The Impact of China Carbon Emission Trading System on Land Use Transition: A Macroscopic Economic Perspective

: The carbon emission trading system (CETS) is a milestone policy in the history of China’s emission trading system, which is of great signiﬁcance to China’s realization of “carbon peak and carbon neutralization”. As an important component of sustainable development, LUT should be related to the CETS. However, in the literature on the CETS, little material deals with its impact on land use transition (LUT). This paper will enrich this literature. Based on 30 provincial regions in China from 2011 to 2017, using the DID and entropy methods, this study investigated the impact of CETS on the trend of LUT from three perspectives: economic effects, environmental effects and Porter effects. The conclusions are that (1) the implementation of the CETS hindered economic development, but optimized energy-use efﬁciency; (2) the implementation of the CETS reduced the emissions of CO 2 and SO 2 ; (3) the implementation of the CETS did not produce a Porter effect; and (4) the inﬂuence of the CETS had the characteristics of a spatial cluster. These ﬁndings offer some guidance for improving CETS policies and formulating similar environmental regulation policies. the implementation of the CETS promotes land use transition in the pilot areas toward economic development, but will have a negative impact on the economy in the short


Introduction
A "regulation" is a general rule or a special act that is developed and implemented by an administrative body with the purpose of directly intervening in the market allocation mechanism, or indirectly changing the supply and demand decisions between enterprises and consumers [1]. The actor of regulation is usually a state administrative agency with administrative power and its behavior towards the regulated object is generally forced. Environmental regulation is one of the regulatory instruments that aims to harmonize the relationship between the economy and the environment. The emission trading system (ETS) is one of the most promising and effective environmental regulation tools [2] and one of the hot spots of research in China. Scholars have focused their research into ETS on corporate decision-making under the system, spillover effects between the system and other markets, and governmental allocations of emission rights [3]. The carbon emission trading system (CETS) is one of the ETSs and its implementation is a general trend. In 2010, China's per capita carbon emissions rose to the G20 countries' average (Source: https: //www.climate-transparency.org/wp-content/uploads/2016/02/CP_CHINA_2015, accessed on 25 October 2021) and by 2016, China's per capita greenhouse gas emissions were 8.8 tons of carbon dioxide equivalent, 17% higher than the G20 average (Source: https: //www.climate-transparency.org/wp-content/uploads/2019/11/B2G_2019_China, accessed on 25 October 2021). Although China's per capita carbon emissions are far lower than those of developed countries, such as the United States, they still need to be taken seriously. The CETS policy is undoubtedly a milestone in China's emission reduction history [4] and it is strategically important for China to successfully achieve the goal of have explored the effect of CETS on LUTs. Our study aims to fill this gap (see Figure 1 for the technical path).
Land 2022, 11, x FOR PEER REVIEW 3 of 1 al. found that there is a negative spatial spillover effect of the CETS [55], and so on. Few papers have explored the effect of CETS on LUTs. Our study aims to fill this gap (se Figure 1 for the technical path). The CETS has economic effects. It not only forms a green economy, but also accom panies industrial structural upgrading and social welfare improvement. As the CETS wi lead to an increase in production costs, enterprises will reduce the cost of environmenta management per unit of product by increasing their production. Alternatively, enterprise will try to improve product quality, increase profit per unit of product, and even ente new industries with low energy consumption and low pollution in order to develop. I addition, the compensation strategy of the CETS will also benefit some enterprises [56 However, as Dong et al. argued (as noted above), the short-term impact of the CETS o the economy is likely to be negative [52], even if the policy compensates for some of th losses of the firms [56], since a shift in production strategies and business policies, etc cannot be accomplished in the short run. Considering the limitation in the scope of th data, the sample interval of this paper is short. Therefore, this paper proposes Hypothesi 1 (H1): the implementation of the CETS promotes land use transition in the pilot area toward economic development, but will have a negative impact on the economy in th short term.
The CETS has an environmental effect in that its required emission restrictions forc firms to improve their resource use efficiency and reduce consumption, it expels from th market those firms that are pollution-intensive but cannot reasonably manage their po lution [57], and it stimulates firms to invest more in pollutant reduction and treatmen [58]. Therefore, this paper proposes Hypothesis 2 (H2): the implementation of CETS pro motes land use transition in pilot areas toward green development.
The CETS also has a Porter effect. China's factor-driven, crude growth approac leads to low resource utilization and low economic efficiency of resources [59], meanin it is not a high-quality economic development model. In contrast, the innovation-drive model is key to improving productivity and driving economic transition [60,61], which i important to achieving high-quality economic development in China [62]. Therefore, thi paper proposes Hypothesis 3 (H3): the implementation of the CETS promotes land us transition in the pilot areas toward high-quality economic development.
