Can Setting Up a Carbon Trading Mechanism Improve Urban Eco-Efficiency? Evidence from China

Abstract

The Carbon Emissions Trading Pilot Policy (CETP) has attracted more scholarly attention. However, most existing studies are only singularly focused on carbon emission reduction or economic development. More research is needed to determine whether it can promote green and sustainable urban development. Therefore, this paper takes the data from 284 prefecture-level cities in China from 2007 to 2016 as the research sample, uses ecological efficiency as the indicator to measure the sustainable green development of cities, and uses the difference method (DID) and the propensity score matching difference method (PSM-DID) to study whether CETP can achieve the sustainable green development of pilot cities. The results show that CETP can improve pilot cities’ ecological efficiency and realize cities’ green and sustainable development by optimizing the industrial structure and promoting technological innovation. In addition, the impact of CETP on different cities is also significantly different. Compared with small and medium-sized cities and non-provincial capital cities, CETP has a greater impact on large cities and provincial capital cities. Compared with central and western cities, CETP has a greater impact on eastern cities. CETP can improve the ecological efficiency of non-resource cities, but it cannot change the ecological efficiency of resource cities. Our models survive numerous robustness checks.

1. Introduction

The United Nations Framework Convention on Climate Change (UNFCCC) [1] was signed in 1992, and countries worldwide have taken a series of measures to reduce emissions to protect the environment. One of these measures is carbon emission trading. The concept of carbon emissions trading dates back to 1968 when the American economist Dales introduced the concept. The basic idea is to grant legal rights to emit pollutants in the form of permits so that the environment and resources can be bought and sold like commodities.

More recently, carbon trading mechanisms have been increasingly well-researched. In this context, the National Development and Reform Commission (NDRC) of China approved for the first time in October 2011 the “two provinces and five cities” of Beijing, Shanghai, Tianjin, Shenzhen, Chongqing, Guangdong, and Hubei to implement a pilot scheme for carbon emissions trading. The pilot policy of carbon emission trading (CETP) is a new pilot policy derived from the previous emission trading policy, which has a broader scope and greater relevance. China’s major pilot policy is to use the market mechanism to reduce emissions. Carbon trading implies a unified unit of measurement that is no longer limited by geography and can greatly expand the scale of emissions reduction trading [2,3].

Ecological efficiency is an effective indicator to measure the coordination of the economy, resources, and environment from energy conservation and emission reduction perspectives. It considers both the cost of input factors and the benefit level of output. Output benefits include expected outputs (such as economic benefits) and unexpected outputs (such as pollution emissions). The essence of ecological efficiency is to achieve the maximum economic benefit output with the lowest resource and environmental input, which can comprehensively and effectively assess sustainable green development. Therefore, it can provide quantifiable standards for green development, intuitive effects for new economic development models, and the basis and guidance for policy adjustment [4].

Therefore, it is very important to explore whether implementing CETP will fully improve the quality and speed of emission reduction and promote the improvement of economic benefits through the market approach to achieve green and sustainable development that considers both the environment and the economy. If so, how large is its impact? What are the characteristics of impact? It is necessary to clarify these issues to realize the coordinated development of the global environment and economy and to provide information for follow-up policies.

Given this, based on the panel data of 284 cities from 2007 to 2016, this paper discusses whether CETP can improve urban ecological efficiency by using the difference in difference (DID) and propensity score matching difference method (PSM-DID) and empirically analyzes the mechanism of action of CETP and the heterogeneity of its role in different cities.

The novelty of this paper lies in the use of eco-efficiency as an indicator for measuring green and sustainable urban development, which enables economic development and environmental protection to be considered simultaneously. In addition, the possible contributions of this paper are as follows: (1) The use of eco-efficiency as an indicator of synergistic economic and environmental development expands the scope of CETP evaluation. (2) The heterogeneity of the role of CETP in different cities is analyzed. (3) This paper demonstrates, from a theoretical and empirical standpoint, that CETP can improve urban eco-efficiency by encouraging urban technological innovation and optimizing industrial structure.

The rest of this paper is arranged as follows. The second part is a literature review; the third part is theoretical analysis and hypothesis; the fourth part is the research design; the fifth part is the result analysis. The sixth part is heterogeneity analysis, the seventh part is mechanism analysis, and the last part is a discussion.

2. Literature Review

This paper reviews the literature from two perspectives: carbon emissions trading and eco-efficiency. This is shown in

Table 1. Literature Review on carbon emissions trading and eco-efficiency.

