Do environmental target constraints promote corporate pollution reduction?

In order to improve the environment quality, in 2007 the Chinese government implemented a policy document on environmental target constraints called the ‘Letter of Responsibility’. Based on this impact, we collect the environmental target constraints (ETC) data of 276 cities in China, and use the Differences-in-Differences (DID) method to evaluate the impact of local government ETC on enterprise pollution. The results show that ETC can significantly curb the pollution emission level of enterprises. This effect varies significantly between enterprises of different regions, scales, ownerships, and total factor productivities (TFPs). ETC can promote enterprise emission reduction by inducing green innovation, improving corporate tax burden and financing constraints, and promoting enterprise exit. Economic growth target constraints will weaken the inhibitory effect of ETC on corporate pollution. This paper provides important empirical evidence for deepening China’s official environmental assessment system and environmental governance.


Introduction
Since the reform and opening up, China has integrated into the global division of labor system, and achieved rapid economic development.In particular, China joined the World Trade Organization (WTO) in 2001 and has undertaken a large number of highly polluting global processing and manufacturing industries (Zhang et al 2011).Although this has promoted the industrialization process, the extensive development model of 'high pollution, high energy consumption and high emissions' has also brought serious environmental pollution problems (Zhang et al 2011, Zhang and Da 2015, Xue et al 2021).In addition, China is the world's largest emitter of greenhouse gases, accounting for more than a quarter of the world's greenhouse gas emissions each year, while greenhouse gases are the main cause of climate change.China's annual economic losses due to environmental pollution range from US$100 billion to US$300 billion, which seriously restricts economic sustainability (Ebenstein et al 2015, Zhang et al 2019, He et al 2020).As the main body of pollutant discharge, enterprises are the key link of environmental governance, and it is of great significance to conduct in-depth research on corporate pollutant discharge behavior (Du et al 2022).
In response to the growing environmental problems, China promulgated its first environmental law in 1979, marking the beginning of environmental protection management.For the first time, the Sixth Five-Year Plan included environmental protection as a separate chapter, showing the importance the government attaches to it.In 1998, the government approved acid rain and sulfur dioxide control zones (two control zones) to promote air pollution prevention.During this period, the comprehensive improvement and quantitative assessment of the urban environment began to be implemented, and a set of urban environmental management models with Chinese characteristics was initially established.However, due to the lack of effective restraint mechanisms and assessment methods, the improvement of environmental pollution under the 'two control zones' policy is only temporary (Jin et al 2016, Li et al 2016, Chen et al 2018a, Wang et al 2021).The reasons may be that in the context of China's unique decentralized political system and the 'GDP-only theory' as an indicator of official performance appraisal, local governments have launched fierce 'promotion tournaments' in order to compete for economic advantages (Li and Zhou 2005).This has led to local officials loosely implementing environmental regulation at the expense of the environment in exchange for regional economic growth (Wu and Cao 2021).The failure of the 'two control zones' policy led the central government to reconsider its environmental governance plan.
During the 11th Five-Year Plan period, the central government strengthened control over major pollutant emission targets and signed a 'Letter of Responsibility' with provincial governments.The 'Letter of Responsibility' is a binding agreement between the provincial government and the Ministry of Environmental Protection on environmental targets, including specific target limits and tasks for pollutant emissions, which is the essence of ETC (Chen et al 2018a).The signing of the 'Letter of Responsibility' marks the first time that the central government has included ETC in the evaluation indicators of officials at the local level, and has made environmental governance an important evaluation criterion for the promotion of officials (Yang et al 2021).Therefore, the signing of the 'Letter of Responsibility' in 2007 is a relatively formal exogenous shock.In the face of such shocks, local governments react differently to the ETC set by higher levels of government.Some regions have actively written ETC conditions into government work reports, and strengthened their binding nature through public commitments.Some regions do not explicitly put forward ETC conditions in the government work reports, leaving flexibility and room for maneuver.The difference in self-restraint caused by the inconsistent response of local officials to the emission reduction targets assigned by their superiors provides a basis for distinguishing between the experimental group and the control group.In China, the administrative contracting system is the norm in the Chinese government (Li and Zhou 2005).In the process of issuing indicators, it will be accompanied by a complex political game for the allocation of specific values of constraint indicators.But for local officials, this game has little impact on whether they will disclose their environmental assessment targets in government work reports.For micro enterprises in its jurisdiction, it is a relatively exogenous impact.The detailed distribution of cities is shown in figure 1.
ETC is a typical environmental regulation with Chinese characteristics, which requires local governments to achieve certain environmental goals within a certain period of time.However, in existing studies, the effect of ETC on corporate emission reduction is not clear, and the mechanisms behind it are also complex and diverse.Can this kind of environmental target constraint with typical Chinese characteristics effectively promote the reduction of pollution by enterprises?What are the possible mechanisms of its impact?How will the government's economic growth target constraints affect the ETC? Taking the signing of the Letter of Responsibility in 2007 as a proxy variable for ETC, this paper evaluates its impact on corporate emissions reduction.The contribution of this paper is that DID, Difference-in-differences-in-differences (DDD) and instrumental variable (IV) methods are used to effectively solve the endogenous problem between enterprise pollution discharge level and urban division, and improve the credibility of the estimation results.Secondly, this paper analyzes the heterogeneity effect between enterprises in different regions, scales, ownerships, and TFPs, and reveals the impact mechanism of ETC on different types of enterprises, which fills the gap of relevant literatures.Thirdly, this paper discusses the behavior of enterprises in response to ETC, and the impact of economic growth target constraints on ETC, which provides more perspective for understanding China's official environmental assessment system and environmental governance.Fourthly, this paper combines environmental policy with government target constraint behavior, and evaluate the impact and impact mechanism of local government environmental target constraint on enterprise pollution emission reduction from the enterprise dimension.
The rest of this paper is organized as below.Section 2 is the literature review.Section 3 introduces the theoretical analysis.Section 4 describes the methodology and data.The empirical results are shown in section 5.The extended analysis is shown in section 6.The last section shows the conclusions and policy recommendations.