The marginal contributions of this paper relate to the driving mechanisms of LUT We found that there is a lack of research on the CETS policies and LUT within the sam research framework and few studies have been conducted from an economic perspective Second, the causal relationship between the CETS and LUT is more convincingly ident fied through the DID estimation method.
The rest of this article is structured as follows. Section 2 focuses on the study meth ods, variable design, and data sources. Section 3 describes the findings of this study an the corresponding testing procedures. Section 4 discusses the findings of the previous sec tions and draws out the conclusions that were obtained from this study. The CETS has economic effects. It not only forms a green economy, but also accompanies industrial structural upgrading and social welfare improvement. As the CETS will lead to an increase in production costs, enterprises will reduce the cost of environmental management per unit of product by increasing their production. Alternatively, enterprises will try to improve product quality, increase profit per unit of product, and even enter new industries with low energy consumption and low pollution in order to develop. In addition, the compensation strategy of the CETS will also benefit some enterprises [56]. However, as Dong et al. argued (as noted above), the short-term impact of the CETS on the economy is likely to be negative [52], even if the policy compensates for some of the losses of the firms [56], since a shift in production strategies and business policies, etc., cannot be accomplished in the short run. Considering the limitation in the scope of the data, the sample interval of this paper is short. Therefore, this paper proposes: Hypothesis 1 (H1). the implementation of the CETS promotes land use transition in the pilot areas toward economic development, but will have a negative impact on the economy in the short term.
The CETS has an environmental effect in that its required emission restrictions force firms to improve their resource use efficiency and reduce consumption, it expels from the market those firms that are pollution-intensive but cannot reasonably manage their pollution [57], and it stimulates firms to invest more in pollutant reduction and treatment [58]. Therefore, this paper proposes: Hypothesis 2 (H2). the implementation of CETS promotes land use transition in pilot areas toward green development.
The CETS also has a Porter effect. China's factor-driven, crude growth approach leads to low resource utilization and low economic efficiency of resources [59], meaning it is not a high-quality economic development model. In contrast, the innovation-driven model is key to improving productivity and driving economic transition [60,61], which is important to achieving high-quality economic development in China [62]. Therefore, this paper proposes: Hypothesis 3 (H3). the implementation of the CETS promotes land use transition in the pilot areas toward high-quality economic development.
The marginal contributions of this paper relate to the driving mechanisms of LUT. We found that there is a lack of research on the CETS policies and LUT within the same research framework and few studies have been conducted from an economic perspective. Second, the causal relationship between the CETS and LUT is more convincingly identified through the DID estimation method.
The rest of this article is structured as follows. Section 2 focuses on the study methods, variable design, and data sources. Section 3 describes the findings of this study and the corresponding testing procedures. Section 4 discusses the findings of the previous sections and draws out the conclusions that were obtained from this study.

Difference-in-Difference Model
To assess the treatment effect of a policy or event, economists often use a counterfactual framework, referring to the difference between data that are treated (the treatment group) and data that are not treated (the control group) as the "treatment effect." Differencein-difference (DID) is the measure that is most widely used to estimate the treatment effect. The original logic of the method was proposed by Snow in his study on the cholera epidemic in London [63] and it was introduced into economics by Obenauer in his study on the treatment effect of the minimum wage method [64]. To estimate the treatment effect and compare the difference before and after a policy or event, the method assesses the before-and-after change in the treatment group minus the before-and-after change in the control group, hence the name "difference-in-difference". Therefore, DID is often used to identify causal relationships between explained and explanatory variables. If the model satisfies the parallel trend assumption, then the difference-in-difference method can be used directly. Stata 15 software was used to execute the model.
When DID uses panel data, a two-way fixed effects model is usually applied. Based on this, the DID model incorporates the interaction term of the treatment group dummy variable (Treat i ) and the treatment period dummy variable (Post t ). The basic form of the model is as follows: where i denotes the individual and t denotes time. Y i,t is the explained variable. The variable µ i is the province fixed effect; γ t is the year fixed effect; and ε i,t is the random error term. Treat i is the treatment group dummy variable, which takes the value of 1 if individual i belongs to the "treatment group" that was subject to the policy shock; otherwise, it takes the value of 0. Post t is the treatment period dummy variable, whereby individuals in the treatment group were subject to the policy shock only during the treatment period. Therefore, if time t belongs to the "treatment period", it takes the value of 1; otherwise, it takes the value of 0.
Finally, the implications of this study are given in Equation (1) and some of the labels were adjusted in order to obtain the model that was used in this paper.
where LUT i,t is the land use transition variable and Treat i × Post t is the treatment variable of the CETS. Figure 2 shows the distribution of the pilot areas of the CETS policy at the provincial level in China.

