The Synergistic Effect of the Carbon Emission Trading Scheme on Pollution and Carbon Reduction in China’s Power Industry

Abstract

The synergistic effect of pollution and carbon reduction can alleviate the dual pressure of improving environmental quality and reducing greenhouse gas emissions in China. The carbon emission trading scheme (CETS) is a crucial market-based tool for carbon emission reduction, and understanding its synergistic impact on air pollution control is essential. Based on data from 30 provincial panels in China spanning from 2007 to 2021, we employ the difference-in-differences (DID) method to analyze the synergistic effects of the carbon emission trading plan in the power industry and its influence mechanisms are examined. We observe that the CETS significantly enhances both pollution and carbon reduction in China’s power sector, particularly demonstrating effective synergy in reducing CO₂, SO₂, and PM_2.5 emissions. Furthermore, mechanism analysis reveals that the CETS achieves joint emission reductions by lowering energy consumption, influencing the power industry’s generation structure, promoting technological innovation among enterprises, and thereby realizing synergistic pollution and carbon reduction effects in China’s power sector. Heterogeneity analysis shows that regions with limited power facility, low electricity generation, and small economic scale exhibit the most pronounced synergistic benefits from pollution and carbon reduction efforts.

1. Introduction

China, being the largest consumer of energy and the top emitter of carbon dioxide globally, also ranks first in the emission of sulfur dioxide, nitrogen oxides, and various other air pollutants [1,2,3]. Based on this, China has set ambitious carbon reduction targets for 2030 [4], with the goal of reaching peak carbon emissions by 2030 and attaining carbon neutrality by 2060 [5]. The implementation plan for reducing pollution and carbon emissions, along with enhancing synergistic efficiency, aims to achieve notable outcomes by 2030. It emphasizes the coordinated efforts in key areas of air pollution prevention and control to promote both carbon peak and air quality improvement in China [6]. Relevant studies confirm that air pollutants and greenhouse gasses share common origins [7,8,9]. During economic development, the combustion of fossil fuels releases significant amounts of air pollutants such as particulate matter, sulfur dioxide, and nitrogen oxides into the atmosphere, alongside greenhouse gasses like carbon dioxide, methane, and nitrous oxide [10,11]. Production and daily activities also contribute to substantial emissions of air pollutants and greenhouse gasses, including agricultural production and household waste disposal [12,13,14,15]. Consequently, collaborative governance of pollution and carbon reduction is essential for China as it navigates a pivotal phase in its transition to a low-carbon economy.

Existing studies have explored the synergistic effect between reductions in air pollutant emissions and greenhouse gasses. Liu et al. (2021) [16], who used a DID model, indicate that China’s carbon emission trading system effectively reduces carbon dioxide emissions while simultaneously coordinating PM_2.5 reductions. This is accomplished by incentivizing businesses to adopt emission reduction strategies and modify their industrial structures, with the most significant impacts noted during the summer months. Dong et al. (2019) [17] appropriately extended the basic Kaya identity model, considering the synergistic effect of carbon dioxide emissions on PM_2.5 emissions, and concluded that reducing carbon emissions significantly reduces sulfur emissions. Some scholars also focus on various environmental policies and analyze their synergies, including the carbon emission trading system [18,19], low-carbon city policies [20], and pollution control regulations [1,21].

As a market mechanism for reducing carbon dioxide emissions, the carbon emission trading scheme (CETS) is widely regarded by scholars as the most effective means to achieve greenhouse gas emission reduction targets at lower social costs [22,23,24,25,26,27,28]. Feng et al. (2024) [29] used the STSA algorithm to predict carbon emissions. Huang et al. (2023) [30] studied the Maritime Emissions Trading Scheme (METS). In recent years, the synergistic effects of emission reductions from the CETS in China have received significant attention, yet consensus remains elusive. Several scholars have validated the synergistic impact of carbon trading on reducing emissions of additional pollutants. Liu et al. (2021) [16] analyzed the impact of the carbon trading pilot on PM_2.5 using the DID method and found that the policy reduced PM_2.5 concentrations in the pilot area by 4.8%. However, studies have found that certain emission reduction measures aimed at a single pollutant may inadvertently lead to increases in other air pollutants [31,32,33,34]. Moreover, some researchers contend that the synergistic effect of carbon trading primarily occurs through the concurrent reduction of SO₂ and carbon dioxide emissions [35,36,37], attributable to the high proportion of sulfur elements in Chinese fossil fuels compared to other pollutants [35]. Thus, significant debate persists about the synergistic effects of implementing the CETS on different atmospheric pollutants and carbon dioxide emissions. Existing studies suggest that the CETS has the potential to promote synergistic emissions reductions, but uncertainty and controversy about its effectiveness remain. Future studies are needed to more deeply explore the interactions between different pollutants and how the policy design can be optimized to achieve broader environmental benefits. Furthermore, researchers should focus on the synergistic effects of CETS across various regions and industries, so as to provide better technical support for the implementation of national and regional policies.