Carbon Emissions Trading
Author name	Time	Contribution
Xiao, Jin [5]	2004	For the first time, it is pointed out that the market for carbon emission trading is very unstable due to irresponsible participants and asymmetric information.
Liu Chuanming et al. [6]	2019	It shows that small and medium-sized enterprises with surplus carbon emission rights can use the benefit guidance effect of market players, the encouragement effect of enterprise technological innovation, and the support effect of local government policies to reduce carbon emissions.
Dong Xuan et al. [3]	2020	Points out that the pilot policy of carbon emission trading measures can significantly reduce carbon dioxide emissions.
Li, Zhiguo et al. [7]	2021	In the policy measures of carbon emission trade, there is a spillover effect for air emission reduction.
Wang, Huiying et al. [8]	2021	The proposed policy measures of carbon emission trading can promote carbon emission reduction through the energy structure.
Liu, Baoliu [9]	2021	In pilot regions, carbon emissions equity trading policies can improve total factor productivity in the green economy.
Tian, Guiliang [10]	2022	It is pointed out that the emission reduction effect of carbon emission trading policies varies from region to region.
Qi, Yawei [11]	2022	The noted carbon emissions trading pilot policy promotes significant green innovation in regional pollution control companies in pilot countries.
Wang, Zhi [12]	2022	It is pointed out that a carbon emission trading policy can significantly improve green total factor productivity in pilot areas.
Ecological efficiency
Author name	Time	Contribution
Schaltegger and Sturm [13]	1990	The Concept of Ecological Efficiency.
Roger L. Burritt, Chika Saka [14]	2005	Measuring Ecological Efficiency by the Ratio Method.
Fu Lina [15]	2013	It is found that the influencing factors of eco-efficiency mainly involve technological innovation, economic development, industrial structure, population density, urbanization, and opening up.
Li Shenglan et al. [16]	2014	Measuring eco-efficiency using packet-end analysis.
Qu Xiao’e et al. [17]	2018	According to the study, China’s overall eco-efficiency follows a pattern of high efficiency in the east and low efficiency in the west, with two levels of differentiation.
Xu Chenglong et al. [18]	2020	Measuring Ecological Efficiency by Stochastic Frontier Analysis.
Zhang Xinlin [19]	2020	The study found that China’s eco-efficiency presents a distribution pattern of high in the east and low in the west.
Chen Minghua [20]	2020	Analyzing the Regional Difference and Spatial-Temporal Evolution of Eco-efficiency in the Yellow River Basin Using the Dagum Gini Coefficient and Geographical Detector.
He Weida [21]	2022	They argued that digital development had a significant impact on green eco-efficiency.
Zahnhar Duman [22]	2022	Pointed out that there were significant spatial spillover characteristics of the impact of green technological innovation on urban eco-efficiency.

.

In summary, most of the existing literature has only explored the impact of CETP on carbon emission reduction or economic development in a single way, and very little literature has studied the two in combination, so it is necessary to open up this research scope. In addition, benefiting from the research of many scholars on industrial structure and technological innovation in recent years provides valuable ideas for the research of this paper. Therefore, this paper will cut through technological innovation and industrial structure optimization to investigate whether CETP can effectively promote cities’ green and sustainable development.

3. Theoretical Analysis and Research Hypotheses

Based on existing policies and documents, this paper discusses the mechanism and characteristics of CETP on urban ecological efficiency from two aspects of technological innovation and industrial structure optimization and then formulates and finally tests the hypothesis in the empirical part.

A study by Han et al. [23] found that optimizing the industrial structure could significantly improve the eco-efficiency of local and adjacent provinces. Subsequently, a study by Tan Jing et al. [24] showed that the carbon trading system effectively optimized the industrial structure of pilot regions. More recently, a study by Zhao Chaoyang et al. [25] found that CETP promoted the optimization and upgrading of industrial structures. Therefore, this paper ventures to speculate that CETP can improve urban eco-efficiency by optimizing the urban industrial structure and, thus, urban eco-efficiency, and is driven by the following pathways: (1) CETP will release signals to stimulate social investment to focus on high-tech, low-carbon industries, which will greatly optimize industrial structure and thus reduce carbon emissions and provide new economic growth points. (2) CETP will lead to higher prices for production factors, and narrow living conditions will threaten the survival of pollution-intensive industries. Then, CETP will eliminate the old "three high" industries and promote the development of new low-carbon industries, thus continuously optimizing the industrial structure and thus improving the eco-efficiency of the city. Finally, CETP helps achieve the emission reduction target and promotes economic growth simultaneously. In summary, the following hypotheses are proposed in this paper:

Hypothesis 1.

CETP improves the eco-efficiency of cities by promoting the optimization of industrial structures.

Modern economic growth theories point out that the root of economic growth is technological innovation. Under the CETP policy, companies of a certain size have difficulty in changing their inherent production patterns and structures, which leads to increased pressure on their operations. The only way out of this dilemma is to innovate in technology to make existing production decarbonized and sustainable. Ultimately, this will promote the efficiency of green innovation, i.e., the realization of the Porter Hypothesis [26]. Recently, many scholars [9,27,28] have shown that CETP can significantly improve the efficiency of green innovation. Therefore, CETP will generate a wave of economic activities in the related STI fields, increase the investment of enterprises in green innovation, make the economic development of the city compatible with the ecological environment, and thus improve the ecological efficiency of the city. Based on this, the following hypotheses are proposed in this paper:

Hypothesis 2.

CETP improves urban eco-efficiency by promoting technological innovation in cities.