Entropy Method
The entropy method, which is a comprehensive indicator method, has the characteristic of giving weights through the deviation between variables. When the uncertainty between variables is greater, the degree of deviation is also greater, indicating that it contains more valid information and it is therefore assigned a higher weight. Conversely, variables that contain less valid information are naturally given less weight. Based on this characteristic, the entropy method has the advantage of alleviating information overlap among variables and giving weights objectively. In addition, the weight of each secondary indicator reflects its contribution to the composite indicator, which allows scholars to further explore the impact of the refined indicators [65]. Referring to the method of Tang et al. [66], the basic steps of the entropy method that was used in this paper are as follows. We used MATLAB R2016a software to realize the measurement.
Step 1: , and t is period t. The term xi,b,t is the element of the bth indicator in period t of region i. We derived the normalized form of xi,b,t by using Equation (3).
Step 2: Information entropy was assumed to be Eb,t, which reflects the intensity of the amount of information contained in the bth indicator in period t in all regions. The number 30 refers to the 30 provinces and municipalities directly under the central government that were involved in this study. The specific equation is as follows.

Entropy Method
The entropy method, which is a comprehensive indicator method, has the characteristic of giving weights through the deviation between variables. When the uncertainty between variables is greater, the degree of deviation is also greater, indicating that it contains more valid information and it is therefore assigned a higher weight. Conversely, variables that contain less valid information are naturally given less weight. Based on this characteristic, the entropy method has the advantage of alleviating information overlap among variables and giving weights objectively. In addition, the weight of each secondary indicator reflects its contribution to the composite indicator, which allows scholars to further explore the impact of the refined indicators [65]. Referring to the method of Tang et al. [66], the basic steps of the entropy method that was used in this paper are as follows. We used MATLAB R2016a software to realize the measurement.
Step 1: , and t is period t. The term x i,b,t is the element of the bth indicator in period t of region i. We derived the normalized form of x i,b,t by using Equation (3).
Step 2: Information entropy was assumed to be E b,t , which reflects the intensity of the amount of information contained in the bth indicator in period t in all regions. The number 30 refers to the 30 provinces and municipalities directly under the central government that were involved in this study. The specific equation is as follows.
Step 3: The weight of the bth indicator in period t (W b,t ) was calculated. The term k is the total number of indicators. LUT can refer to the process of land use pattern changes that occur over time corresponding to the transition between economic and social development stages, which are inseparable from economic activities. In this paper, the change tendency of LUT is measured in three dimensions: the economic, environmental and Porter effects of the CETS. The construction of the indicators for these three dimensions of LUT is described below (Table 1). 1.
The economic level reflects the impact of the CETS policy on the economic development level of the pilot region. After the implementation of the CETS, it was noted that the output of pilot enterprises was negatively affected in the short term. This was due to the difficulty experienced by enterprises in attempting to quickly adjusting their production patterns in the short term. Considering the contradiction between the cost and liquidity of enterprises, the innovative R & D behavior of enterprises may also slow their operation, thus leading to a decline in the economic level in the region. In the long run, the production efficiency of enterprises in the region was improved and the products were more competitive, which may lead to a rise in the economic level in the region. In any case, the implementation of the CETS policy changed the economic level of land use in the region. The economic level consists of four secondary indicators: per capita GDP, GDP per unit of land area, added value from secondary industry that is generated by power, and proportion of added value from tertiary industry. (I) Per