In 2017, China’s National Development and Reform Commission released the Plan for the Construction of the National Carbon Emission Trading Market, specifically aimed at the power generation sector, marking the official inception of the national carbon market [38]. The power sector is a major contributor to air pollutants and greenhouse gas emissions across production and consumer industries, generating significant amounts of air pollutants and carbon dioxide emissions that contribute to environmental degradation and climate change. In 2020, emissions from coal-fired power generation in China comprised 43.5% of CO₂, 13.1% of NO_x, 13.2% of SO₂, and 3.8% of PM from total industrial emissions, respectively [39,40,41]. Historically, most studies have concentrated on industries that generate the highest levels of carbon emissions [42,43], neglecting the power sector which, alongside transportation and industry, comprises China’s top three carbon-emitting sectors.

Building on this, this study centers on the carbon emission trading mechanism, using the electric power industry as its research subject. It analyzes carbon dioxide emissions and major atmospheric pollutants (SO₂, NO_x, PM_2.5) data from 2007 to 2021 across provinces and cities, employing the difference-in-differences method to quantify the synergistic emission reduction effects of the CETS. Furthermore, it examines the pathways through which the CETS affects both pollution reduction and carbon reduction within the power industry. Results indicate that under the CETS implementation, a synergistic effect exists between carbon emissions and atmospheric pollutant emissions in China’s power sector, with the strongest synergy observed between CO₂ and PM_2.5. The synergistic reduction in pollution and carbon emissions in China’s power industry primarily occurs through pathways affecting energy consumption, power generation outcomes, and technological innovation. Heterogeneity analysis reveals that variations in power facility levels significantly impact CO₂ and SO₂ emissions across different regions. Differences in electricity generation also notably affect SO₂ emissions, while variations in economic scale affect SO₂ and NO_x emissions. Hence, regions with lower power facility levels, electricity generation, and economic scale exhibit the most significant synergistic effects. These findings can facilitate coordinated monitoring of air pollution control and carbon dioxide emission reduction across various industries.

The primary contributions of this paper are as follows. First, regarding the research subjects, existing studies on synergistic emission reduction effects primarily concentrate on the national level or specific regions, with only a limited number examining particular industries [43,44,45]. The research on collaborative emission reduction in the power industry with huge emissions still needs to be improved. This study addresses this gap by focusing specifically on China’s power industry, providing insights into emission reduction technology options and enhancing the implementation of the CETS across various industries. Second, existing research predominantly examines the CETS’s impact on carbon emission reduction [46,47,48], overlooking its synergistic effects on both carbon emissions and air pollutants. This paper examines how the CETS facilitates synergistic reductions in pollution and carbon emissions in China, emphasizing the importance of energy consumption, power generation structure, and technological innovation within the power industry. Third, many power industry enterprises may increase energy consumption and pollutant production due to installing desulfurization, denitrification, and dust removal facilities [49]. This paper tackles the debate over the synergistic effects of pollution and carbon reduction in the power industry by analyzing the impact of the CETS, confirming its beneficial role in promoting both pollution and carbon reduction. Finally, we examined the varied impact of the CETS on the synergy between pollution and carbon reduction in China’s power industry. By examining power facility levels, electricity generation, and economic scale across regions, we identified significant synergistic effects in regions with lower facility levels, electricity generation, and economic scale. This insight guides the selection and prioritization of CETS implementation areas.

The rest of this paper is structured as follows. Section 2 presents the research background and theoretical hypotheses. Section 3 outlines the research design. Section 4, Section 5 and Section 6 detail the empirical results; and the final section offers conclusions and policy implications.

2. Research Background and Theoretical Assumptions

2.1. Policy Background

The concept of the carbon emissions trading mechanism, aimed at reducing carbon dioxide emissions [50], originated in the early 1990s. Coase (1960) [51] argued that while carbon taxes can reduce carbon dioxide emissions to some extent, government intervention disrupts market resource allocation. He proposed an alternative: treating negative externalities as commodities tradable in a market with clear property rights, satisfying the interests of all relevant parties. This approach allows negative externalities to be traded as commodities with clear property rights, satisfying the interests of all parties involved in the market mechanism. This method not only fulfills the market’s resource allocation function but also enhances overall societal welfare. The carbon emissions trading system emerged from this theoretical basis. As environmental issues have gained prominence, economists have increasingly explored market mechanisms to address them, including greenhouse gas emissions reduction.