Literature review
The direct effect of formal environmental regulation on corporate emission reduction is an important issue that scholars have focused on.Formal environmental regulation can significantly curb corporate pollution emissions (Hettige et al 2000).Greenstone et al (2021) found that concentrations of major air pollutants in China declined significantly after peaking in 2013.Guan et al (2022) found that during China's 11th Five-Year Plan, goal-based performance appraisal systems effectively strengthened local environmental governance.Scholars have also studied the impact of a specific environmental policy on corporate pollutant emissions, including two-control zone policy and smart city construction policy (Tanaka 2015, Chen et al 2018a).However, formal environmental regulation may have a green paradox effect.Wang et al (2018) found that the 'three rivers and three lakes' policy had no substantial impact on the chemical oxygen demand (COD) emissions of surviving enterprises.
In addition, scholars have also studied the impact of formal environmental regulation on firms' emission reduction through intermediate channels, focusing on how environmental regulation affects firms' production and innovation behavior, thereby indirectly affecting firms' emission levels.According to the 'neoclassical theory of economic development', environmental regulation will increase the cost of pollution control and curb corporate emissions by inhibiting production.This inevitably comes at the expense of economic development (Matthews and Denison 1981, Gray 1987, Gray and Shadbegian 2003).However, 'Porter's hypothesis' theory argues that appropriate environmental regulation can not only promote corporate green technology innovation, but also produce an 'innovation compensation effect', partially or completely offsetting the negative impact of cost factors, and thereby reducing corporate emissions through technological progress (Porter 1991).'Porter hypothesis' is the first to elaborate the possibility of a 'win-win' outcome between environmental protection and economic growth, which has attracted wide attention in the academic community (Qiu et al 2018, Ouyang et al 2020).There is no consensus on the applicability of 'Porter hypothesis' in China.The relationship between environmental regulation and firm innovation is uncertain, and the net effect of environmental regulation on technological innovation depends on the relative size of 'innovation compensation' and 'compliance cost' (Ambec et al 2013).There may also be a 'nonlinear' relationship between the two, such as an 'inverted U-shaped curve' (Marino et al 2019).In addition, based on the 'pollution paradise hypothesis', scholars believe that financing constraints (Cai et al 2016) and firm entry and exit (Liu et al 2022) are key channels for environmental regulation to affect firms' emission reduction.
Since the central government established environmental protection indicators as binding indicators for the performance evaluation of local officials, more and more scholars have paid attention to the potential benefits of etc Zhang et al (2019), Yang et al (2022) and Shi et al (2022) respectively examined the impact of energy target constraints on energy reduction intensity, resource allocation efficiency and enterprise energy efficiency in the 11th Five-Year Plan official performance appraisal target responsibility system.Zhang et al (2022) examines the impact of ETC on the marginal emission reduction costs of industrial enterprises by taking the emission reduction target constraints in the 11th Five-Year Plan as an explanatory variable.The actions of local governments are driven by the results of their mandated tasks (Chen et al 2018a).ETC takes environmental governance as a key evaluation indicator for official promotion, which can reshape the political incentives of local officials, make them pay more attention to environmental governance, and then have an impact on corporate pollution.However, research in this area is extremely lacking.In summary, the existing literature has made rich discussions on the relationship between environmental regulation and enterprise pollution discharge, but has not yet reached a unified conclusion.Moreover, few literatures have explored the causal relationship between local governments' autonomous constraint behavior and enterprise emission reduction from the perspective of ETC with typical Chinese characteristics.