capita GDP (PGDP), which directly reflects the economic level within a region but is influenced by the size of the population [67], is divided by the number of residents (i.e., the population) in order to limit this effect. (II) Regions with high economic levels should be able to utilize each piece of land more fully in order to generate more economic returns. Although the type of land is closely related to its economic value, the trend of enhancing the economic return per unit of land output was constant. At the same time, the entropy method can better mitigate the impact of the size of the indicators' value. Therefore, in this paper, GDP divided by land area, i.e., GDP per unit of land area (LGDP), was used as one of the indicators in order to reflect the economic level of a region. (III) The added value from secondary industry that is generated by power consumption per unit of industrial production (SIPC) reflects the economic benefits that can be generated per unit of energy and also reflects the economic activities that are undertaken by enterprises in the region in response to the CETS. (IV) The proportion of added value from tertiary industry (PIT) is the ratio of tertiary industry to total industry value added. The value added from industry was reduced because CETS may induce some highly polluting and low-profit enterprises to leave the market. At the same time, CETS will encourage some firms to flow into the tertiary industry with low pollution and high profit margins, thus causing the PIT to increase [68,69]. When firms in the region are influenced by CETS and initiate innovative R & D, they will increase their material investment (that is, RDIE will be higher), which will drive economic growth [70]. (III) The number of non-industrial design patent applications per capita (NDPA) reflects the level of innovation in a region [71]. When a pilot region is subject to CETS and starts innovation activities, the end result is a significant increase in the number of utility model patent and invention patent applications (both are collectively referred to as nonindustrial design patents). (IV) The proportion of science and technology spending in the general public budget expenditure (PSTE) reflects the importance that local governments place on local innovation development. When regions are affected by the CETS, local officials will enact a series of targeted incentives in order to guide the economic activities of local enterprises, improving their own performance and catering to the policy requirements [72]. The core explanatory variable in this study is the interaction term of the treatment group dummy variable and the treatment period dummy variable (Treat i × Post t ). According to the previous analysis, Treat i × Post t needs to reflect whether region i in period t is a pilot region that has implemented CETS. When Treati × Postt = 1, this indicates that region i is a pilot region and period t occurred in the year of the CETS policy's implementation or after. Otherwise, Treat i × Post t = 0. Equation (6) is the constructive equation of Treat i × Post t .

Explanatory Variable
The core explanatory variable in this study is the interaction term of the treatment group dummy variable and the treatment period dummy variable (Treat i × Post t ). According to the previous analysis, Treat i × Post t needed to reflect whether region i in period t was a pilot region that had implemented the CETS. When Treat i × Post t = 1, this indicates that region i was a pilot region and period t occurred in the year of the CETS policy's implementation or after. Otherwise, Treat i × Post t = 0. Equation (6) is the constructive equation of Treat i × Post t .
Treat i × Post t = 1, i is the pilot area and CETS has been implemented in period t 0, otherwise ,

Data Sources
This study examined 30 provinces and municipalities that were directly under the control of the central government in China from 2011 to 2017. Considering data integrity and statistical caliber, the Tibet Autonomous Region, Hong Kong and Macao Special Administrative Regions, and Taiwan were excluded from the scope of the study. The primary data that were used in this study were mainly sourced from the China Statistical Yearbook, China Environmental Statistical Yearbook, provincial statistical yearbooks, and the China Research Database Service Platform (CNRDS). The data on carbon emissions were obtained from the China Emission Accounts and Datasets (CEADs), the measurement of which can be found in Shan et al. [73,74]. Table 2 reports the basic information of all of the variables that are involved in this study.