China’s carbon emissions trading system started later and is smaller in scale compared to other CETS markets worldwide, but it is rapidly developing [52]. China initiated its carbon emissions trading pilot program in 2011. Launched in 2013, the pilot program encompassed seven provinces and municipalities: Beijing, Tianjin, Shanghai, Chongqing, Guangdong, Hubei, and Shenzhen, following comprehensive considerations of regional conditions and foundational work. In 2017, China’s National Development and Reform Commission (NDRC) issued the “National Carbon Emissions Trading Market Construction Programme (Power Generation Sector),” officially establishing the national carbon market [53]. In 2015, China introduced the concept of “coordinated control of air pollutants and greenhouse gas emissions” in the Air Pollution Prevention and Control Law for the first time [54]. The “Implementation Programme of Pollution Reduction, Carbon Reduction and Synergistic Enhancement” proposes that by 2030, synergistically promoting carbon peaking and enhancing air quality in critical areas of China’s air pollution prevention and control is expected to yield significant results [6]. Currently, China’s ecological civilization construction faces two strategic imperatives: essentially enhancing the ecological environment and reaching carbon neutrality. Advancing synergistic pollution and carbon reduction has emerged as a crucial strategy for the comprehensive green transformation of China’s economic and social development in this new phase of growth.

2.2. Theoretical Assumptions

2.2.1. Synergistic Effect of the CETS on Pollution and Carbon Reduction

China’s carbon emissions trading market has effectively contributed to a reduction in air pollution and carbon dioxide emissions through a variety of means, including economic incentives, technological innovation, data transparency, international cooperation, and legal and policy support. Hu et al. (2020) [55] pointed out that as the cost of carbon emissions rises, firms have an incentive to invest in more efficient energy technologies and low-carbon technologies, such as renewable energy and energy efficiency enhancement technologies, thereby reducing their dependence on fossil fuels and their carbon emissions. Moreover, the carbon market requires companies to report their carbon emissions data on a regular basis and through third-party verification, which increases the transparency and accuracy of the data [56]. Greenhouse gasses and atmospheric pollutants originate predominantly from fossil fuel combustion [57,58], necessitating integrated control strategies rather than isolated approaches [1,21]. The alignment of policy frameworks and strategies for reducing carbon emissions underscores their fundamental consistency [59]. Beyond end-of-pipe dust control measures, policies emphasize coordinated emissions management [60]. This integrated approach represents a critical climate policy for effectively reducing carbon emissions [50,61,62,63,64,65]. Given its status as China’s primary carbon emitter, the power industry significantly contributes to air pollution [66,67,68,69,70,71]. Building on the preceding analysis, we propose Hypothesis 1:

Hypothesis 1.

The implementation of the CETS may produce a synergistic effect on pollution and carbon reduction within China’s power sector.

2.2.2. Mechanisms of CETS Pollution and Carbon Reduction

The CETS aims to reduce overall energy consumption and achieve “double control” over both total energy consumption and its intensity. Enterprises participating in the CETS initially implement measures to curtail carbon emissions to meet allocated quotas [72,73]. Energy consumption is the primary source of CO₂ and air pollutants [58,74], thus reducing energy usage correlates with decreased emissions of air pollutants [75]. As the largest energy consumer across various industries in China [76], the power industry’s adoption of a CETS has the potential to concurrently lower CO₂ and air pollutant emissions through reduced energy consumption. Building on the preceding analysis, we propose Hypothesis 2:

Hypothesis 2.

The implementation of the CETS facilitates synergistic pollution and carbon reduction in China’s power sector by lowering energy consumption.

Following CETS implementation, the cost of traditional fossil energy sources such as coal has escalated [77], compelling enterprises to increase their reliance on clean energy and decrease coal consumption to meet emission targets [19]. In the power sector, Pahle et al. (2011) [78] and Denny et al. (2009) [79] observed that the CETS introduction prompted power plants to phase out thermal power facilities due to increased costs. Consequently, with the CETS in place, the proportion of clean energy generation—such as hydropower, wind power, and solar power—will rise, while thermal power generation will decline [80], thereby optimizing the overall power generation structure of the region. China’s reliance on coal in its energy structure has resulted in a power generation landscape dominated by thermal power in the industry [81,82]. As thermal power is a significant contributor to carbon emissions and air pollutants [83,84], reducing its share will realize synergistic effects in pollution and carbon reduction. Building on the preceding analysis, we propose Hypothesis 3:

Hypothesis 3.

The implementation of the CETS achieves synergistic pollution and carbon reduction in China’s power sector by influencing the power generation structure.