Environmental target constraints and enterprise emission reduction
The actions of local governments are driven by the results of their mandated tasks (Li andZhou 2005, Li et al 2019).According to the 'Porter hypothesis', when the compliance cost caused by strict environmental regulations is higher than the cost required for enterprises to innovate green technology, enterprises will make strategic plans in advance to reduce production costs, and carry out green technology innovation to produce innovation compensation effect and achieve Pareto improvement (Porter 1991).Green technology innovation induced by environmental regulation can be divided into energy-saving and emission-reducing innovation and technological transformation innovation.The former innovation can improve energy efficiency and reduce pollution emissions per unit of output (Huang and Zou 2020).The latter innovation can promote the optimal allocation of production factors, improve the productivity, resource allocation level and technological innovation ability, and enhance the relative competitiveness and adaptability (Demirel and Kesidou 2011).
According to the 'pollution paradise' hypothesis, strict environmental regulation may have an 'industrial transfer effect' on pollution-intensive enterprises (Cai et al 2016).In order to avoid the rising cost of environmental compliance, highly polluting enterprises may move to countries or regions with a low level of environmental regulation (Copeland and Taylor 2004).In China, neighboring regions often resort to competing to lower environmental standards to attract companies in order to maintain their competitive advantage in the region (Ljungwall andLinde-Rahr 2006, Dean et al 2009).By raising the environmental protection threshold to drive away highly polluting enterprises and introduce cleaner production industries, the reallocation of resource factors and the reduction of pollution discharge levels can be achieved (Cheng et al 2018).The lack of external financing can constrain investment and development, thereby affecting their emission reduction levels and environmental performance.In order to cope with ETC, local governments may increase their disposable income by reducing tax costs, thereby encouraging enterprises to spend more money on green technology innovation and environmental governance.Therefore, this paper proposes the following hypotheses.
H1: Environmental target constraints will squeeze out highly polluting enterprises through cost-oriented effects, and attract clean enterprises to settle in, which in turn leads to the reallocation of resource elements.
H2: Environmental target constraints promote emission reduction by reducing the tax burden and financing constraints of enterprises, and forcing enterprises to engage in green innovation.

Source control or end-of-pipe treatment
The central government has particularly emphasized the importance of reducing emissions by enterprises, especially at source control.Although strengthening the source control of enterprises is more effective, it also faces the embarrassing situation of long cycle and large investment (Gutiérrez andTeshima 2018, Liu et al 2022).In the context of a 'promotion tournament', local governments have a strong incentive to encourage companies to strengthen end-of-pipe emission reductions in order to achieve performance in the short term (Hering and Poncet 2014).Therefore, this study proposes the following hypothesis: H3: The impact of environmental target constraints on corporate innovation behavior is still mainly to promote end-of-pipe treatment in the short term.

Moderating effect of economic growth target constraints
For a long time, China has placed economic growth at the heart of local government target management.Due to the need for political promotion, local governments often have the phenomenon of 'layering overweight'4 when setting economic growth targets, and use 'self-pressure' to transmit 'capability signals'5 to the central government (Li and Zhou 2005).Although this development model of 'competition for growth' and 'competition for investment' has promoted the rapid development of China's economy, it has also brought serious environmental pollution problems (Xue et al 2021).In order to obtain a greater probability of promotion, local governments have a strong incentive to complete economic growth targets (Li et al 2019), make strategic adjustments to environmental goals, and attract more polluting enterprises to promote regional economic growth, thereby weakening the inhibitory effect of ETC on enterprise pollution discharge (Jin et al 2005).Thus, this paper proposes hypothesis H4: H4: Economic growth target constraints will weaken the emission reduction effect of environmental target constraints on enterprises.

