Price Mechanism, Government Constraints and Carbon Trading Pilot Policy for Emission Reduction

Based on the data of 247 cities at the prefecture level in China from 2007 to 2019, this paper analyzes the impact of the carbon emissions trading (CET) pilot policy on carbon emission reduction from the perspective of the price mechanism and government constraints. The results show that the carbon emissions and carbon intensity in the pilot areas are significantly reduced by adjusting the industrial structure and promoting green technology innovation. In terms of regions, the emission reduction effect of the pilot policy in regions with a high proportion of industry is obviously weaker than that in other regions. The aim of the carbon emission trading policy in China that achieve carbon emission reduction is by coordinating the carbon emission trading price that fail to fulfill this aim independently and the degree of government punishment for enterprises.


INTRODUCTION
Since the 20 th century, human economic activities have accelerated global warming and seriously threatened the ecological environment and human health both physically and psychology.Since the reform and opening up, China's economy has witnessed rapid development in economy, while bringing serious environmental pollution.Research shows that imperative environmental regulation policies Existing research has provided a great deal of discussion on this topic, but no consensus has been reached.The Porter Hypothesis [3] holds that appropriate and reasonable environmental regulation policies can achieve emission reduction while maintaining economic growth.The carbon market based on the Coase theorem provides a realistic case which scholars can test this hypothesis.The carbon market internalizes pollution discharge, which increases the production cost of enterprises and reduces carbon emissions through a reverse mechanism, which has been widely verified in developed countries such as European and American countries [4,5].Most of research on the pilot policy of carbon emission trading in China foucus on the carbon emission quantity [6,7], regulatory industry spillover effect [8], company value [9], carbon emission efficiency [10], green technology innovation [11,12] and coordinated emission reduction [13] in the pilot area.There are still some controversies surrounding the path through which China's carbon market seeks to emission reduction.Some studies have found that the carbon market can reduce carbon emissions [14] by promoting the green innovation of enterprises, industrial transformation and energy structure adjustment, thus achieving the dual purposes of economic growth and environmental protection.Some suggest that the carbon pilot policy has a "pollution paradise" effect, leading to enterprises moving out from stricter regulation areas to lenient one and slumping regional economic growth [15], and that the quota allocation method based on historical law indirectly subsidizes polluting enterprises, thus causing them can not to achieve carbon emission reduction [16].In addition, some scholars have also evaluated the effectiveness of the market from the aspects of the carbon trading price, market liquidity, quota turnover frequency and information transparency and find that the carbon market in seven pilot areas of China has not achieved even weak efficiency, the market has a phenomenon of "emphasizing contract rather than trading" [17], and the imperfect trading system and illiquid market hinder the carbon market.
In summary, existing research has not reached an agreement on the mechanism through which carbon pilot policy can reduces emissions, and there are few empirical tests on whether this policy produces a "pollution paradise" effect and reduces economic growth.At the same time, there are few studies on the most important attributes of the carbon market price incentive mechanism and the mechanism of government constraints in terms of reducing emissions.Therefore, from the perspective of price incentives and government constraints, based on panel data from 247 cities in China from 2007 to 2019, this paper concludes that the carbon market, as a market-oriented environmental regulation policy, can achieve a win-win situation in China and empirically tests the role of the carbon trading price and government punishment in emission reduction, enriching the research on the carbon market in China.