The innovation compensation effect [85] of the CETS enables enterprises to gain a competitive advantage. To ensure long-term development, enterprises adopt innovative technologies to reduce carbon emissions. Technological innovation enhances energy efficiency [86], decreases per-unit emissions [87], optimizes energy consumption structures, and consequently reduces both carbon and air pollutant emissions [88,89,90]. Research indicates that the CETS stimulates technological innovation in the power industry, thereby enhancing energy structures and yielding positive economic and environmental benefits [45,78,79,91]. Building on the preceding analysis, we propose Hypothesis 4:

Hypothesis 4.

The implementation of the CETS achieves synergistic pollution and carbon reduction in China’s power sector by stimulating technological innovation.

3. Research Design

3.1. Methodology

Compared with previous models, the difference-in-differences (DID) model as a representative model is valid and reliable. The advantage of the DID model lies in its ability to control for non-time-varying heterogeneity and mitigate endogeneity due to omitted variables, with policy evaluations typically unaffected by reverse causality issues. Therefore, we use the CETS as a “natural experiment” to construct difference-in-differences models, using 2013 as the policy shock date. The seven pilot regions of China’s carbon emissions trading policy are designated as the experimental group, while other provinces and cities serve as the control group. The following model is then constructed.

$$Y_{it}={\beta }_0+{\beta }_1{N}_{it}{T}_{it}+\lambda X_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(1)

where i denotes the region and t denotes the year; the dependent variable $Y_{it}$ is the CO₂ or air pollutants emissions from the power sector in region i at time t, which are, respectively, expressed as CO₂ emissions, SO₂ emissions, NO_x emissions, and PM_2.5 concentrations; $N_{it}$ is a policy dummy variable, which is taken as 1 in the pilot region and 0 otherwise; $T_{it}$ is a time dummy variable that takes 0 before the policy occurs and 1 after; $N_{it}{T}_{it}$ is the interaction term of dummy variables $N_{it}$ and $T_{it}$, representing policy implementation; ${\beta }_1$ is the coefficient of interest, which measures the net effect of the policy, with negative coefficients indicating that a negative coefficient indicates that policy implementation reduces pollutant emissions and vice versa; $X_{it}$ represents a series of control variables, including industrial structure, urbanization level, economic development level, openness to the outside world, energy structure, R&D investment, environmental regulation, and power generation structure; ${\mu }_i$ denotes a regional fixed effect; ${\delta }_t$ denotes a time fixed effect; ${\epsilon }_{it}$ is the random perturbation term.

3.2. Data Description

By collecting, screening, and calculating data from sources such as China’s National Statistical Yearbook, the CEADs database, and the MEIC database, this study obtained the average annual emissions of CO₂ and three air pollutants (SO₂, NO_x, and PM_2.5) from the power industry across 30 provinces and municipalities in China (excluding the Tibet Autonomous Region) for the years 2007 to 2021, along with related control variables. CO₂ emissions from the power industry were calculated using the carbon emission factors established by the United Nations Intergovernmental Panel on Climate Change (IPCC). The specific definitions and sources of the data are detailed in

Table 1. Description of variables and data sources.

Number	Typology	Variable Name	Define	Source of Data
1	Explanatory variable	CO2 emissions	CO2 emissions from the power sector, in logarithms	China Carbon Accounting Databases (CEADs)
2		SO2 emissions	SO2 emissions from the power sector, in logarithms	MEIC database
3		NOx emissions	NOx emissions from the power sector, in logarithms	MEIC database
4		PM2.5 emissions	PM2.5 emissions from the power sector, in logarithms	MEIC database
5	Control variable	Industrial structure	Tertiary sector output/secondary sector output	China Statistical Yearbook
6		Urbanization level (of a city or town)	Urban population/total population	China Statistical Yearbook
7		Level of economic development	GDP per capita, logarithmic	China Statistical Yearbook
8		Degree of openness to the outside world	(Total exports and imports of goods × US dollar to renminbi exchange rate)/gross regional product	China Statistical Yearbook
9		Energy structure	Coal consumption/total consumption	China Statistical Yearbook
10		R&D intensity	Internal expenditure on R&D/gross regional product	China Statistical Yearbook
11		Environmental regulation	Completed investment in industrial pollution control/industrial added value	China Statistical Yearbook
12		Generation structure	Thermal power generation/total power generation	China Statistical Yearbook

. To mitigate the impact of heteroscedasticity on the estimation results, all variables (except ratio and dummy variables) were log-transformed.

3.3. Descriptive Statistics

Table 2. Descriptive statistics of variables.