Models
This paper uses the signing of the Letter of Responsibility in 2007 as the background to examine the impact of ETC on corporate emission reduction.The Letter of Responsibility is a policy measure introduced by the Chinese government in 2007 to achieve the environmental goals of the 11th Five-Year Plan, setting specific limits and tasks for pollutant emissions in each province and implementing specific ETCs.The policy impact of the Letter of Responsibility varies from city to city, with some governments clearly stating digital targets for reducing industrial pollutants in their government work reports, while others have political targets.This provides a rare opportunity to use the DID method to identify the impact of ETC on corporate emission reduction.Therefore, this paper takes the cities that clearly put forward the targets of industrial pollutant emission reduction in the government work report as the experimental group, and constructs model (1).
LnPoll it indicates the pollutant discharge level of industrial enterprise i in the year t, measured by the natural logarithm of sulfur dioxide emissions from industrial enterprises (Chen et al 2018b).treat is a dummy variable, which is set to 1 if the location of the enterprise is in the experimental group, and 0 otherwise.post is a dummy variable, which takes the value of 1 when t 2007, and 0 otherwise.X it represents a series of control variables.This paper selects control variables at both the enterprise and city levels.Firm-level control variables include gearing ratio (Lev), firm scale, enterprise value, and enterprise age.City-level control variables include population density, economic development level, transport level, city scale (Popu), and industrial structure (Stru).g , i d t and z j are dummy variables that fix the individual, year, and binary industry fixed effects (FEs), respectively.m it is a random term.The coefficient a reflects the average impact of ETC on the company's emissions.

Data
Our data consist of the following three parts.The first part is the cross-database matching data of the Chinese industrial enterprise database and the Chinese industrial enterprise pollution emission database, from 2003 to 2013.The second one is the China Urban Statistical Yearbook.The third part is the environmental target constraint data and economic growth target constraint data which are manually sorted out and obtained according to local government work reports.The pollution emission data of China's industrial enterprises comes from the Ministry of Ecology and Environment, and its statistics are on industrial enterprises that account for more than 85% of the total emissions of each region.This paper accurately identifies enterprises, and matches the Chinese industrial enterprise database and the Chinese industrial enterprise pollution emission database according to the unique identification code formed by the identity information (Brandt et al 2012).Enterprises with less than 8 employees, or total industrial output value, fixed assets, total liabilities and main business income less than 0, or fixed assets greater than total assets, or opened earlier than 1949 are deleted.The definitions of variables and descriptive statistics are shown in table 1. shows the results of adding a series of control variables mentioned above, indicating that the overall explanatory power of the model has been enhanced.In addition, to further demonstrate the robustness of the results, the regression analysis is reperformed after replacing the fixed effect with Year-industry FE and Year-province FE in columns (3)-( 4).The coefficients of the interaction terms are significantly negative, indicating that the environmental target constraint has a significant inhibitory effect on the emission of enterprise pollution.ETC increases the incentive for local governments to improve the environment.Under the supervision of the central government and the public, local governments will increase pollution prevention and control, encourage green innovation, and introduce cleaner production policies to encourage enterprises to develop green technologies, improve production processes, and optimize energy structures, thereby reducing emissions.It is worth mentioning that a single sulfur dioxide index may not fully reflect the level of enterprise emissions, so this paper replaces the variable being explained with the pollution composite index and nitrogen oxides in the subsequent robustness test.

Parallel trend test
An important premise for causal identification using DID is that the pollution emissions of enterprises in the experimental group and control group must meet the assumption of parallel trend before the policy is implemented.If there is a difference in time trends, the source of emission reductions may be other factors rather than policy implementation.Therefore, this paper constructs model (2) based on the event research method.
d t is a time dummy variable of the t th year.In order to avoid the impact of multicollinearity, this paper takes the year before policy implementation as the base period.The regression results are shown in figure 2. Before the implementation of the environmental target constraint policy, b is not significant, indicating that there is no significant difference between the experimental group and the control group, and the parallel trend hypothesis is satisfied.After the implementation of the policy, the environmental target constraint has a significant inhibitory effect on enterprise pollution, and its impact is still deepening, indicating that the policy has achieved good implementation results, which further confirms the robustness of our results.The reasons may be that in order to gain an advantage in the 'promotion tournament', local governments have resorted to a series of short-term, quick-acting administrative interventions.