4. Study Design

4.1. Research Procedures

In this study, under the assumption of meeting the parallel trend, cities implementing CETP are considered experimental groups, and vice versa. Firstly, the net effect of CETP on urban ecological efficiency is judged by the DID method. This paper adopts the propensity score-matching method, which selects the city sample closest to the experimental group to conduct DID again to make the results more accurate and adds robustness tests such as placebo tests. In addition, given the heterogeneity of cities, this paper also discusses the different impacts of the policy on different types of cities. Finally, it is necessary to obtain the effect of the policy, but it is more realistic to explore its reasons. Therefore, this paper also analyzes and demonstrates the mechanism of CETP. The research procedure is shown in Figure 1.

4.2. Core Model Setting

To verify whether CETP can improve the eco-efficiency of Chinese cities, this paper draws on the experimental approach of Zhang Hua [29]. It considers the pilot program for trading carbon emissions as a quasi-natural experiment. This method has been widely used in policy evaluation in recent years [30,31,32]. It can effectively overcome the interaction between independent and dependent variables and avoid endogenous problems. Since the first batch of pilot cities was identified in 2011 and Fujian Province was only included in 2016, Fujian Province was used as the control group in the sample from 2007 to 2016. Since the pilot policy on carbon emissions trading was promulgated in late 2011, most of the pilot cities were officially launched in 2012, and a few of them were launched after 2012, so in the double-difference model, using 2012 as the implementation year can more truly examine the impact of the pilot policy on urban eco-efficiency. Therefore, the panel data of 284 prefecture-level cities in China from 2007 to 2016 were selected as the sample, and the first batch of “five cities and two provinces” was used as the experimental group, while the remaining cities were used as the control group to quantify the impact of CETP from the perspective of urban eco-efficiency. The following is the model:

$$ue{e}_{ij}={\alpha }_1+{\theta }_{ij}ifctp{c}_i*tppi{n}_j+{\theta }_1{X}_{ij}+{\lambda }_i+{\gamma }_j+{\eta }_{ij}$$

(1)

where $ue{e}_{ij}$ is the eco-efficiency level of the ith city in year j. $ifctp{c}_i$ is a dummy variable for CETP, assuming that this city is the pilot city published in the pilot program, the value must be 1, otherwise it is 0. As a dummy variable before and after the pilot policy reform (with 2012 as the pilot year), the value is 1 during the policy implementation period (2012 and beyond), otherwise it is 0.${\theta }_{ij}$ calculates the carbon emissions exchange in the pilot cities at the time of policy implementation, $X_{ij}$ is the control variable, ${\lambda }_i$ is the city fixed effect, ${\gamma }_j$ is the year fixed effect, and ${\eta }_{ij}$ is the random disturbance term.

4.3. Parallel Trend Test Model

The premise of unbiased DID estimation is that the hypothesis of a parallel trend between the experimental and control groups is satisfied. This paper’s parallel trend is that before CETP, and the ecological efficiency of the experimental group and the control group should have the same trend. Otherwise, the model will produce a biased estimate of the policy implementation results [33].

The experiment was conducted as follows: the four years before the launch of the pilot policy (i.e., 2007) was used as the base year for comparison, and interaction terms between the dummy variables for the four years before the launch of the pilot policy, year of launch, the year four years after the launch, and the corresponding dummy variables for the pilot policy were established, and the regression equations were as follows:

$$Y_{i.j}={\alpha }_2+{\displaystyle \sum _{t=1}^4{\theta }_{pre\_t}{T}_{pre\_t}+{\theta }_{cur}{T}_{cur}+{\displaystyle \sum _{t=1}^4{\theta }_{post\_t}{T}_{post\_i}+{\theta }_2{X}_{i.j}+{\lambda }_i+{\gamma }_j+{\eta }_{i.j}}}$$

(2)

where $T_{pre\_t}$,$T_{cur},\mathrm{and}{T}_{post\_t}$ represent the interaction between the dummy variables and the corresponding policy dummy variables for the years before (4 years only), at and after the official launch of the pilot policy, respectively, ${\theta }_{pre\_t},{\theta }_{cur},\mathrm{and}{\theta }_{post\_t}$ are the corresponding coefficients, respectively, while the other control variables are the same as in the regression equation.

4.4. Mediating Model

Based on Hypothesis 1 and 2, this paper draws on the Baron and Kenny [34] partial mediation approach in the paper by Wen Zhonglin et al. [35] to discuss how CETP improves the quality of the environment by promoting industrial structure upgrades and technological progress, with the transmission mechanism set as follows:

Step 1: Test the effect of establishing CETP on the mediating variables.

$$m{v}_{ij}\left(is,pd\right)={\sigma }_1+{\xi }_{ij}liipc+{\xi }_1{X}_{ij}+{\lambda }_i+{\gamma }_j+{\epsilon }_{ij}$$

(3)

Step 2: Testing the effects of mediating variables on urban eco-efficiency.

$$ue{e}_{mv}={\sigma }_2+{\xi }_{i.j}m{v}_{ij}\left(is,pd\right)+{\xi }_2{X}_{ij}+{\lambda }_i+{\gamma }_j+{\epsilon }_{i.j}$$

(4)

where $m{v}_{ij}\left(is,pd\right)$ denotes the mediating variable and the parameter choice is indicates advanced industrial structure; the choice $pd$ indicates technological progress. $ue{e}_{mv}$ is the interpreted variable. $liipc$ is an interactive item of the CETP policy. ${\xi }_{ij}$ is the interaction term coefficient. ${\xi }_{i.j}$ is the estimated coefficient of the mediating variable. ${\sigma }_1$ and ${\sigma }_2$ are constant terms. ${\xi }_1$ and ${\xi }_2$ are the estimated coefficient of the control variable. $X_{ij}$ is the control variable, ${\lambda }_i$ is the city fixed effect, and ${\gamma }_j$ is the year fixed effect. ${\epsilon }_{ij}$ and ${\epsilon }_{i.j}$ are the random disturbance term.