Variant	Experimental Group					Control Group
	Sample Size	Average Value	(Statistics) Standard Deviation	Maximum Values	Minimum Value	Sample Size	Average Value	(Statistics) Standard Deviation	Maximum Values	Minimum Value
V1	105	2501.164	1541.661	6440.087	77.872	345	2364.195	1737.911	7166.941	0.364
V2	105	163,880.373	205,365.410	1,130,800.673	204.870	345	195,850.011	212,043.218	1,275,800.277	2926.197
V3	105	165,726.455	128,133.676	576,852.586	25,205.490	345	228,662.623	199,705.109	991,383.810	12,789.192
V4	105	15,865.344	13,761.990	81,725.376	35.824	345	24,585.120	22,679.221	112,960.050	630.388
V5	105	1.550	1.074	5.297	0.631	345	1.025	0.406	3.214	0.500
V6	105	0.718	0.136	0.896	0.443	345	0.525	0.096	0.739	0.282
V7	105	21,028.718	11,482.860	48,075.000	6642.330	345	9714.704	3353.454	21,037.600	3561.510
V8	105	0.662	0.459	1.721	0.080	345	0.170	0.151	1.021	0.008
V9	105	0.033	0.026	0.104	0.012	345	0.033	0.022	0.093	0.004
V10	105	0.028	0.015	0.065	0.009	345	0.012	0.006	0.030	0.002
V11	105	0.002	0.002	0.008	0.000	345	0.004	0.004	0.031	0.000
V12	105	3.959	0.855	6.089	2.501	345	3.873	0.716	5.737	2.359

shows the descriptive statistics of the variables for the period 2007–2021. A significant disparity exists between the experimental group and the control group, with CO₂ and SO₂ concentrations lower in the pilot region than in the nonpilot region, but the concentrations of NO_x and PM_2.5 are greater, suggesting that the composition of the sources of NO_x and PM_2.5 may be different from that of other pollutants. The level of economic development, industrial structure, urbanization level, and degree of opening up to the outside world of the pilot areas are significantly greater than those of the nonpilot areas, the structure of power generation is reasonable, and investment in scientific and technological research and development is substantially higher than in the nonpilot areas, but the investment in environmental governance in the nonpilot areas is somewhat greater than that of the pilot areas.

4. Empirical Results

4.1. Benchmark Results

To ensure the reasonableness and robustness of the results, we used a regression with progressively increasing fixed effects and control variables, and the main results are presented in

Table 3. Basic regression results.

Variant	(1)				(2)				(3)
	CO2	SO2	NOx	PM2.5	CO2	SO2	NOx	PM2.5	CO2	SO2	NOx	PM2.5
Treat × time	−0.908 **	−0.780 ***	−0.446 **	−0.800 ***	−0.547 ***	−0.353 ***	−0.104 **	−0.429 ***	−0.753 **	−0.311 ***	−0.184 ***	−0.507 ***
	(0.381)	(0.277)	(0.223)	(0.286)	(0.145)	(0.080)	(0.047)	(0.094)	(0.366)	(0.099)	(0.066)	(0.122)
Time	0.054	−1.247 ***	−0.399 ***	−0.761 ***	0.044	−2.518 ***	−0.747 ***	−1.932 ***	−0.863	−1.924 ***	−0.598 ***	−3.070 ***
	(0.180)	(0.126)	(0.103)	(0.129)	(0.168)	(0.090)	(0.053)	(0.106)	(0.564)	(0.298)	(0.198)	(0.360)
Treat	−1.083 ***	−0.726 ***	−0.275 *	−0.722 ***	−2.188 ***	−3.791 ***	−1.838 ***	−3.419 ***	1.136	0.096	−0.212	−5.997 ***
	(0.289)	(0.197)	(0.157)	(0.205)	(0.249)	(0.137)	(0.080)	(0.162)	(1.495)	(0.733)	(0.498)	(0.843)
Constant term	7.425 ***	12.395 ***	12.341 ***	10.150 ***	8.126 ***	12.964 ***	12.770 ***	10.549 ***	6.230	23.6275 ***	11.793 ***	−0.051
	(0.140)	(0.096)	(0.073)	(0.096)	(0.201)	(0.110)	(0.064)	(0.130)	(4.075)	(2.243)	(1.510)	(2.782)
R-squared	0.152	0.354	0.148	0.253	0.8822	0.945	0.957	0.912	0.903	0.964	0.966	0.948
Time fixed effect	NO	NO	NO	NO	YES	YES	YES	YES	YES	YES	YES	YES
Location fixed effects	NO	NO	NO	NO	YES	YES	YES	YES	YES	YES	YES	YES
Control	NO	NO	NO	NO	NO	NO	NO	NO	YES	YES	YES	YES

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Values in parentheses are standard errors.

. Model (1) conducts a simple baseline regression analysis and provides simple regression results for the core explanatory variables, Model (2) accounts for city fixed effects and time fixed effects in addition to those in Model (1), while Model (3) incorporates eight control variables to enhance the reliability of the findings.