LITERATURE REVIEW AND RESEARCH HYPOTHESIS
Previous studies have focused mainly on the following three aspects.1) Carbon market for economic development and environmental protection.For a long time, the problem of environmental governance was solved by the government mainly through administrative intervention [18].However, the administrative environmental regulation policy not only inefficient and unsustainable but also causes a series of problems, such as rent-seeking [19].In the case in which traditional administrative environmental regulation becoming increasingly worse, developed countries such as European and American countries take the lead in putting forward market tools to internalize external environmental governance issues through price mechanisms, such as the carbon tax and carbon emission trading market [20].At this stage, China is still vigorously reforming its economic system, and whether the market-based environmental regulation policy can effectively reduce emissions while maintaining economic growth needs further verification.On the one hand, China, as the largest developing country in the world, is still in the process of urbanization and industrialization and has a high level of demand for fossil energy consumption.The carbon pilot policy increases the energy use cost of enterprises in pilot areas and reduces the profit margin of enterprises; Thus, enterprises choose to reduce the current output to meet the requirements of emission reduction.Because the carbon pilot policy has increased the environmental binding force of the pilot areas, some enterprises choose to move from these pilot areas to nonpilot areas, which has a "pollution paradise" effect and reduces the economic growth of the pilot areas.On the other hand, the carbon pilot policy can increase the degree of R&D investment of enterprises through a "reverse mechanism", improve the total factor productivity of enterprises, and bring about new growth momentum to the region through the development of green industries.For this, this paper suggests: Hypothesis 1: The carbon pilot policy significantly reduces the amount of carbon emissions in the pilot area while maintaining economic growth, that is, reducing the carbon intensity.
2) Emission reduction mechanism of the carbon market.Economic restructuring and green technology progress are two important channels through which a country can achieve the goals of energy conservation, emission reduction and low-carbon economic transformation, among which the rationalization of the industrial structure is most important [21].Since the 21st century, the acceleration of industrialization has been one of the main reasons for the rapid growth of carbon emissions.The carbon pilot policy internalizes pollution externalization, and the "three-high" (high-pollution, highenergy-consumption and high-emission) enterprises with relatively weak emission reduction capacity will face more severe survival pressure, forcing enterprises to pay attention to energy conservation and emission reduction in the production process and actively carry out the structural adjustment of products.In addition, as a market-oriented environmental regulation tool, the carbon pilot policy affects the behaviors of enterprises, such as resource allocation, R&D and investment [22].Some scholars believe that the carbon pilot policy increases the production cost of enterprises and has a crowding-out effect on investment in green innovation.For example, Chen and Zhang [10] believe that the market-based environmental regulation policy has no significant impact on the green innovation of reducing the level of pollution emissions.In contrast, some scholars have explained that the carbon pilot policy has a significant effect on increasing green investment in pilot cities from the perspective of profit [23].The carbon pilot policy reduces the uncertainty of enterprises' investment in the field of green technology, weakens those concerns about innovation risks, and drives enterprises' innovation in the field of green technology [24].In view of this, this paper suggests: Hypothesis 2: The carbon pilot policy will reduce emissions through industrial restructuring and the improvement of green innovative technologies.
3) Influence mechanism of price incentives and government restraint mechanisms on the emission reduction effect.The carbon emission trading market needs a reasonable price formation mechanism and appropriate level of government management to achieve emission reduction throughout the market [25].As shown in Figure 1, many factors affect the operation of the carbon market, such as the carbon emission accounting mechanism, number of initial quotas, quota distribution, degree of marketization, formation of the price mechanism, access mechanism, and laws and regulations.In practice, there is generally imperfect competition in the carbon market.A few enterprises may purchase and store carbon emission rights far exceeding their own emission quotas to seek monopoly benefits or use them in the future, which may lead to a decline in market liquidity as well as other consequences [26].Thus, an important prerequisite for carbon pilot policies is that they reduce carbon emissions to build a carbon emissions trading market that is close to perfect competition under reasonable government control.Through the perfect market-oriented mechanism, the factors have successfully shifted from departments with a low utilization rate to those with a high utilization rate, thus realizing the most effective allocation of resources.Compared with European and American countries, carbon pilot work in China started late and is still in the groping process from pilot to comprehensive promotion.A carbon market mechanism has not been effectively established, and some problems, such as the lack of a functional market, unreasonable quota allocation and lagging legislation, exist [27].As shown in Figure 2, in each pilot city, the carbon market shows the phenomena of a smaller price fluctuation, a larger daily trading volume and a higher degree of market liquidity near the settlement period because, when the market mechanism has not been fully established, emission control entities are under pressure to reduce emissions, showing the characteristics of performance-driven trading rather than being based on their own costs.Considering the importance of the Chinese government in environmental governance, this paper suggests: Hypothesis 3: Price incentives and government constraints can achieve emission reduction through coordination, and government constraints play an important role in the case that the carbon market mechanism is not yet perfect.
In general, although more and more studies have begun to focus on China's carbon market, and scholars have studied it from multiple dimensions and achieved rich research results, on the one hand, there are still divergent empirical conclusions on the carbon emission reduction effect of the carbon market, and on the other hand, there are no studies to clarify the relationship between the inefficient  carbon trading market mechanism and the carbon emission reduction effect.This paper has the marginal contribution in the following three aspects.First, by examining carbon emissions at the prefecture-level city level, it is found that market-based environmental regulation can achieve a win-win situation in China's economic growth and environmental performance.Second, this paper uses the data of 247 cities in China for empirical analysis, and the application of this data set is helpful to discover some problems and conclusions that cannot be found from the provincial level research, which has certain value for expanding the theoretical research in this field; Last, innovative research perspective, considering that the government still plays a key role in environmental governance in the process of China's market economy transformation from the perspective of price incentives and government constraints, this paper empirically examines the role of carbon trading price and government punishment in emission reduction.