4.5. Sample Selection and Data Sources

This paper selects data from 284 prefecture-level cities and CETP regions in China from 2007 to 2016 as samples. The data of this paper are mainly from China’s traditional statistical institutions, such as the China National Bureau of Statistics, and various statistical yearbooks. A small part of data is from the Guotai’an Database. Figure 2 of the pilot policy visualization is shown below:

4.6. Variable Selection

(1)

Explained variable

The explained variable is urban eco-efficiency. This paper measures urban eco-efficiency by referring to the model in the article by Tone [36]. The SBM-DEA model with unexpected output can consider both economic benefits (expected output) and adverse environmental impacts (unexpected output), so it is suitable for analyzing the efficiency of green economic development. In addition, the model can effectively solve the problem of variable relaxation. It is shown as follows:

$$min{\rho }^{*}=\frac{1+\frac{1}{m}{{\displaystyle \sum }}_{m=1}^M\frac{s_m^x}{x_{jm}^t}}{1-\frac{1}{l+h}\left({{\displaystyle \sum }}_{l=1}^L\frac{s_j^y}{y_{jl}^t}+{{\displaystyle \sum }}_{h=1}^H\frac{s_h^b}{b_{jh}^t}\right)}$$

(5)

$$s.t\left\{\begin{array}{c}{x}_{jm}^t\ge {\displaystyle {\displaystyle \sum }_{j=1,j\ne 0}^n}{\lambda }_j^y{x}_{jm}^t+s_m^x\\ y_{jl}^t\ge {\displaystyle {\displaystyle \sum }_{j=1,j\ne k}^n}{\lambda }_j^t{y}_{jl}^t-s_l^y\\ b_{jh}^t\ge {\displaystyle {\displaystyle \sum }_{j=1,j\ne k}^n}{\lambda }_j^t{y}_{jh}^t-s_h^b\\ {\lambda }_j\ge 0,s_m^x\ge 0,s_l^y\ge 0,j=1,2,\dots ,n\end{array}\right.$$

(6)

where ${\rho }^{*}$ is the measured value of overall urban eco-efficiency; n denotes the number of decision units; m, h, and l denote slack variables for inputs, expected outputs, and non-expected outputs, respectively; $s_m^x, s_l^y,\mathrm{and}{s}_h^b$ represent slack variables for the corresponding investments, expected outputs, and non-expected outputs, respectively; $x_j^t, y_j^t, \mathrm{and}{b}_j^t$ represent investments, expected outputs, and non-expected outputs of the decision units at time t, respectively; λ is the weight of the decision units vector. The specific indicators are selected as shown in the table below.

By reviewing the relevant literature [37,38,39], the following variables were finally selected as indicators for calculating eco-efficiency in this paper. The inputs include capital, labor costs, and resource inputs; the unintended outputs include the amount of industrial wastewater, the amount of industrially produced sulfur dioxide, and the amount of industrially produced flue gas emissions; the intended outputs are expressed in terms of gross local product. This is shown in

Table 2. Calculation of indicators of explanatory variables.

Tier 1 Indicator	Secondary Indicators	Indicator Components
Input indicator	Financial inputs	Capital stock, the number of employees
	Labor cost inputs	Number of employees at the end of the year
Output indicator	Resource inputs	Total urban electricity consumption
	Unintended output	Volume of industrial wastewater, volume of industrial production of sulfur dioxide, the amount of industrial production flue gas emissions
	Expected production	Gross local product

:

(2)

Explanatory variables

The explanatory variable is whether it is the city where CETP is located. It is a dummy variable. For cities in the policy implementation area, the value must be 1 in the year of policy implementation or after; otherwise, it is 0.

(3)

Mediating variables

Based on assumptions 1 and 2, referring to the practice of Xiping Liu et al. [40], the number of patents granted by each prefecture-level city was selected as the technological innovation level of each city. According to the practice of Malin Song and others [41], the ratio of the added value of the secondary industry to the added value of the tertiary industry in each city was selected to measure the optimization level of the industrial structure.

(4)

Control variables

For control variables, the following variables were selected through a literature review:

(1) Zhang Wei et al. [42] pointed out that coordinated ecological and environmental growth is mainly driven by economic development. Therefore, economic development is closely related to eco-efficiency, and the GDP per capita of each city, indicating the level of economic development, was selected as a control variable in this paper.