Baseline Regression Results
This paper examines the effects of the causal relationships between the CETS on economic level tendency (ELT), green development tendency (GDT) and economic quality Land 2022, 11, 41 9 of 15 tendency (EQT) of LUT through the DID model of Equation (2). Since the six pilot provinces of CETS were opened one by one between June 2013 and June 2014 and there were time lags in the economic quality tendency and economic level tendency of LUT, we took 2013 and 2014 as the start-up times of the CETS policies. This paper used the Stata 15 software computing model. Table 3 reports the baseline regression results.  Table 3(I) is −2.6208, which passes the significance test at the 5% level of significance. This indicates that the CETS significantly reduced the ELT of the pilot provinces, regardless of whether 2013 or 2014 was used as the start of the CETS policy (Therefore, H1 holds). Meanwhile, by comparing the magnitude of the t-values and the magnitude of the coefficients, it can be tentatively determined that the CETS policy had a greater impact on ELT in 2014 and after.
The coefficient of the Treat i × Post 2013 term in the first column of Table 3(II) is −0.3088, significant at the 1% significance level, indicating that the GDT of the pilot regions declined with the implementation of the CETS policy (Therefore, H2 holds). Since GDT is a negative indicator, and the lower the GDT, the lower the environmental pollution, the CETS improved environmental pollution in the pilot areas. The coefficient of the Treat i × Post 2014 term in the second column of Table 3(II) does not pass the 10% significance level test; although it is also negative, indicating that the CETS cannot significantly improve environmental pollution if 2014 is used as the start of the CETS policy. This indirectly reflects that the impact of the CETS policy on green development started in 2013.
The coefficients of the Treat i × Post 2013 and Treat i × Post 2014 items in Table 3(III) are 0.0523 and 0.3416, respectively, and consistent with our expectations. However, both scores failed the significance test at the 10% significance level, indicating that there is no significant causal relationship between CETS and the improvement in economic quality tendency in the pilot regions, regardless of whether 2013 or 2014 is used as the starting point for the implementation of the CETS policy (Therefore, H3 does not hold).

Parallel Trend Assumption Test
Parallel trend assumption is a necessary condition for the validity of the DID model's conclusions. It assumes that the treatment and control groups must be comparable and consistent before the policy or event occurs. Accordingly, this paper designed the parallel trend assumption strategy: before the CETS policy was implemented, the treatment and control groups had basically similar trends of change; but after the CETS policy was implemented, a significant difference emerged between the treatment group and the control group due to the impact of the policy shock. In this paper, the event analysis framework was used to examine this strategy. The regression equation is shown in Equation (7).
where Dum τ is the year dummy variable and β τ is the main examined coefficient of Equation (7). If none of the β values corresponding to the pre-CETS policy implementation are significant, while most of the β values corresponding to the post-CETS implementation pass at least the 10% significance test, then the model's results satisfy the parallel trend assumption (and vice versa, which would indicating that the study's results are not valid).
The results of the baseline regressions that are described above show that the CETS policy with 2013 as the start-up time had a significant effect on ELT and GDT, while the CETS with 2014 as the start-up time only had a significant effect on ELT. Therefore, this section performs parallel trend assumption tests for each of the three cases mentioned above. Figure 3 plots the β τ estimations and their 90% confidence intervals for each year.
significant, while most of the β values corresponding to the post-CETS implementation pass at least the 10% significance test, then the model's results satisfy the parallel trend assumption (and vice versa, which would indicating that the study's results are not valid).
The results of the baseline regressions that are described above show that the CETS policy with 2013 as the start-up time had a significant effect on ELT and GDT, while the CETS with 2014 as the start-up time only had a significant effect on ELT. Therefore, this section performs parallel trend assumption tests for each of the three cases mentioned above. Figure 3 plots the βτ estimations and their 90% confidence intervals for each year. Figure 3a shows the parallel trend test of CETS (with 2014 as the start time) for ELT. It can be observed that the β values corresponding to 2011, 2012 and 2013 fail the 10% significance test, while the β values corresponding to 2014, 2015 and 2017 pass the significance test at least the 10% significance level. While in 2016, the β values were not significant, it is possible to conclude that, overall, the CETS significantly reduced the ELT in the pilot provinces and this finding is valid. In addition, the result showing that the CETS (with 2013 as the start time) had a significant negative effect on ELT does not pass the parallel trend assumption test and therefore is not included in Figure 3.