DDD method
Since the DID model requires strict premise assumptions, and requires that the signing of the Letter of Responsibility cannot affect the emission of enterprises in the control group, but it is difficult to meet in reality.In order to alleviate the endogenous problem, this paper further uses the DDD model for analysis.Since the cleaner production industry is hardly affected by the Letter of Responsibility, this paper divides enterprises into cleaner production industry and non-cleaner production industry (Akbostanci et al 2007).Model (3) is dirty is a dummy variable, which equals 1 when the enterprise is pollution-intensive, and 0 when the enterprise is cleaner production.The remaining variables are the same as in model (1).The regression results are shown in table 3. a are significantly negative, consistent with the benchmark regression results, which further verifies the robustness of the results.It is worth noting that the above setting does not completely overcome the endogeneous problem.Therefore, the conclusions will be further tested by instrumental variable method.

Placebo test
To exclude the interference of random factors on the effect of policy assessment, a placebo test is used to verify the robustness of the results (Abadie et al 2010).In this paper, cities with the same number as the original experimental group are randomly selected as the 'pseudo-experimental group', and the model (1) is re-estimated with 2007 as the policy intervention node.The coefficient of Post * Treat are recorded, and the above operation is  repeated 500 times.Figure 3 shows the results of the placebo test.Most of the landing points are far away from the benchmark regression results, indicating that the probability of obtaining the above benchmark regression coefficients based on random samples is low, which further verifies the robustness of the results.

Replace the explanatory variables
In order to avoid the problem of insufficient confidence in the regression results caused by the single measurement method of the explanatory variable, the robustness test is performed by replacing the explanatory variables with the composite pollution index (Lne) and the natural logarithm (LnNO) of NOx emissions (Popp 2006).The construction process of the comprehensive pollution index is as follows.
Firstly, the three main pollutant emissions of enterprise SO 2 emissions, wastewater discharge and soot emissions are standardized.pol eit indicates the emission of pollutant e produced by the enterprise i in the period of t. max and min represent the maximum and minimum emissions of pollutants per unit of output of all enterprises each year, respectively.We calculate the adjustment coefficient of each pollutant in enterprises by using the ratio of the standardized emission of each pollutant to the average level of each pollutant emission output of all enterprises in the country.
rpol eit is the average amount of pollutant emissions per unit output of all industrial enterprises in the industrial enterprises database.According to the standardized value and adjustment coefficient of each pollutant discharge, the enterprise pollution emission intensity index is constructed, and the natural logarithm of the index is taken.The higher the value of Lne, the greater the pollution emission intensity of the enterprise.
The regression results are shown in table 4. Column (1) reports the results of replacing the explanatory variables with the composite pollution index.Column (2) shows the results of replacing explanatory variables with NOx emissions.The coefficients are significantly negative, indicating that the environmental target constraint significantly reduces the emission level of enterprises, which is consistent with the benchmark regression results.

Advance policy intervention time
In order to exclude the influence of temporal stochastic factors on the results, this paper tests the robustness of the policy intervention time in advance.This paper brings forward the policy intervention year to 2006 and 2005 and uses model (1) for regression.The results are shown in columns (3)-( 4) of table 4. The coefficient of Post * Treat is not significant after the advance policy intervention time, indicating that no other policy interferes with the results before the implementation of etc.This verifies the robustness of our results.

PSM-DID test
In reality, policy is often a non-randomized experiment, and self-selection bias is inevitable.For example, some regions may be classified as experimental groups due to energy-rich and pollution-intensive industry agglomeration, resulting in excessive differences between some samples in the control group and the experimental group.Therefore, the propensity score matching (PSM) method is used for further calibration in this research.This method can match each experimental group sample to a specific control group sample, making the sample more accurate.Seen from figure 4, the absolute value of the standardization deviation of each variable after matching is significantly reduced, indicating that the data has a good matching effect.Column (5) of table 5 reports the results after PSM treatment.The coefficient of the interaction term is still significantly negative at the level of 1%, indicating that the difference between samples has no serious impact on the results.This verifies the robustness of our results.