(2) The results of Wang Rong et al. [43] show that foreign direct investment significantly impacts regional eco-efficiency in China, the indirect impact is greater than the direct impact, and there is heterogeneity between regions. Therefore, this paper selects the ratio of real FDI to GDP as a control variable.

(3) A study by Wang Wenbin et al. [44] pointed out that S&T investment can positively affect regional eco-efficiency. Therefore, in this paper, the ratio of science and technology expenditure to the GDP of each city was selected as a control variable.

(4) A study by Li Yan et al. [45] found that an increase in the level of government intervention can positively affect the eco-efficiency of cities. Therefore, this paper uses the ratio of fiscal expenditure to GDP as a control variable.

4.7. Data Status

Considering the data availability of some cities and the integrity of the overall data, we finally chose to delete the data of some prefecture-level cities, perform linear interpolation on the remaining partial, incomplete data, and perform logarithmic processing on the non-ratio data to eliminate the impact of heteroscedasticity. The descriptive statistics and variable definitions of the final data are shown in

Table 3. Descriptive statistics and variable definitions.

	Variable	Variable Name	Variable Definition	N	Mean	sd	Min	Max	Unit
Explained variable	uee	Eco-efficiency	Calculated by SBM-DEA model	2840	0.439	0.214	0.182	1.209	No
Explanatory variable	liipc	Carbon trading city pilot	Whether it is the city where CETP is located	2840	0.0651	0.247	0	1	No
Intermediate variable	is	Advanced level of industrial structure	The ratio of the added value of the secondary industry to that of the tertiary industry	2840	1.366	0.846	0.201	10.16	Percentage
	pa	Technological advances	Number of patents granted by each prefecture level city	2840	3154	8184	5	101,864	Individual
Control variable	pcgdp	Level of Economic Development	GDP per capita	2840	5.17	3.51	0.384	4.68	Yuan
	nofdie	Foreign Direct Investment	Ratio of real foreign direct investment to GDP	2840	76.38	351.0	1	6007	Individual
	pofe	Level of Scientific Research Expenditure	Ratio of science and technology expenditure to GDP	2840	0.160	0.105	0.00161	2.702	Individual
	posrf	Level of government control	Ratio of fiscal expenditure to GDP	2840	0.00241	0.00245	0.298	0.0503	Individual

.

5. Result Analysis

5.1. Baseline Regression

In order to consider the impact of CETP on urban eco-efficiency, the stepwise regression method is introduced in this chapter to regress the benchmark model. The basic idea of stepwise regression is to introduce variables into the model one by one, run a t-test after each explanatory variable is introduced, and then run a t-test for the explanatory variables that have been selected one by one. When the initially introduced explanatory variables become no longer significant due to the introduction of later explanatory variables, it indicates that the original explanatory variables are unsuitable for regression. The results are presented in

Table 4. Baseline regression results.

Variable	(1)	(2)	(3)	(4)	(5)
liipc	0.0664 ***	0.0624 ***	0.0617 ***	0.0618 ***	0.0628 ***
	(0.0218)	(0.0214)	(0.0211)	(0.0212)	(0.0211)
pcgdp		55.46 **	57.46 **	59.47 **	60.06 **
		(26.98)	(26.95)	(27.11)	(27.00)
nofdie			3.27 × 10−5	3.26 × 10−5	3.31 × 10−5
			(2.57 × 10−5)	(2.58 × 10−5)	(2.54 × 10−5)
pofe				0.0432	0.0466
				(0.0883)	(0.0900)
posrf					−0.964
					(1.608)
Constant	0.435 ***	0.406 ***	0.403 ***	0.395 ***	0.396 ***
	(0.00142)	(0.0142)	(0.0151)	(0.0213)	(0.0219)
Fixed city	Yes	Yes	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes	Yes	Yes
Observations	2840	2840	2840	2840	2840
R-squared	0.721	0.722	0.722	0.722	0.722

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

, with items (1), (2), (3), (4), and (5) indicating the regression results when different numbers of control variables are added. With the same restrictions on city-fixed effects and year-fixed effects, the estimated coefficients of variation are significantly positive, and the results indicate that the impact of CETP on urban eco-efficiency is significantly positive, which indicates that market-based environmental regulation has a strong incentive effect on urban economic growth and eco-environmental protection. The magnitude and significance of the estimated coefficients of LIIPC did not change significantly when control variables such as economic development level were gradually added, further indicating that the pilot policy can promote the green development of the urban economy. This is similar to the findings of Lu Min and Wu Wenjie et al. [46,47].

As for the control variables, the effect of economic development level on eco-efficiency is more significant, while all other control variables are insignificant, which is broadly consistent with the findings of Yang Yong et al. [48] and Xing Zhengcheng et al. [49].