Model construction
Among the methods through which to evaluate the effect of policy implementation, the differencein-differences method is a widely used econometric method.The basic idea of this method is to regard institutional change and new policy implementation as "natural experiments" or "quasi-experiments" that are exogenous to the economic system.Because the carbon pilot policy meets the meaning of exogenous in both time and space a , China's carbon pilot policy can be regarded as a "quasi-natural experiment", so that the difference-in-differences method can be used to evaluate the policy effect.
Because the times at which the carbon market is initiated across pilot area are different, this paper uses the multi-period difference-in-differences method to estimate the impact of pilot policies on the levels of regional carbon emissions.To avoid the bias of missing variables for coefficient estimation, this paper adopts fixed effects, the corresponding model of which is as follows: where subscripts i and t represent regions and years, respectively.Y it denotes the carbon emission quantity and carbon intensity of the region.DID is the core explanatory variable: DID = treatment i × post t, where treatment i represents whether or not the city is a pilot city.Post t represents the policy implementation time.Control it represents the control variable that affects the change in the level of carbon emissions across time and cities; γ i represents urban fixed effects, which control for the influencing factors that change with individuals but not with time.γ t represents time fixed effects, which control the factors that affect carbon emissions that change with time but not with cities. γ ir represents the interaction between region and year, which controls for the factors that change with time.ε it represents the error term.
a Although the General Office of the National Development and Reform Commission issued a notice on the pilot work of carbon emissions trading in 2011, the specific implementation time of the pilot policy is still uncertain.In addition, the selection of pilot cities is decided by the State Council, and local governments cannot influence the decision of the central government through their own efforts; thus, the problem of sample self-selection is negligible.
The value rules of treatment i are as follows: when i represents Beijing, Tianjin, Shanghai, Chongqing and prefecture-level cities belonging to Guangdong, Hubei and Fujian, the value is 1; when i represents other regions, its value is 0. Considering the specific time point at which to initiate the carbon market b , the rules for the value of post t are as follows: when the region represents Beijing, Tianjin, Shanghai and the prefecture-level cities belonging to Guangdong and t ≥ 2013, the region represents Chongqing and the prefecture-level cities belonging to Hubei and t ≥ 2014, or the region represents the prefecture-level cities belonging to Fujian and t ≥ 2016, the value is 1.Otherwise, the value is 0. If the carbon pilot policy significantly reduces the amount and intensity of carbon emissions in pilot areas, then β 1 is significantly negative.

Explained variables and core explanatory variables
The explained variables in this paper are regional carbon emissions and carbon intensity, and the core explanatory variable is whether to initiate the carbon market, that is, the DID variable.The main variables are described in Table 1.Because carbon emissions are closely related to economic development and energy dependence, referring to Chen [28], to make the carbon emissions of the treatment and control groups comparable, the heterogeneity characteristics that may affect whether the area becomes a pilot project are controlled.In this paper, the control variables of the characteristics of economic development and environmental dependence in relevant regions are presented below.
The level of economic development is denoted by the logarithmic values of regional gross domestic product (GDP) the square of regional GDP and the logarithm of the number of industrial enterprises above a designated size.According to the theory of the environmental Kuznets curve, an inverted U-shaped relationship may exist between the economic development level and amount of environmental pollution.When the level of regional economic development is low, it is necessary to help people improve their living standards through rapid industrialization and urbanization, which inevitably negatively affect the ecological environment.At this time, there is a positive correlation between the economic development level and degree of environmental pollution, and carbon emissions increase with the improvement of the economic development level.However, when the level of economic development is high, people need to solve environmental problems more urgently, and more social resources thus flow in the direction of energy conservation and emission reduction for green technology innovation.At this time, there is a negative correlation between the level of regional economic development and degree of environmental pollution, and carbon emissions decrease with the improvement of the economic development level.
The industrial structure is embodied in the proportion of the value added of the secondary and tertiary industries to GDP.At present, the low efficiency of energy utilization is an important reason for the serious environmental problems in China.The secondary industry is dominated by "three-high" enterprises, and the problem of the low efficiency of energy utilization is particularly prominent.The corresponding carbon emissions in areas with a high proportion of secondary industry are relatively high.
The economic structure is embodied in the ratio of the retail sales of social goods to GDP.Higher consumption levels during economic development further stimulate enterprise production, thus increasing the level of carbon emissions.The degree of opening to the outside world is embodied in the ratios of total imports and exports to GDP.Moreover, opening to the outside world produces a technology spillover effect, enhances the efficiency of regional energy utilization, and then reduces the level of carbon emissions.
The degree of economic agglomeration is embodied in the logarithm of population density and the per capita income level.People in areas with higher per capita income levels have stricter requirements for air quality and higher expectations for China as a whole.
Dependence on energy is embodied in the number of employees in the extractive industry per thousand people.In particular, energy richness cannot measure the dependence of a region's economic development on energy.The resource curse refers to the relationship between energy dependence and the environment.When a region has and uses many natural resources, factors of production and economic resources flood into the resource department, resulting in insufficient degrees of capital investment in other departments, a single economic structure and serious environmental pollution.This paper introduce the annual average transaction price of the carbon trading market as the proxy variable for the price incentive.When the market transaction price is high, the unit emission reduction cost faced by the covering subject is high.With reference to Liu and Woodward [13], this paper takes the administrative punishment of enterprises that fail to fulfill their obligations in the pilot areas as the proxy variable for the government restraint mechanism.c This paper introduces the following related variables for mechanism analysis and subsequent inspection.Industry refers to the ratio of the value added of the tertiary and secondary industries to GDP, the patents of the number of applications for green utility models in that year, the volume of the total annual regional transactions to carbon emissions in that year, and the degree of development of the factor market.