Eliminating policy interference
In order to eliminate the impact of potential factors, this paper analyzes other competitive policies or historical shocks that may affect corporate emission reduction.The implementation of the SO2 emission trading pilot policy implemented in 2007 also relies on local officials for its assessment objectives, which may have an impact on the identifying of net effect of environmental target constraints on export products quality.In the third  'Environmental Assessment Storm' operation in 2007, the State Environmental Protection Administration implemented 'regional and watershed approval restrictions' in some regions, while the environmental protection departments of some provinces also implemented regional approval restrictions in some areas within the province, resulting in stricter enforcement and punishment of environmental regulatory policies in these areas, which may have an impact on the results of this paper6 .Table 5 reports the results of regression analysis after removing these two policy interferences.ETC still has a significant inhibitory effect on corporate pollution behavior, indicating that competitive policies and historical shocks have not had an impact on the results of this paper.

Endogenous testing
When setting up environmental goals, local governments will tend to consider their own development.In order to ensure rapid economic development, local governments also have different attitudes towards environmental governance.Therefore, the selection of the experimental group may not be completely exogenous.From the perspective of physical geography, the number of large rivers flowing through each province in China is selected as the instrumental variable (IV).The regression results are shown in table 6.The Cragg-Donald Wald F-statistics value of the weak IV test is 95.89, which is greater than the critical value of 16.38, indicating that the IV does not have the problem of weak IV selection.Kleibergen-Paap Lagrange multiplier (KP-LM) statistic-P and Hansen J statistic-P are less than 0.05, which rejects the null hypothesis, indicating that the IVs are exogenous and the model does not have insufficient recognition.The results of the first stage show that there is a significant positive correlation between whether prefecture-level cities include environmental targets in government work reports and IV.It is consistent with the above expectations.The results of the second stage show that after considering the potential endogenous problems, ETC can still significantly inhibit corporate emissions, which is consistent with the benchmark regression results.
5.4.Heterogeneity analysis 5.4.1.Ownership heterogeneity Under China's economic system with public ownership as the main body, there are great differences between enterprises of different ownerships.Due to complex political relationships, there may be a 'human relationship' effect when governments set emission reduction targets for companies.State-owned enterprises with political ties can receive more policy preferences than non-state-owned enterprises with weak political ties, and then focus more on corporate performance (Hering and Poncet 2014).Therefore, this paper divides the sample into state-owned enterprises, foreign-funded enterprises, private enterprises and collective enterprises for group regression.The results are shown in columns (1)-( 4) of table 7.In state-owned enterprises, foreign-funded enterprises and private enterprises, ETC has a significant inhibitory effect on enterprise emissions.Foreign investors including Hong Kong, Macao and Taiwan holdings have the greatest inhibitory effect and the strongest significance.The possible reason is that these enterprises have more technology accumulation, and can play a rapid role in green technology transformation to significantly reduce pollution emissions.Although the collectively owned enterprises are not significant, the coefficient is also negative.

Geographical heterogeneity
China is a vast country, and there are great differences in economic development level, natural resource endowment, humanistic and environmental awareness among different regions (Xu and Lin 2019).Therefore, this paper divides the sample into three regions including eastern, central, and western regions.The regression results are shown in columns (5)-( 7) of table 7. The environmental target constraint significantly reduces the pollution level of enterprises in the central and eastern regions.The three samples also show significant spatial heterogeneity, with the largest abatement effect in the eastern region and lower in the central and western regions.The reason may be that the eastern region has a high economic level and talent advantages, and the agglomeration effect of various resources is more obvious, and it can quickly achieve emission reduction by virtue of its advantages of economies of scale and resource allocation.The results also reflect the great contributions of the eastern region to industrial emission reduction.

Scale heterogeneity
ETC will lead to increased operating costs for enterprises and raise the threshold for market access in polluting industries.Due to the existence of economies of scale, large-scale enterprises often have a high level of technical reserves and capital flow advantages, and can quickly optimize their production processes in a short period of time to reduce pollution emissions, and can more easily cope with the increase in environmental compliance costs.This paper takes the total assets of enterprises as a classification criterion, and divides enterprises into large-scale and small-scale samples according to the median, so as to analyze the heterogeneity.The results are shown in columns (8)-( 9) of table 7. The emission reduction effect of large-scale enterprises is more obvious and significant, which further verifies the above theoretical analysis.