Discussion of the Combination of the Empirical Results and the Entropy Method Results
This sub-section combines the regression results of Section 3 with the weight changes of each secondary indicator as reported in Table 1. 1. Hypothesis 1 of this paper holds-that is, the implementation of the CETS was found to promote the land use transition in the pilot areas toward economic development, but this had a negative impact on the economy in the short term. The implementation  the significance test at least the 10% significance level. While in 2016, the β values were not significant, it is possible to conclude that, overall, the CETS significantly reduced the ELT in the pilot provinces and this finding is valid. In addition, the result showing that the CETS (with 2013 as the start time) had a significant negative effect on ELT does not pass the parallel trend assumption test and therefore is not included in Figure 3. Figure 3b shows the parallel trend test of the CETS (with 2013 as the start time) for GDT. The β values corresponding to both 2011 and 2012 do not pass the 10% significance test, while the β values corresponding to 2013, 2014, 2016 and 2017 pass the significance test at least the 10% level of significance. Although for 2015 the β values were not significant, it seems that the conclusion that CETS can improve environmental pollution in the pilot regions as a whole is valid.

Discussion of the Combination of the Empirical Results and the Entropy Method Results
This sub-section combines the regression results of Section 3 with the weight changes of each secondary indicator as reported in Table 1. 1.
Hypothesis 1 of this paper holds-that is, the implementation of the CETS was found to promote the land use transition in the pilot areas toward economic development, but this had a negative impact on the economy in the short term. The implementation of the CETS significantly reduced the economic growth trend in 2014 and in the years after. This suggests that, in order to reduce pollution emissions, firms may choose to reduce production or not to produce, while the demand for cash flow from innovative R & D practices will also drag down firms' production, thus reducing the total output in the region. Based on the trend of weight changes in the secondary indicators, per capita GDP (PGDP), GDP per unit of land area (LGDP), added value from secondary industry generated by power consumption per unit of industrial production (SIPC) and the proportion of added value from tertiary industry (PTI) in Table 1, it is clear that LGDP had the highest weight and remained above 50% for many years. This indicates that many regions with high output per unit of land, especially the pilot regions, have started to reduce production, indirectly reflecting that past output may have been caused by high-pollution industries. At the same time, it is noted that the weight value of LGDP has decreased significantly since 2014, while the weight value of SIPC has increased significantly, which, on the one hand, reflects the influence of CETS and, on the other hand, indicates that enterprises in the pilot areas have started to focus on production efficiency and obtain more economic benefits with less energy consumption.

2.
Hypothesis 2 of this paper holds, i.e., the implementation of the CETS promoted the LUT in pilot areas toward green development. The implementation of the CETS has been effective in reducing CO 2 emissions (CE) and SO 2 emissions (SE) in the pilot areas since 2013. This shows that the emitting companies have effectively reduced their related emissions and shifted towards a green development strategy. At the same time, according to the weights of CE and SE in Table 1, the values of the two weights are comparable and do not change significantly over time. This indicates that the discrete changes of the two are roughly the same, which is consistent with the previous conclusion that carbon dioxide and sulfur dioxide emissions are consistent, indirectly reflecting that the CETS also has an emission reduction effect on non-carbon pollution.  Table 1, did not show a clear trend change over time. This reflects that it is currently difficult for the CETS to effectively contribute to high-quality economic development and form an economic-environmental win-win situation.

Spatio-Temporal Analysis of LUTs
Based on the above conclusions and discussions, this sub-section presents the spatiotemporal evolutionary trends of ELT and GDT data on the provincial-level regional distribution map of China (Figure 4) through the use of ArcMap software. In order to better demonstrate the trends of the variables, this study processed the ELT and GDT data as follows (taking ELT as an example).
Step 1: Taking 2013 as the threshold, the average of ELT in 2011 and 2012 was taken as the proxy variable Pre i for ELT before the policy shock. Similarly, the average of ELT from 2013 to 2017 was taken as the proxy variable Pos i for ELT after the policy shock.
Step 2: We calculated the Change i variable, which reflects the degree and direction of change in ELT before and after the policy shock. Change i = Pos i -Pre i -if the sign of Changei is negative, the ELT decreases after the CETS policy shock in region i.
Step 3: The Change i variables were input into the distribution map of provincial areas in China (Figure 4a) through the ArcMap software and, at the same time, the eight areas with the most reduced tendencies were specifically marked out, representing the eight provinces with the most reduced economic values before and after the CETS policy was implemented.
We undertook the same measures for the GDT data in order to obtain Figure 4b. If the sign of the Change i variable is negative, this means that the GDT of region i decreased after the CETS policy shock and the environmental pollution level in region i was improved.
Land 2022, 11, x FOR PEER REVIEW 12 of 15 Step 3: The Changei variables were input into the distribution map of provincial areas in China (Figure 4a) through the ArcMap software and, at the same time, the eight areas with the most reduced tendencies were specifically marked out, representing the eight provinces with the most reduced economic values before and after the CETS policy was implemented.
We undertook the same measures for the GDT data in order to obtain Figure 4b. If the sign of the Changei variable is negative, this means that the GDT of region i decreased after the CETS policy shock and the environmental pollution level in region i was improved.  According to Figure 4a, the economic level decreased in 20 provinces before and after the CETS policy took effect. Six CETS pilot regions declined economically, with Beijing, Tianjin, Shanghai, and Guangdong being among the top eight provinces in terms of decline. These four provinces and cities have higher economic levels and are therefore more affected by the CETS. Similarly, it can be found that the provinces with declining ELTs are mainly concentrated in the eastern coastal region and its adjacent areas and there are clustering characteristics. Figure 4b, shows that there were 14 provinces with improved environmental pollution before and after the CETS policy took effect. Six CETS pilot areas showed improved environments, with Beijing, Tianjin, Shanghai, Hubei, and Chongqing in the top eight provinces in terms of change. This indicates that, although there are non-pilot areas where the environment improved, the degree of improvement was overall less than that in the pilot areas affected by the CETS policy. In addition, the environmental improvement provinces are mainly concentrated in the eastern and central regions and are adjacent to each other, with clustering characteristics.