Data description
This paper uses panel data from 247 cities in China from 2007 to 2019 to evaluate the emission reduction effect of the carbon pilot policy.Among these data, those on the amount of carbon emissions come from the China Carbon Emissions Accounting Database (CEAD), those on the number of green patents come from the China National Intellectual Property Administration, those on the development degree of the factor market come from the China Marketization Index Database, and other data come from the China City Statistical Yearbook, China Energy Statistical Yearbook, China Regional Statistical Yearbook and EPS Database.Limited by the available data, we can obtain data on the price of carbon trading and the development degree of the factor market only at the provincial level.Regional GDP and per capita output are adjusted to 2007 as the base period.
The whole data process and regression outcomes are conducted by stata17.0 software.c Specific values: Beijing, Shanghai and Shenzhen take 6, Chongqing takes 5, Hubei takes 4, Guangdong takes 3, Fujian takes 2 and Tianjin takes 1.

Benchmark regression result
Table 2 lists the benchmark regression results of formula (1), and all columns report the clustering standard errors at the provincial level.Column (3) of Table 2 shows that the estimated coefficient overestimates the impact of the carbon pilot policy on carbon intensity when the control variables are not included.The possible reason for this is that in the pilot area, there are accelerated emission reduction measures in addition to the carbon pilot policy, which is not the case in non-pilot cities, which leads to errors in the estimated coefficient due to missing variables.From columns (2) and ( 4), it can be seen that the regression coefficients of carbon emissions on the carbon pilot policy are significantly negative at 1%, at -0.133 and -0.176, respectively.After adding control variables and fixing the interaction items for city, region and year, the carbon market is shown to significantly reduce the amount and intensity of carbon emissions in pilot areas, which is similar to the emission reduction effect identified in previous studies.This result shows that there is no "pollution paradise" phenomenon in the pilot area because due to the characteristics of bank lending in China, enterprises need to re-establish contact with local banks and government officials after moving out of the area, and this cost is far greater than that brought about by the carbon pilot policy.Among the control variables, the per capita income level is significantly negative at 1%, which shows that with the increase in the per capita income level, people pay more attention to environmental pollution.The logarithm of the square of regional GDP is significantly negative at 5%, which is in line with our expectation of the environmental Kuznets curve; that is, with the increase in regional GDP, the amount of carbon emissions first increases and then decreases.

Parallel trend test and policy dynamic effect analysis
The key premise of the double-difference model is the parallel trend hypothesis; that is, before the implementation of the policy, the carbon emission change trends of the pilot and nonpilot areas should be the same.In addition, the results of the benchmark regression are the average impact of carbon pilot policies on carbon emissions, which does not reflect the heterogeneous results over time.Referring to the research framework of Jacobson and LaLonde [29], this paper tests the parallel trend assumption and analyzes the dynamic effects of policies based on event analysis.In this paper, the sixth year (2007) before the formal implementation of the carbon pilot policy is taken as the reference year, and the interactive items of the virtual variables of the years before the implementation of the carbon pilot policy and the virtual variables of the corresponding policies are put into equation (1) for regression.The regression results are as follows: In formula (2), s represents the time when the policy is implemented, and the value before 2013 is negative; then, the value is positive, and the value in 2013 is 0. DID S represents the dummy variable in the s year when the policy is started.The meanings of the other symbols are the same as those in formula (1).
The regression results show that before the implementation of the pilot policy, the coefficient is not significant (Figure 3), which shows that the change trends of carbon emissions in the pilot and nonpilot areas are basically the same, and a parallel trend assumption is established.After the implementation of the policy, the coefficient is significantly negative and tends to decrease with the passage of time, indicating that the effect of the policy is slowly expanding with time, which is in line with the promotion of the pilot policy in China and effectively responds to the suggestion that the pilot carbon policy in China has only a short-term emission reduction effect [30].The long-term policy may be effective because of the learning effect of local governments and the judgment of enterprises on the sustainability of policies.In the process of carrying out the pilot policy in China, the central government asks each pilot area to learn from the other and, through trial and error in each area, forms an experience that can be popularized.When it is found that the policy effect in a pilot area is good, the central government sends the heads of other areas to the local organization to study and commend them through meetings; thus, incorrect policy implementation is corrected, and effective behaviors are promoted and implemented.In addition, the effectiveness of environmental regulatory policies is closely related to the expectations of enterprises for such policies [31].When an enterprise expects future policy to be more relaxed, it will not invest in energy-saving and emission reduction technologies but will choose short-term production reduction, and the environmental policy will be effective only in the short term.When the enterprise expects the policy to continue, it will change the direction of its production investment and increase its investment in green production technology, which is effective in the long run.Note: The values in brackets are the clustered standard errors at the provincial level, and *, * * and * * * represent significance levels of 10%, 5% and 1%, respectively.

Placebo test
To enhance the reliability of the benchmark regression results, this paper uses a non-parametric replacement test to conduct a placebo test [32].Specifically, all provinces and policy time points are sampled without repetition, and 8 provinces and corresponding policy time points are selected at a time.Eight provinces are selected as virtual experimental groups, and the remaining areas are used as virtual control groups.Random sampling ensures that the core explanatory variable DID that constructed in this paper has no impact on carbon emissions.This process is repeated 500 times, and then, the DID regression coefficients of interaction between 500 virtual processing groups and virtual policy points are obtained.According to the kernel density diagram of the kernel explanatory variable DID (Figure 4), the estimated coefficient is concentrated around 0, which shows that the coefficient is not significant and indirectly shows that the possibility of some variables missing in formula (1) is small.At the same time, the true estimates (-0.147, -0.176) in this paper are located at the low tail of the kernel density map, which shows that the benchmark regression results are robust.