TFP heterogeneity
TFP reflects the technical level, management efficiency, and scale effect of the enterprise, so there are significant differences in production efficiency and innovation ability of different TFP enterprises.Therefore, this paper measures the TFP of enterprises, and divides enterprises into two groups of high TFP and low TFP based on the median to analyze their heterogeneity (Levinsohn and Petrin 2003) (LP method).The results are shown in columns (10)-( 11) of table 7. Enterprises with high TFP have a stronger emission reduction effect, which is in line with expectations.Enterprises with high TFP tend to have better human capital level, technology accumulation, R&D and innovation capabilities, and stronger ability to absorb and innovate new technologies.Enterprises with high TFP often have strong market competitiveness and have the ability to modernize equipment to achieve emission reduction targets.

Green innovation
The mediation effect model is used to test whether ETC reduces pollution levels by promoting green technology progress, and models are constructed in model ( 7), ( 8) and (9).The level of green innovation is measured by the number of green patent applications in prefecture-level cities (city Green _ ).Effluent levels are measured by the natural logarithm of total industrial SO 2 emissions (citylso2).The results are shown in table 8. Through the results of model (8), ETC significantly promote the progress of green technology.When city Green _ is added to model (9), technological innovation has a significant negative impact on pollution levels, indicating that technological progress and innovation by industrial enterprises have indeed reduced SO 2 emissions.The coefficient of ETC is reduced compared with model (7), which satisfies all the conditions of the mediation effect.ETC promotes companies to reduce pollution by inducing green innovation, so the Porter hypothesis is verified.

Corporate tax burden and financing constraints
Since the improvement of the external business environment will eventually be reflected in the operation of enterprises, this paper uses the method of direct regression of key explanatory variables to mediating variables for empirical testing (Jiang and Bai 2022).The natural logarithm proportion of corporate income tax to the output value of industrial sales in the current year (lnssys) is chosen to measure the level of corporate taxation.The natural logarithm of proportion of accounts receivable to industrial sales output value in the current year (lnrzys1) and the natural logarithm of proportion of corporate accounts receivable to the sales revenue of main products in the current year (lnrzys2) are taken as the proxy variables of enterprise financing constraints.All the above variables are reduced by 0.5% tail treatment.From the results in table 9, the coefficients of the interaction terms are significantly negative, indicating that the environmental target constraint significantly promotes the government to provide green subsidies and tax incentives to enterprises, and reduces the tax burden and financing constraint level of enterprises.Moreover, the effect of corporate tax incentives and financing constraints on corporate emission reduction has been verified (Zhang andVigne 2021, Yu et al 2022).

Exit effect of enterprises
Subject to pressure from the central government, local officials are likely to adopt a 'one-size-fits-all' approach to quickly complete assessment goals for the sake of their own career development.This raises the barrier of the entry for some highly polluting industries, and will also force some companies with less backward technology and less resilience to withdraw or relocate.In this paper, only companies that have been tracked continuously for more than 10 years during the sample period are retained, and the balance panel data is constructed to examine whether ETC promotes the exit or relocation of enterprises.It is worth noting that after constructing the data as balanced panel data, the median size of enterprises increased from the original 54101 thousand yuan to 118681 thousand yuan, indicating that many smaller enterprises with relatively backward clean production capacity may be forced to withdraw from the market due to strict environmental regulations, and thus cannot continue to exist during the sample period.As a number of polluting enterprises withdraw from the market, clean enterprises will take on more production tasks, which will lead to the reallocation of factors in the market.The results are shown in table 10.Not only the coefficients of the interaction items are significantly negative, but the absolute values are also larger than the results in the benchmark regression.The reasons may be that companies that survived in the sample period have strong financial strength, technical reserves, and risk resistance, and the exit of some enterprises has also increased their profits, so they are more capable of transforming to cleaner production.
6. Further analysis .Although the coefficient of source control is negative, it is not significant.The results of end-of-pipe treatment are significantly positive, indicating that after the implementation of ETC, local governments implement a series of short-term and fast-achieving end-of-pipe treatment measures to rapidly improve local environmental performance.