The results in Table 3 indicate that all regression coefficients are negative, with most being statistically significant at the 1% level. This suggests that the policy implementation had a substantial emission reduction effect on CO₂ and the three air pollutants, regardless of whether control variables are included. The coefficients of the cross terms in Model (3) show a significant improvement in model explanation upon adding control variables, with policy implementation variables yielding coefficients less than zero at the 0.01 significance level. This indicates that carbon emissions trading significantly reduces CO₂ and air pollutant emissions; specifically, CO₂ emissions decreased by 75.3%, SO₂ by 31.1%, NO_x by 18.4%, and PM_2.5 by 50.7%. Notably, the synergistic emission reduction effect of CO₂ with SO₂ and PM_2.5 is the most pronounced.

The data indicate that the implementation of the carbon emissions trading market has a notable synergistic effect on pollution and carbon reduction within the power industry.

4.2. Robustness Test

4.2.1. Parallel Trend Test

The estimation of policy effectiveness using the double difference method relies on the prerequisite that the experimental and control groups exhibit identical growth trends prior to experiencing the policy shock. Therefore, conducting a parallel trend analysis of the explanatory variables becomes imperative. In this study, lagged virtual variables were introduced for each year preceding the implementation of the carbon emission trading market policy. These variables were subsequently interacted with by the experimental group’s virtual variables. The resulting interaction terms, combined with the core DID explanatory variables, were utilized in the regression analysis of carbon dioxide and atmospheric pollutant emissions.

Figure 1a–d illustrate the results of emission tests for CO₂, SO₂, NO_x, and PM_2.5. The coefficient ${\beta }_{\mathrm{t}}$ (t = −5 to 0) is not significant, confirming that the DID model satisfies the parallel trend assumption. Analyzing the annual dynamic changes post-implementation of the CETS reveals that the regression coefficients ${\beta }_{\mathrm{t}}$ (t = 1 to 7) for CO₂, SO₂, and PM_2.5 emissions are generally significantly negative, indicating an overall decreasing trend in these emissions, as shown in Figure 1a,b,d. This suggests that the CETS has a dynamic impact on reducing CO₂, SO₂, and PM_2.5 emissions. Although the regression coefficient ${\beta }_{\mathrm{t}}$ (t = 1 to 7) for SO₂ is negative, it is not statistically significant, indicating that while the CETS contributes to reducing SO₂ emissions, it is not a primary factor and lacks a dynamic effect.

4.2.2. Placebo Test

For different provinces and cities with many other different traits, some of their traits may have different impacts over time, thus affecting the identification assumptions. Therefore, we first establish controls for a set of observable key city characteristics, including industrial structure, urbanization level, economic development level, openness to the outside world, energy structure, R&D intensity, environmental regulations, and power generation structure. However, for the effects of uncontrollable characteristics, especially unobservable characteristics, we employ an indirect placebo test. The results are shown in Figure 2, where it can be seen that the distribution of the test regression coefficients for all four models is concentrated at approximately 0, which shows that the randomly sampled sample mix has no impact on the effect of policy implementation. Therefore, it can be concluded that the baseline regression results, which differentiate between the experimental and control groups based on participation, are robust.

5. Further Analysis

According to the above analysis, the CETS significantly reduces carbon dioxide and air pollutant emissions, achieving synergistic emission reduction in China’s electric power industry and enhancing environmental quality. This supports Hypothesis 1. Further, to test Hypotheses 2–4, we explore the mediating effects of energy consumption, power generation structure, and technological innovation on the synergistic effect of pollution and carbon reduction through the influence of the CETS. The following mediation effect model is constructed with reference to Reuben M. Baron and David A. Kenny [92]:

$$Y_{it}={\beta }_0+{\beta }_2{N}_{it}{T}_{it}+\lambda {X^{\prime }}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(2)

$$M_{it}={\beta }_0+{\beta }_3{N}_{it}{T}_{it}+\lambda {X^{\prime }}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(3)

$$Y_{it}={\beta }_0+{\beta }_4{N}_{it}{T}_{it}+{\alpha }_1{M}_{it}+\lambda {X^{\prime }}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(4)

Since some of the mediating variables are included in the control variable $X_{it}$ in Model (1), to ensure consistency of the test, ${X^{\prime }}_{it}$ in Model (2)–(4) removes the energy structure and industrial structure compared to $X_{it}$, and the final control variables include urbanization level, level of economic development, degree of openness to the outside world, R&D intensity, environmental regulation, and power generation structure. $M_{it}$ is the mediating variable.

First, Model (2) is regressed such that if ${\beta }_2$ is significant, it indicates that policy affects $Y_{it}$ emissions; second, Model (3) is regressed such that if ${\beta }_3$ is significant, it indicates that policy acts on the mediator variable; final, in regressing Model (4) to compare the coefficients of ${\beta }_4$ with those of the policy implementation without the inclusion of the mediating variable $M_{it}$, the size of ${\beta }_2$, if ${\beta }_4$ is smaller than ${\beta }_2$ or the significance level decreases, this suggests that there is a mediating effect.