Other robustness tests
Benchmark regression results show that the carbon pilot policy has achieved an effective level of emission reduction, but a series of tests are still needed to confirm the robustness of the results.To ensure the robustness of the research results, this paper conducts a multidimensional analysis by excluding the interference of other important policies during the research sample period and screening the sample time.
The interference of other policies must be eliminated, as other policies also affect carbon emissions during the sample period, resulting in bias in the benchmark regression results.To solve the increasingly serious environmental problems, China has adopted numerous policies to control the atmosphere, such as its official promotion of the pilot policy for emission trading in 2007 and the Air Pollution Prevention Action Plan issued in 2012, which have an impact on the identification of the effects of the carbon pilot policy.To eliminate the interference of other relevant policies during the sample period, this paper adds corresponding explanatory variables to formula (1) to control for the influence of other environmental regulatory policies on the benchmark regression results.The regression results show that the regression coefficients of multiple double-difference variables DID are all significantly negative at 1%, which shows that the benchmark regression results in this paper are robust.In addition, although the regression coefficients of other policies fail the significance test, they are all negative, indicating that other policies have a synergistic effect on carbon emission reduction.This paper also tests the robustness of the benchmark regression results by excluding some special samples that could influence the results.Because there are many missing values for the urban economic variables in 2017, it may interfere with the accurate identification of carbon pilot policy effects.In this paper, the sample data from 2017 are excluded, and regression is conducted.The results are shown in columns (1) and (2) of Table 3, and the coefficients of the core explanatory variables are significantly negative at 1%.As the pilot policies, include Beijing, Shanghai and Shenzhen, three large cities with advanced economies, these three cities may adopt stricter environmental governance policies during the sample period, which may interfere with the accuracy of the benchmark regression.Therefore, this paper excludes the three samples from Beijing, Shanghai and Shenzhen, and the regression results are shown in columns ( 3) and ( 4) of Table 3.The regression coefficients of multiple double-difference variable DID are significantly negative at 1%. Notably, the regression coefficient obtained by excluding special years and special samples does not change significantly, which further shows that the benchmark regression results in this paper are robust.
The next point involves the robust estimator using overlapping DID.An important assumption of overlapping DID under the framework of two-way fixed effects is the homogeneity of the policy effect; that is, the policy effect is the same within the experimental group and at different time points in the same group.However, in reality, the homogeneity assumption is often not established because of the different sample characteristics of the experimental group.Relevant studies have pointed out that in the case of heterogeneity, the group that is treated first is regarded as the "wrong" experimental group, and under the condition that the parallel trend assumption is established, the error increases with the increase in the number of samples treated successively [33].Specifically, there are 186 regions that have never been treated, accounting for 75.3% of the total sample, and only 24.7% of the samples have been treated successively.Therefore, it can be inferred that the bias of the two-way fixed effect coefficient obtained in the benchmark regression is very small.For the sake of scientific rigor, this paper uses a two-stage DID method to identify the effect of the carbon pilot policy on regional emission reduction.The regression results are shown in Table 4, in which the control variables are not added in columns ( 1) and ( 3) and are added in columns ( 2) and ( 4).The carbon pilot policy has a significant effect on regional emission reduction under the condition of using an overlapping DID robust estimator, and the regression coefficient is very close to the benchmark regression result.
Regarding the synergistic effect, carbon dioxide comes mainly from the combustion of fossil energy such as coal, oil and natural gas, from which other pollutants in the atmosphere (such as sulfur dioxide and nitrogen oxides) come, and air pollutants show the phenomenon of having the same roots and homologies.To further test the robustness of the benchmark regression results, this paper takes sulfur dioxide as the explained variable for regression, and the results show that d the coefficient of the core explanatory variable DID is significantly negative at 5%, and the quantity and intensity of sulfur dioxide emissions in the carbon trading pilot areas obviously decrease after the implementation of the policy.This finding shows that the carbon pilot policy can reduce the amount of air pollutant emissions such as sulfur dioxide, nitrogen oxides and PM2.5 while reducing carbon dioxide emissions, and the construction of a carbon trading market can control air pollution, which is of positive significance for improving the quality of the atmospheric environment and promoting the green transformation of economic and social development.

Mechanism analysis of carbon pilot policy affecting carbon emission reduction
To explore the transmission mechanism of policies to reduce carbon emissions, this paper conducts an empirical test on carbon pilot policies through industrial structure adjustment and green innovation.In addition, the double-difference regression results can show only the average treatment effect (ATE), and the provinces and cities that implement the carbon pilot policies include cities such as Beijing and Shanghai, which are driven mainly by service industries, and Tianjin, Guangdong and Chongqing, which are driven mainly by manufacturing.Will the implementation effect of the policy change be due to the different proportions of urban industrialization?This paper conducts a classified investigation to answer this question.