5.2. Parallel Trend Test

Figure 3 shows the results of the parallel trend test. It can be seen from the figure that the estimated coefficients of urban ecological efficiency were very small between the first four years and the first two years of the time point for the investigation of CETP, and remained insignificant at the 95% level confidence interval. However, in the year before the investigation time, that is, 2011, the estimated coefficients temporarily became positively significant. It was analyzed that this is because there are many samples of low-carbon pilot cities in the experimental group. The pilot policy was issued in 2010 and officially acted in 2011. Its basic purpose is also for the green and sustainable development of cities. In addition, considering that the official promulgation time of CETP is 2011, this paper takes 2012 as the official inspection time point, which may also be due to the selection of inspection time point. In general, the parallel trend test of the experimental group and the control group basically passed. After reviewing the time point, the positive impact of carbon emission trading policy began to appear, and there was a trend of continuous improvement, which was consistent with the results of benchmark regression.

5.3. Testing the Results Based on the PSM-DID Method

Although the DID method can obtain the net effect of CETP, the error caused by sample selection will be inevitable, so we chose the PSM-DID method to strengthen the robustness of the benchmark regression conclusion. The annual adjustment method based on Liu Ye et al. [50] was adopted to find the appropriate control group based on the nearest neighbor-matching method. The matching results show that except for pofe variables, the control and experimental groups’ control variables have no significant difference, and the standardization deviation is significantly reduced, which proves the effectiveness of the PSM-DID method used in this paper. The matching is shown in

Table 5. Results after matching.

Variable	Status	Mean Value	Mean Value	Mean Value	Mean Value	Mean Value	Mean Value
pcgdp	Before matching	0.00049	0.00048	2.7	38.2	0.33	0.741
	After Match	0.00049	0.00048	1.6		0.21	0.831
nofdie	Before matching	80.529	60.809	10.5	34.0	1.30	0.192
	After Match	80.529	67.523	6.9		0.89	0.376
pofe	Before matching	0.13797	0.12619	20.4	−13.8	2.52	0.012
	After Match	0.13797	0.12457	23.2		3.02	0.003
posrf	Before matching	0.0027	0.00248	8.0	80.3	1.00	0.318
	After Match	0.0027	0.00266	1.6		0.19	0.853

.

After matching for the control group, further regression analysis was performed, and the results are shown in

Table 6. PSM-DID regression results.

Variable	uee	uee	uee
	(1)	(2)	(3)
liipc	0.0867 ***	0.0869 ***	0.0867 ***
	(0.0302)	(0.0309)	(0.0307)
pcgdp		39.43	37.66
		(50.27)	(49.15)
nofdie		−1.50 × 10−5	−1.51 × 10−5
		(3.45 × 10−5)	(3.46 × 10−5)
pofe		0.0610
		(0.167)
posrf		−0.0973	0.420
		(4.423)	(4.013)
Constant	0.410 ***	0.384 ***	0.392 ***
	(0.00775)	(0.0392)	(0.0305)
Fixed city	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes
Observations	550	550	550
R-squared	0.809	0.810	0.810

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

. Because the matching for the pofe variable was less effective, excluding the pofe variable performed one additional regression. Ultimately, the results using PSM-DID are mainly consistent with those from the benchmark regression; the regression coefficients of the policy dummy variables of three models (1), (2), and (3) are significantly positive. Thus, market environment regulation promotes ecological efficiency and confirms the robustness of the benchmark regression results.

5.4. Other Robustness Tests

(1)

Placebo test

A placebo test was then conducted further to strengthen the credibility of the baseline regression results. Referring to Fang Hui’s [51] approach of randomly selecting experimental groups, 39 cities were first randomly selected from 284 cities as the “pseudo-experimental”, and it was assumed that these 39 cities were all pilot cities for CETP. In contrast, the remaining cities were used as the control group, thus constructing a city-randomized test. The above process was repeated 500 times to obtain 500 regressions to plot the kernel density distribution of the estimated coefficients against the p-values, as shown in Figure 4. This was used to verify whether the eco-efficiency of Chinese cities was significantly influenced by factors other than CETP. As seen in Figure 4, the regression coefficients of the original benchmark are all 0.063 (the red vertical line). However, the estimated coefficients of some of the randomly obtained group experiments are concentrated around 0. Their p-values are mostly greater than 0.1, indicating that establishing CETP has no significant effect on the randomized experimental group, thereby increasing the robustness of the benchmark regression’s core findings.

(2)

Excluding the effects of other policies

Is the policy effect of CETP on urban eco-efficiency a “net effect”? Are cities’ eco-efficiency also affected by other related policies? This may interfere with the policy effects estimated by the model. By sorting out the relevant policies from previous years and integrating them with the relevant literature, the low-carbon cities pilot policy and the innovative cities pilot policy were finally selected to construct the control policy groups, respectively. The results excluding cities included in these two pilot policies are (1) and (2) in

Table 7. Robustness tests: excluding other policy disturbances and expected effects.

Variable	Exclusion Policy	Exclusion Policy	Expected Effect	Expected Effect
	(1)	(2)	(3)	(4)
liipc (2011)			0.0727 ***	0.0693 ***
			(0.0188)	(0.0183)
liipc (2010)			0.00531	0.00577
			(0.0131)	(0.0134)
liipc	0.311 ***	0.266 ***
	(0.0372)	(0.0509)
Control variable	Uncontrolled	Control	Uncontrolled	Control
Fixed city	Yes	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes	Yes
Observations	1920	1920	2840	2840
R-squared	0.742	0.747	0.721	0.723

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

, respectively.

