term emission reduction by enterprises is significant. It shows that CEPI in the treatment of air pollution does play a certain governance effect; there are certain defects in the treatment of water pollution. Therefore, the government should further deepen the supervision intensity of CEPI in water pollution management.
We examined the effectiveness of CEPI's governance at the regional level. We collected emission data such as industrial wastewater emissions, industrial SO 2 emissions, and industrial soot emissions, as well as patent data such as the number of patent applications, the number of patents granted, and the number of invention patents granted, for each region by compiling the China City Statistical Yearbook. After that, we use the logarithmic regional emission indicators Ind_Water/ Ind_SO2/Ind_Dust and the indicators Inv_App/Inv_Aut/ Inv_Ant of patent applications (total number of invention patents and utility model patents, number of invention patents, and utility model patents) to test the impact of CEPI on regional pollutant emissions. The results in columns (1)-(3) in Table 12 demonstrate that CEPI significantly reduces regional wastewater and exhaust emissions. The results in columns (4)- (6) show that CEPI significantly increases the number of patent applications and grants in the region. The results show that the environmental treatment effect of CEPI in the region is significant.

Conclusions
China's rapid economic development has also brought about many ecological and environmental problems, and the enforcement effects of existing environmental policies still need further evaluation. In this paper, we use the environmental policy of CEPI as a quasi-natural experiment and combine data from listed companies in polluting industries. From the perspective of environmental information disclosure, we use difference-in-difference estimation to examine whether CEPI mitigates firms' stock price crash risk by reducing the hoarding of negative information. This paper finds that CEPI reduces the firms' stock price crash risk in polluting industries in the region compared to non-inspector regions. The disclosure effect of CEPI is mainly found in the sample of private firms, firms with lower information transparency and disclosure quality, and firms in non-national civilized cities and regions with stronger incentives for officials to be promoted. This may be due to the fact that CEPI enhances the disclosure of negative environmental information in such samples, which inhibits the hoarding of negative information and thus reduces the stock price crash risk. CEPI mainly reduces the firms' stock price crash risk in polluting industries by reducing the stock price bubble. Among the batches of CEPI, the second, third, and fourth batches had better information disclosure effects, and the effect of each batch of inspectors gradually deepened. CEPI differs in the effectiveness of water pollution and air pollution treatment. CEPI increased the long-term treatment investment and short-term emission reduction effect of polluting industry enterprises in air pollution treatment, and the effect of inspectors in water pollution treatment was not significant. This paper examines the enforcement effectiveness of CEPI from the perspective of environmental information disclosure, using CEPI as a quasi-natural experiment. It not only helps to clarify the effectiveness of environmental governance after the enactment of environmental policies, but also extends the research on CEPI. It also deepens the regulators' knowledge and understanding of the linkages between the central government, local governments, and local enterprises. And thus, it has important practical implications for the promulgation and implementation of environmental policies.
Author contribution Mengyao Wen designed the experimental protocol, carried out the experiments, wrote the manuscript, revised the manuscript, and read and approved the manuscript.
Data availability Not applicable.

Declarations
Ethics approval Not applicable.

Consent to participate Not applicable.
Consent for publication Not applicable.

Competing interests
The author declares no competing interests.