Green innovation effect
The carbon pilot policy can reduce carbon emissions by affecting green innovation investment.Enterprises face many costs to carry out green technology innovation, and the carbon pilot policy provides additional benefits for green enterprises with strong emission reduction capacities; they can obtain corresponding profits by selling excess quotas on the trading market, and at the same time, the policy also increases the burden for enterprises with weak emission reduction capacities.Therefore, the carbon pilot policy stimulates enterprises to carry out green innovation through the roles induced by costs and benefits.In this paper, the number of practical patent applications for green innovation in this year is taken as the proxy variable for green innovation, and the model is as follows: The mediator represents the number of practical patent applications for green innovation, and β 1 measures the impact of pilot policies on green innovation.If the pilot policy aims to reduce carbon emissions by encouraging regional green innovation technologies, then β 1 should be significantly positive.
From column (2) of Table 4, it can be seen that the coefficient is significantly positive after adding control variables and fixing the interaction items of individual, year and region, indicating that compared with nonpilot areas, the carbon pilot policy has significantly improved the intensity of green investment in p.ilot areas [34].

Effect of industrial structure adjustment
Structural adjustment is the fundamental way in which to save energy and reduce emissions.The carbon pilot policy affects the regional industrial structure in two main ways.First, through the definition of property rights, the pollution costs generated in the production process of enterprises are internalized, which indirectly increases the costs of producing pollution-intensive products.For this reason, enterprises choose to reduce the number of high-pollution products produced and increase the number of clean products produced.Second, by setting the maximum total regional emissions to stimulate enterprises to eliminate backward production capacity, the increase in production and management costs of "three-high" enterprises makes it difficult for them to survive in the market.At the same time, relevant departments intensify their macro-control to achieve the corresponding emission reduction targets, requiring pilot areas to build low-carbon industries that meet their own conditions and leading the regional industries to rationalize and make green adjustments.The industrial structure shows that the value added of the secondary industry accounts for a decrease in the proportion of GDP.Therefore, this paper selects the ratio of the value added of the tertiary industry to that of the secondary industry as the industrial structure variable with which to investigate whether the carbon pilot policy can reduce emissions through industrial structure adjustment.The model is the same as formula (4), in which the mediator represents the industrial structure.To avoid multicollinearity, control variables do not include the proportion of value added of the secondary and tertiary industries to GDP.
As shown in Table 5 e , in columns (3), control variables are not added, and the fixed effect is used for estimation.In columns (4), the control variables and fixed effect of individual, year, and region interaction items are added.The estimation coefficient of the core variable is significantly positive at the 5% level, indicating that the carbon pilot policy has significantly improved the industrial structure variables.The proportion of the value added of the secondary industry is closely related to and the "main force" behind carbon emissions.The reduction in this proportion significantly reduces the level of carbon emissions.

Heterogeneity analysis
Because there are obvious differences across pilot areas, the influence of policies on different pilot areas may be heterogeneous.The discussion on this issue is helpful for deeply understanding the mechanism and boundary conditions of the carbon pilot policy.Considering that the proportions of the secondary industry in Tianjin, Guangdong and Chongqing in the pilot areas are obviously different from those in other areas, whether the policy effect will be influenced by the proportion of industries in pilot cities is investigated.The model is as follows: e To eliminate the influence of the financial crisis and the "4 trillion" policy on different regions, samples before 2010 are excluded.
Table 5. Mechanism analysis of carbon pilot policies affecting carbon emissions.

Variable
Application for green innovation and practical patent Industrial structure where IND is a dummy variable.When the pilot area is Tianjin, Chongqing or located in Guangdong Province, the value is taken as 1; otherwise, the value is taken as 0. The interaction coefficient β 2 of the core explanatory variables DID and IND measures the impact of the carbon pilot policy on emission reduction with different levels of regional industrialization, and β 1 represents the policy effect when the IND variable is 0, that is, the emission reduction effect of the pilot policy in other regions.Other variables are the same as in formula (1).
The regression results are presented in columns ( 1) and ( 2) of Table 6.When control variables are added and the interaction items of individual, year, and region are fixed, the β 2 coefficient is significantly positive at 5%, while the β 1 coefficient is significantly negative and smaller than that in the benchmark regression results, which shows that carbon pilot policies in other regions have better emission reduction effects than do those in Tianjin, Guangdong and Chongqing.Possible reasons for this finding are as follows: ① Because the production line of large factories is fixed and the cost of industrial transformation is high, it takes a long time to research and develop green innovative technologies, and with high uncertainty.Faced with the exogenous cost increase brought about by the carbon pilot policy, manufacturers' first response is to reduce production and carbon emission levels while maintaining normal production operations, rather than initiating green transformation and upgrading.② There are many nationalized heavy industrial enterprises in Tianjin, Guangdong and Chongqing.Although the current rules and regulations regard environmental factors as an important factor in terms of whether the leadership of state-owned enterprises can be promoted, considering that state-owned enterprises play a "ballast stone" role in economic and social development, the effect of market-oriented environmental regulation policies on the emission reduction of state-owned enterprises is quite limited [35,36].