The results of the mediating effect model test are shown in

Table 4. Mechanism analysis 1: Energy consumption.

Variant	(1)	(2)	(3)	(4)	(5)	(6)
	LNCO2	Energy Consumption	LNCO2	LNSO2	Energy Consumption	LNSO2
Treat × time	−1.4883 ***	−0.032 **	−0.3609 *	−1.6561 ***	−0.032 **	−0.2476 **
	(−6.4779)	(0.016)	(−1.7694)	(−9.1627)	(0.016)	(−2.0721)
Energy consumption			0.0017 ***			0.0016 ***
			(3.4659)			(5.5702)
Constant term	7.3360 ***	1.938 ***	16.4711 ***	11.6877 ***	1.938 ***	16.3691 ***
	(85.3361)	(0.030)	(6.3219)	(177.0871)	(0.030)	(10.6929)
R-squared	0.086	0.028	0.507	0.167	0.028	0.748
Time fixed effect	YES	YES	YES	YES	YES	YES
Location fixed effects	YES	YES	YES	YES	YES	YES
Control	YES	YES	YES	YES	YES	YES

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Values in parentheses are standard errors.

,

Table 5. Mechanism analysis 2: Generation structure.

Variant	(1)	(2)	(3)	(4)	(5)	(6)
	LNCO2	Generation Structure	LNCO2	LNSO2	Generation Structure	LNSO2
Treat×time	−1.4883 ***	−0.010 ***	−0.4190 **	−1.6561 ***	−0.010 ***	−0.2206 *
	(−6.4779)	(0.003)	(−2.0876)	(−9.1627)	(0.003)	(−1.7595)
Generation structure			−0.0009 *			0.0021 ***
			(−1.6596)			(6.4890)
Constant term	7.3360 ***	−0.232 ***	13.7486 ***	11.6877 ***	−0.232 ***	18.0859 ***
	(85.3361)	(0.033)	(5.3234)	(177.0871)	(0.033)	(11.7289)
R-squared	0.086	0.159	0.542	0.167	0.159	0.767
Time fixed effect	YES	YES	YES	YES	YES	YES
Location fixed effects	YES	YES	YES	YES	YES	YES
Control	YES	YES	YES	YES	YES	YES

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Values in parentheses are standard errors.

and

Table 6. Mechanism analysis 3: Technological innovation.

Variant	(1)	(2)	(3)	(4)	(5)	(6)
	LNCO2	Technological Innovation	LNCO2	LNSO2	Technological Innovation	LNSO2
Treat×time	−1.4883 ***	0.0040 ***	−0.3609 *	−1.6561 ***	0.0040 ***	−0.2535 **
	(−6.4779)	(4.4366)	(−1.7694)	(−9.1627)	(4.4366)	(−2.0071)
Technological innovation			0.0017 ***			5.0510
			(3.4659)			(0.7836)
Constant term	7.3360 ***	−0.0629 ***	16.4711 ***	11.6877 ***	−0.0629 ***	16.3264 ***
	(85.3361)	(−5.3817)	(6.3219)	(177.0871)	(−5.3817)	(9.9858)
R-squared	0.086	0.752	0.507	0.167	0.752	0.730
Time fixed effect	YES	YES	YES	YES	YES	YES
Location fixed effects	YES	YES	YES	YES	YES	YES
Control	YES	YES	YES	YES	YES	YES

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Values in parentheses are standard errors.

.

5.1. Energy Consumption

The dummy variables for the treat × time in columns (2) and (5) of Table 4 are all significantly negative, indicating that the CETS has a substantial negative effect on total energy consumption. This means that total energy consumption in the pilot areas decreases significantly following the policy implementation. In columns (3) and (6), both the core dummy variable treat × time and the mediator variable energy consumption are included simultaneously. The regression coefficients for treat*time remain significantly negative, while those for energy consumption are significantly positive, indicating a strong correlation between energy consumption and both carbon dioxide and sulfur dioxide emissions. This demonstrates that a higher total energy consumption correlates with increased emissions. Compared to columns (1) and (4), treat × time remains negative at the 1% significance level, but its absolute value slightly decreases. This suggests that the CETS achieves the synergistic effect of pollution and carbon reduction in the electric power industry by lowering total energy consumption, thus confirming the mediating role of energy consumption and verifying Hypothesis 2.