According to items (1) and (2) in Table 7, the interaction coefficient of CETP is significantly positive, which is consistent with the results of benchmark regression. Therefore, after excluding the interference of other policies, CETP can still significantly improve the ecological efficiency of pilot cities.

(3)

Policy exogeneity

Following the removal of the influence of other policies on the results, the model results should be checked to ensure that they are not influenced by policy exogeneity, that is, that the control group cities do not form reasonable expectations with the experimental group cities prior to the implementation of CETP. This regression uses 2011 as the year of policy implementation to better match the expected effects of the actual policy, as the year of implementation is 2011. The dummy interaction for the year prior to the implementation of CETP, liipc (2010), was then added to the baseline regression, and the 2011 policy dummy interaction, liipc (2011), was substituted for the original 2012 policy interaction, liipc, as the explanatory variable. The regression results without and with control variables are shown in columns (3) and (4) of Table 7, in that order.

As can be seen from columns (3) and (4) of the table, the regression coefficient of the core explanatory variable liipc (2011) remains significantly positive before and after the inclusion of the control variables, while the estimated coefficient of liipc (2010) in the previous year is insignificant, which indirectly confirms that CETP does not have predictive effects, thus enhancing the reliability of the original baseline recovery findings.

6. Heterogeneity Analysis

6.1. City Size Heterogeneity

Differences in city size can lead to significant differences in resources between cities, thus affecting the credibility of the baseline regression results. On the one hand, small and medium-sized cities are less talented than large cities, and their infrastructure could be better, creating significant differences in the composition of the social landscape, thus creating potential differences between cities. The impact of such differences on the experimental results is challenging to identify. On the other hand, large cities’ industrial structure and openness have certain advantages over small and medium-sized cities. The efficiency of their resource allocation and the rationality of their social operations are significantly better than those of small and medium-sized cities, so the final results are also very different when dealing with economic development and environmental protection. In this chapter, two types of cities are distinguished, namely large cities, medium cities, and small counties, according to the document issued by the State Council of the Communist Party of China. The regressions are carried out in turn. The regression results are presented in columns (1) and (2) of

Table 8. Regression results of city size heterogeneity and city level heterogeneity.

Variable	Large and Medium-Sized Cities	Small Cities	Non-Core Cities	Core Cities
liipc	0.238 ***	0.0620 ***	0.0552 **	0.236 ***
	(0.0488)	(0.0160)	(0.0221)	(0.0748)
Control variable	Control	Control	Control	Control
Fixed city	Yes	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes	Yes
Observations	1090	1090	2500	140
R-squared	0.785	0.656	0.730	0.709

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

.

As seen from the table, the coefficient of the CETP interaction term is positive and significant for both large and small cities. The coefficients for large and medium-sized cities are more significant, which verifies that their resource use efficiency is better than that of small cities and can be more advantageous in improving eco-efficiency.

6.2. Heterogeneity at the City Level

In addition, cities at different levels have different government control methods, and core cities need to take this more into account when making government decisions, thus generating some potential operational differences from non-core cities and generating bias in the regression. Therefore, this paper divides the sample of 284 cities into core and non-core cities, where core cities are provincial capitals, sub-provincial cities, and municipalities directly under the central government. The results are shown in columns (3) and (4) of Table 8.

As seen from the table, when the government regressions are run separately for central and non-core cities, the interaction term of CETP is still very significant, and the regression coefficients are all positive, indicating that the baseline regression results are still convincing when differentiating between city levels.

6.3. Heterogeneity of Urban Location

The influence of geographic location on urban development is self-evident, whether in terms of economic development or cultural heritage. These may indirectly affect the effect of CETP to varying degrees, so this paper will further examine the impact of CETP on the eco-efficiency of cities in different locations. Based on this, this paper divides the 284 sample cities into three regional samples, East, Central, and West, based on the division of China’s national territory. Furthermore, if the experimental and control group samples allow, the regressions exclude policy interference. The final regression results are shown in columns (1), (2), and (3) of

Table 9. Regression results of geographic location heterogeneity and heterogeneity in the extent of urban resource endowment in cities.

Variable	East	Central	Western	Resource-Based	Non-Resource Based
liipc	0.234 ***	0.0869 **	0.122 ***	0.0525	0.0643 ***
	(0.0517)	(0.0352)	(0.0347)	(0.0497)	(0.0225)
Control variable	Control	Control	Control	Control	Control
Fixed city	Yes	Yes	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes	Yes	Yes
Observations	500	1090	1090	610	2230
R-squared	0.813	0.785	0.656	0.706	0.726

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

.

The results show that the regression coefficients for the pilot cities in the three divisions, East, Central, and West, are all positive and show significance. The coefficients are more significant in the eastern region and the second largest in the western region. This shows that the effect of China’s CETP on the environment in each geographic region is noticeable, which is also inseparably related to the level of economic and social development in each region.