Considering the policy effect from the perspective of the price incentive and government constraint
At present, the development of the carbon pilot market in China is not perfect, and there is a lack of functionality of the market mechanism.For example, the electricity market has not been completely reformed in terms of price, and there are few subjects involved.In addition, environmental regulation in China used to be administrative.Does the price mechanism in the trading market play a role in policy emission reduction?This question needs further investigation.The details are as follows: where price it is an indicator for measuring the price mechanism, which is embodied in the annual average of the transaction price in the carbon trading market.Price it is a more detailed version of the virtual variable DID, and β 1 is its corresponding coefficient, which is the key coefficient with which to judge whether the price incentive mechanism is effective.The meanings of the other symbols are the same as in formula (1).
Table 7 shows the relevant regression results.Columns ( 2) and (4) show that β 1 is significantly negative at 1% and 5%, respectively, which means that the carbon trading price has significantly enhanced the carbon emission effect of the carbon market and that the price mechanism has played a role in emission reduction.To prove the influence of the price incentive mechanism on emission reduction, this paper selects the proportion of annual transaction volume to regional carbon emissions in that year and the development degree of the factor market as proxy variables for market perfection and interacts them with the core explanatory variable DID.The specific model is as follows: where market it represents the ratio of total annual transaction volume to regional carbon emissions in that year and the development degree of the factor market (input).The interaction coefficient β 2 between DID and market it measures the influence of market mechanism perfection on the emission reduction effect of the pilot policy.If the coefficient is significantly negative, then the emission reduction effect of the carbon pilot policy is enhanced with the improvement of the market.Other variables are the same as in formula (1).As the table 8 shows that the regression coefficients are all significantly negative at 5%, which further proves that the price incentive mechanism has played a role in carbon pilot policy emission reduction.
To verify whether the emission reduction effect of the carbon pilot policy in pilot areas comes from nonmarket mechanisms other than price incentives and whether nonmarket mechanisms are related to the binding force of the government, price it in formula (6) with penalty i is an index to measure the government's binding degree, specifically the punishment enacted on enterprises when their carbon emissions exceed the allowable amount, and β 1 is the corresponding coefficient, which can explain whether the government's binding degree affects the key coefficient of the carbon emission reduction effect in the carbon market.The meanings of the other symbols are the same as in formula (1).

Data
Columns ( 2) and (4) of Table 9 show the relevant regression results.The regression coefficient β 1 is significantly negative at 1%, which shows that the greater the government binding force of the nonmarket mechanism is, the more obvious the emission reduction effect in pilot areas.The reason for this is that the government strictly examines the materials that the enterprise will use to receive the project subsidy after the period ends, which will make it more difficult for the enterprise to obtain government subsidies.Moreover, the government will reduce the quota for its free carbon emission trading rights, further increasing the burden on the enterprise.Some enterprises do not consider the cost-benefit mechanism in carbon market transactions but regard punctual performance as a means through which to improve their relations with the government.Therefore, the binding force of the government also plays a very important role in the emission reduction of the pilot policy.
Because there may be some subjectivity in assigning the value according to the punishment enacted by provincial and municipal governments on enterprises that fail to complete the contract in time, the regression results obtained may be inaccurate.Therefore, this paper selects the absolute value of the ratio of the fiscal deficit in the local budget to the fiscal revenue in the budget to measure the binding force of local governments on local enterprises.Local governments facing greater financial pressure are easily captured by local elites, and then, they relax the intensity of environmental regulation and passively enforce the law [37].Specifically, the variable of marketization in formula ( 7) is replaced by the variable

Data Intelligence
fin_burden it to measure the financial pressure faced by local governments.If the DID*fin_burden coefficient of the interaction term is positive, then the carbon pilot policy will lead to a decline in the emission reduction effect in areas with high degrees of financial pressure because local governments cannot provide a good law enforcement environment.As shown in Table 10, the interaction coefficient is significantly positive at least at 5%, which further strengthens the conclusion that government constraints play an important role in the carbon pilot policy.
In short, according to the above empirical results, the carbon market can significantly reduce carbon emissions in pilot areas, and the price mechanism plays an important role such a reduction, but the power of government constraints cannot be ignored.Accordingly, the empirical research conclusion of this paper supports that at present, the incentive effect of the price mechanism based on the trading market and the government restraint effect of the non-trading market in the carbon pilot area of China cooperate with each other to urge emission control subjects to fulfill their obligations, thus reducing carbon emissions.Based on the experience of China's emission trading policy, the government needs to intervene through the use of administrative force before the carbon market becomes mature.With the gradual improvement of the carbon market system, the price mechanism will play a leading role in emission reduction.