Moderating effects of economic growth target constraints
In order to explore the impact of the conflict between economic growth targets and environmental objectives on the emission reduction effect of enterprises, this paper uses the economic growth target data of city government work report in 2003-2013.We add the interaction term of economic growth target constraint (grow) and environmental target constraint in the econometric model to investigate the moderating effect of economic growth target constraint on environmental target constraint.The results are shown in column (3) of table 11.
The coefficient of the triple interaction term is significantly positive at the level of 1%, meaning that local governments have relaxed their controls on environmental governance while facing constraints on economic growth.Economic growth targets have an inhibition effect on environmental governance.

Conclusions and recommendations
This research collects the environmental target constraint data of 276 Chinese cities, and uses the DID model to study the emission reduction effect of etc The results show that the environmental target constraint significantly reduces the pollution discharge level of industrial enterprises, and the end-of-pipe treatment is still the main factor in emission reduction.Foreign-invested enterprises, eastern enterprises, larger enterprises, and high TFP enterprises have stronger emission reduction effects.ETC can force companies to green innovation.Under the pressure of achieving economic growth targets, local governments may attract more polluting enterprises to promote regional economic growth, thereby weakening the inhibitory effect of ETC on corporate pollution emissions.
Based on our findings, we suggest the following policy implications for improving China's environmental regulation and governance.The central government should deepen the reform of the target responsibility system and improve the official appraisal and evaluation system.Simply using GDP as an assessment indicator may make local governments more inclined to adopt short-term strategies for pollution control, such as promoting the entry and exit of enterprises and strengthening terminal governance.The central government should design more comprehensive and balanced indicators to assess the environmental performance of local governments, such as green GDP, ecological compensation, and public satisfaction.The government should control the discharge of pollutants at the source and the generation of pollutants in the production process.Since ETC have a limited effect on source control, which is more effective and efficient than end-of-pipe treatment in reducing pollution, local governments should actively introduce corresponding tax policies, give special subsidies to enterprises with relatively tight finances for source control, and encourage enterprises to adopt cleaner production technologies, optimize energy structures, and improve production processes.Local governments should fully consider the affordability of different heterogeneous enterprises, give more technical and economic support to growing small and medium-sized enterprises, and avoid the 'one-size-fits-all' governance ways.Since ETC have different effects on enterprises of different regions, scales, ownerships, and TFPs, the government should tailor its policies according to the characteristics and needs of different types of enterprises, and help them overcome the barriers of green innovation and emission reduction.Local governments should leave room when setting economic growth targets to avoid environmental constraints caused by too high economic growth targets.Local governments should balance the two objectives and avoid pursuing economic growth at the expense of the environment.They should also consider the long-term benefits of environmental improvement for economic sustainability and social stability.

6. 1 .
Source control or end-of-pipe treatmentIn this paper, the SO 2 production amount per unit of total industrial output value ( ) of end-of-pipe treatment.The regression results are shown in columns (1) and (2) of table10

Table 1 .
Descriptive statistics of variables.

Table 2 .
Benchmark regression results.We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** indicates significance at 1%.

Table 5 .
Eliminating policy interference.We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** indicates significance at 1%.
*We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** and ** indicate significance at 1% and 5%, respectively.

Table 7 .
Heterogeneity analysis.We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** , ** , * indicate significance at 1%, 5%, and 10% levels, respectively.

Table 8 .
Mechanism test of green innovation effect.

Table 9 .
Mechanism test of tax burden and financing constraint.

Table 10 .
Mechanism test of enterprise exit.We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** and ** indicate significance at 1% and 5% levels, respectively.

Table 11 .
Further analysis results.We report robust standard errors in parentheses and cluster robust standard errors at the individual level in this table.*** indicates significance at 1% level.Fund Talent Launch Project of Zhejiang Agricultural and Forestry University: 2021FR050, 2023FR003; Project of Zhejiang Province Research Center for the Xi Jinping Chinese Characteristics Socialism Thought in the New Era: 22WH70066-8Z; the Fundamental Research Funds for the Provincial Universities of Zhejiang: W20220052; Zhejiang Province Philosophy and Social Science Planning Leading Talent Cultivation Project 24YJRC09ZD and 24YJRC09ZD-2YB.