6.4. Heterogeneity of Urban Resources

Based on the factor endowment theory proposed by Heckscher and Olin, the development of cities is also influenced by the degree of resource endowment. In other words, resource-rich cities have relatively lower market costs in the development process. In contrast, CETP, as a form of market environmental regulation, is undoubtedly influenced by the degree of resource endowment. For this reason, this paper divides the 284 cities in the sample into 61 resource-based cities and 223 non-resource-based cities according to the divisions in the Chinese State Council plan and runs the regressions separately. The regression results are presented in columns (4) and (5) of Table 9.

As seen from the table, the impact of CETP on resource-based cities is not significant, while the impact on non-resource-based cities is very significant. This suggests that resource-based cities are more dependent on their resource-rich advantages than non-resource-based cities and that CETP, as a form of market-based environmental regulation, is a barrier to resource-based cities that rely on resources for their development. Non-resource-based cities are forced to maintain their development through technological innovation and industrial restructuring due to resource disadvantages.

7. Analysis of Impact Mechanisms

As seen from

Table 10. Intermediary analysis results.

Variable	is	uee	pa	uee
	(1)	(2)	(3)	(4)
liipc	0.129 **		4106 **
	(0.0648)		(1739)
is		−0.0236 ***
		(0.00625)
pa				4.51 × 10−6***
				(6.37 × 10−7)
Control variable	Uncontrolled	Control	Uncontrolled	Control
Fixed city	Yes	Yes	Yes	Yes
Fixed year	Yes	Yes	Yes	Yes
Observations	2840	2840	2840	2840
R-squared	0.824	0.036	0.802	0.049

z-statistics in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1, the same below.

, the coefficients of the mediating variables all show significance, and the policy interaction term has a positive effect on the mediating variables, indicating that CETP indirectly improves urban ecological effects through technological progress and industrial structure upgrading. Hypotheses 1 and 2 are confirmed, consistent with the findings of Min Lu [46] and Wenjie Wu et al. [47]. Implementing low-carbon pilot policies promotes regional economic growth by promoting technological innovation on the one hand and reducing regional pollution emissions through industrial structure upgrades on the other, thus enhancing eco-efficiency.

8. Discussion

8.1. Conclusions

To better explore whether CETP can promote urban green sustainable development, this paper uses ecological efficiency as an evaluation indicator. It uses the samples of 284 prefecture-level cities in China from 2007 to 2016. Through DID and PSM-DID methods, this paper finally draws the following conclusions: ① CETP can significantly enhance urban eco-efficiency. The results persist after long-term robustness tests with parallel trend tests, PSM-DID models, and placebo tests. This means that implementing CETP can effectively promote the green and sustainable development of the city. ② The results of heterogeneity analysis show that compared with small and medium-sized cities, CETP can improve the ecological efficiency of large cities more significantly. Compared with the central and western regions of China, the ecological efficiency of CETP in the eastern coastal areas has been improved more. In addition, CETP can not improve the ecological efficiency of resource-based cities, but can improve the ecological efficiency of non-resource-based cities. ③ The results of the mechanism test show that CETP can improve urban ecological efficiency by promoting urban technological innovation and optimizing industrial structure.

8.2. Policy Suggestion

First, there is no doubt that CETP can effectively improve urban efficiency to achieve green and sustainable urban development. Therefore, a more comprehensive carbon trading system should be established, and the pilot scope of CETP should be continuously expanded, which will help China achieve the goal of “carbon neutrality” and high-quality economic development faster with the help of the economic effect and emission reduction effect of CETP.

Secondly, in the process of continuously expanding the scope of the CETP pilot, because there are significant differences between different cities, it is essential to make full use of the successful experience of early pilot cities and improve policies in combination with the city’s conditions. For cities in central and western China, we should first lay a solid economic foundation, and for cities in eastern China, we should invest more in environmental protection publicity. Compared with small and medium-sized and non-provincial capital cities, more investment should be spent on large and provincial capital cities, maximizing the policy benefits of CETP. For cities that rely on their resources to develop, they should restrict the excessive exploitation of resources and explore new development models.

Finally, the government of CETP cities should support and guide local enterprises to innovate more because CETP promotes sustainable urban development by promoting technological innovation. Therefore, we should adopt more generous enterprise innovation incentives, which will help CETP add value to its benefits. In addition, it is essential to limit the development of pollution-intensive industries. We should develop more clean energy and low-carbon industries, which will help transform and upgrade the industrial structure and better realize cities’ green and sustainable development.

8.3. Research Limitations and Prospects

(1)

Research limitations

To study whether CETP can promote the sustainable green development of cities, this paper selects ecological efficiency as a measurement indicator. Still, there is more than one way to measure whether cities can achieve sustainable green development, and there is more than one way to calculate ecological efficiency. Different indicator selection and calculation methods for ecological efficiency may lead to different research conclusions.

(2)

Research prospect

Although the research in this paper discusses whether CETP can effectively promote the sustainable development of pilot cities through ecological efficiency, it needs to study the spatial effect of the policy, and the selection of research indicators is subjective and not unique. Therefore, it is urgent to call for more relevant and effective indicators to broaden the evaluation scope of CETP and analyze the spatial spillover effect of CETP.