Policy Recommendations
This paper has obtained the following conclusions: (1) Under the influence of the CETS, land use in the pilot area was transitioned in the direction of economic level reduction in the short term. However, during this period, the efficiency of industrial energy use was improved. This economic disincentive effect had spatial clustering characteristics and was mainly concentrated in the coastal areas of southeast China and economically developed regions. (2) Under the influence of the CETS, the emissions of CO2 and SO2 in the pilot areas significantly decreased and land use transitioned towards green development. This emission reduction effect had spatial clustering characteristics and was mainly concentrated According to Figure 4a, the economic level decreased in 20 provinces before and after the CETS policy took effect. Six CETS pilot regions declined economically, with Beijing, Tianjin, Shanghai, and Guangdong being among the top eight provinces in terms of decline. These four provinces and cities have higher economic levels and are therefore more affected by the CETS. Similarly, it can be found that the provinces with declining ELTs are mainly concentrated in the eastern coastal region and its adjacent areas and there are clustering characteristics. Figure 4b, shows that there were 14 provinces with improved environmental pollution before and after the CETS policy took effect. Six CETS pilot areas showed improved environments, with Beijing, Tianjin, Shanghai, Hubei, and Chongqing in the top eight provinces in terms of change. This indicates that, although there are non-pilot areas where the environment improved, the degree of improvement was overall less than that in the pilot areas affected by the CETS policy. In addition, the environmental improvement provinces are mainly concentrated in the eastern and central regions and are adjacent to each other, with clustering characteristics.

Policy Recommendations
This paper has obtained the following conclusions: (1) Under the influence of the CETS, land use in the pilot area was transitioned in the direction of economic level reduction in the short term. However, during this period, the efficiency of industrial energy use was improved. This economic disincentive effect had spatial clustering characteristics and was mainly concentrated in the coastal areas of southeast China and economically developed regions. (2) Under the influence of the CETS, the emissions of CO 2 and SO 2 in the pilot areas significantly decreased and land use transitioned towards green development. This Accordingly, the policy guidance obtained from this paper is as follows: The implementation of the CETS has been successful for China and its benefits outweigh its disadvantages. Under the influence of the CETS, CO 2 and SO 2 emissions have been significantly reduced and the environmental effects of these reductions have been practically realized. At the same time, although the CETS has inhibited the economic development of the pilot regions, the pilot enterprises are independently improving their energy output per unit. This essentially optimizes the economic development, eliminates high-pollution and low-efficiency enterprises, and encourages the enterprises to transition towards innovative development. The most important deficiency during the implementation of the CETS is that it did not effectively increase the innovation output of the pilot enterprises, which is not in line with the goal of "building China into an innovation power" and is not conducive to the promotion of China's sustainable development path.  Data Availability Statement: Raw data is available from a paid database, please contact the corresponding author if needed.

Conflicts of Interest:
The authors declare no conflict of interest.