Price Mechanism, Government Constraints and Carbon Trading Pilot Policy for Emission Reduction
Data 23

CONCLUSIONS
Against the background of severe and the deterioration of resources and the environment, can the carbon pilot policy effectively reduce emissions, help China achieve its goal of "double carbon" and assume the important responsibility of environmental governance while maintaining economic growth?The answer is yes.Based on the data of 247 prefecture-level cities in China from 2007 to 2019, this paper effectively controls for potential endogeneity problems by using the difference-in-differences identification framework.The research conclusion shows that on the whole, the carbon market has significant emission reduction effects on pilot areas, including those related to the amount of carbon emissions and carbon intensity.Through mechanism analysis, it is found that the carbon pilot policy realizes the emission reduction effect mainly by adjusting the industrial structure and encouraging green innovation.Heterogeneity analysis shows that the higher the degree of industrialization is, the smaller the emission reduction effect of the carbon pilot policy.At present, the carbon market is still in the process of trial and error, and the role of price incentives is limited.The effective implementation of the carbon market needs the coordination and help of the government, and the carbon market realizes carbon emission reduction through the coordination of price incentives and administrative restraint mechanisms.Based on the above findings, the policy recommendations in this paper include the following: 1) Adhere to market orientation and promote the continuous expansion of market players.According to the conclusion of this paper, the higher the carbon emission trading price, the stronger the incentive for regional emission reduction in the carbon trading market, and the government can gradually reduce the proportion of free allowances and increase the proportion of paid auctions to increase the carbon trading price.In addition, the single industry covered by the carbon market and the insufficient number of market players are the main reasons for the sharp fluctuation of carbon emission trading prices and the lack of market liquidity.We should continue to strengthen the institutional construction of the carbon market, such as allowing third-party institutions to monitor, report and verify carbon emission data, and form a reasonable reward and punishment mechanism to restrain the behavior of market participants, so as to create a good prospect for the market.Build a well-organized and well-functioning trading platform, and encourage appropriate innovation in carbon financial products.At present, China's carbon market trading environment is still immature and not closely related to the reform of electric power, so it is necessary to establish and improve the policy and system of carbon emission trading to promote the simultaneous advancement of coal power reform.
2) Improve the relevant laws and regulations and provide local governments with special green funds.Local governments need to strengthen the management and guidance of the carbon trading market and establish a long-term system.It is advisable to pass legislation on the management of carbon emission trading as soon as possible, and the national carbon market should further improve the total control mechanism, quota allocation mechanism, quota allocation method, trading system, and promote the operation of the national carbon trading market to have laws to follow.Combined with the introduction of relevant laws and regulations, the establishment of a national carbon trading market joint supervision system and departmental coordination mechanism, the establishment of independent regulatory agencies and manage carbon trading.Clarify the responsibilities of relevant departments to ensure the stable development of the national carbon market.Local government funding plays an important role in promoting regional emission reductions, so the central government should implement peer-to-peer green allocations in areas with high fiscal pressure, so as to prevent local governments from relaxing environmental regulations for their own performance.
3) Fully consider the relationship between fairness and efficiency of carbon emission reduction.The construction of the carbon market should coordinate the relationship between fairness and efficiency.In terms of inter-sector fairness, it is necessary to expand the sector coverage of the carbon market as much as possible, explore the emission reduction spillover effects of regulated industries on upstream and downstream industries, and take into account the fairness of covered and uncovered industries, which will help improve the efficiency of the carbon market.For inter-regional fairness, it is necessary to take into account the differences in industrial base, emission reduction capacity, and fiscal revenue between regions, and adopt flexible emission reduction targets and transfer payments to give certain support to regions with greater difficulty in reducing emissions.This paper further deepens the research on the implementation of price-based environmental regulation policies in developing countries, but some shortcomings still exist, such as whether the carbon transaction price has a threshold effect on emission reduction (it cannot play a role in emission reduction when the price is too low) and how China can deepen its reform of the power sector to improve the carbon market.These aspects could be considered in future studies.

Figure 1 .
Figure 1.Composition of the carbon trading market.

Figure 2 .
Figure 2. Changes in prices in various carbon markets over time.

b
The specific pilot time is June 2013 in Shenzhen; November 2013 in Beijing; December 2013 in Tianjin, Shanghai, and Guangdong; April 2014 in Hubei; June 2014 in Chongqing; and December 2016 in Fujian.

Table 1 .
Description of main variables.

Table 2 .
Benchmark Regression of Carbon Pilot Policy Emission Reduction.
Note: The values in brackets are the clustered standard errors at the provincial level, and *, * * and * * * represent significance levels of 10%, 5% and 1%, respectively.

Table 4 .
Robust estimator of DID.
Note: The values in brackets are the clustered standard errors at the provincial level, and *, * * and * * * represent significance levels of 10%, 5% and 1%, respectively.

Table 6 .
Regression results of heterogeneity analysis.The values in brackets are the clustered standard errors at the provincial level, and *, * * and * * *represent significance levels of 10%, 5% and 1%, respectively.

Table 8 .
Regression results of market perfection on emission reduction.

Table 7 .
Regression results of the effect of price incentives on emission reduction.The values in brackets are the clustered standard errors at the provincial level, and *, * * and * * *represent significance levels of 10%, 5% and 1%, respectively.

Table 9 .
Regression results of government constraints on emission reduction.

Table 10 .
results on the impact of government fiscal pressure on emission reductions.