5.2. Generation Structure

The dummy variable treat*time in columns (2) and (5) of Table 5 is significantly negative, indicating that carbon emissions trading policies have a substantial negative effect on the generation structure. Specifically, after the policy implementation, the share of thermal power generation in total generation within the pilot region decreases significantly. In columns (3) and (6), both the core dummy variable treat*time and the mediating variable generation structure are included, with the regression coefficients for treat*time remaining significantly negative, while those for generation structure are also significant. This suggests a significant relationship between the generation structure and both carbon dioxide and sulfur dioxide emissions. Compared to columns (1) and (4), the regression coefficient for treat*time remains significantly negative, although its absolute value slightly decreases. This indicates that the CETS facilitates pollution and carbon reduction in China’s power sector by adjusting the generation structure and reducing the proportion of thermal power generation, thereby confirming the mediating role of generation structure and verifying Hypothesis 3.

5.3. Technological Innovation

The dummy variable treat*time in columns (2) and (5) of Table 6 is significantly negative, indicating that the CETS has a substantial positive impact on technological innovation, suggesting that the implementation of the CETS has fostered technological advancements in the pilot region. In columns (3) and (6), both the core dummy variable treat × time and the mediator variable technological innovation are included, where the regression coefficient for treat × time remains significantly negative, consistent with the findings in columns (1) and (4). However, the absolute value of its coefficient decreases slightly, indicating that the CETS achieves pollution and carbon reduction in China’s power sector by promoting technological innovation. This supports the mediating role of technological innovation, thereby verifying Hypothesis 4. Overall, the results in Table 4, Table 5 and Table 6 reveal that the mediating effect of SO₂ is notable, and similar findings are obtained for NO_x and PM_2.5. For brevity, Table 4, Table 5 and Table 6 show the regression results for CO₂ and SO₂.

6. Heterogeneity Test

The ideal situation for the DID model, as a natural experiment of sorts, is that the pilot and nonpilot cities are chosen randomly. However, the choice of relevant policies is mostly not random. Similarly, the determination of the list of pilot cities is not random: the list of pilot areas is closely related to the geographical location of the city, the existing economy, the level of social development and the degree of openness, etc., and these preexisting differences between the cities may have different impacts on the urban environment over time trends, which may result in biases in the estimates.

6.1. Differences in Power Facility

As essential infrastructure supporting urban survival, power facilities serve as the sole provider of electricity, crucial for accelerating economic development and ensuring public security. The installed power generation capacity serves as a pivotal metric to gauge the extent of public power infrastructure, reflecting the scale of the power system. Regional disparities in power facility size may influence their carbon and air pollutant emissions. To this end, the interaction term M1DID (M1DID = M1 × treat × time) of the dummy variable for the power facility and the core explanatory variable treat × time is further added to control the influence on the result estimation.

6.2. Differences in Power Generation

Coal power is still the dominant energy source in China’s current electricity supply, giving full play to the role of touting and guaranteeing supply; coal power is also the main source of carbon emissions and air pollutants emissions in the power sector, and this difference may have an impact on the benchmark results as power generation is higher in the five relevant provinces of China, Inner Mongolia, Guangdong Province, Jiangsu Province, Shandong Province, and Xinjiang, as compared to the other provinces. Therefore, the interaction term M2DID (M2DID = M2 × treat × time) between the dummy variable for electricity generation and the core explanatory variable treat × time is added to control for the effect on the estimation of the results.

6.3. Differences in Economic Development

Differences in the scale of provinces may lead to different manifestations of the impact of carbon emissions, and these differences may stem from a variety of factors such as the level of economic development, industrial structure, energy consumption patterns, geographic location, climatic conditions, policy orientation, and technological level of each province. As a policy instrument of environmental regulation, the carbon emissions trading mechanism may have different impacts depending on the size of provinces and cities. Therefore, the interaction term M3DID (M3DID = M3 × treat × time) between the dummy variable for the size of the province and city and the core explanatory variable treat*time is added to control for the effect on the estimation of the results.

The results of M1, M2, and M3 are presented in

Table 7. Heterogeneity tests: power facility, power generation, and economic development.

Variant	M1				M2				M3
	CO2	SO2	NOx	PM2.5	CO2	SO2	NOx	PM2.5	CO2	SO2	NOx	PM2.5
MDID	1.222 ***	−0.426 **	−0.078	−0.175	0.143	0.197 *	−0.010	0.035	−0.406	0.333 *	0.431 ***	−0.317
	(0.347)	(0.185)	(0.132)	(0.180)	(0.228)	(0.109)	(0.063)	(0.135)	(0.370)	(0.177)	(0.097)	(0.217)
Constant term	17.507 ***	19.711 ***	12.972 ***	16.161 ***	4.580	22.955 ***	11.656 ***	12.259 ***	4.048	23.391 ***	12.865 ***	11.099 ***
	(2.870)	(1.528)	(1.087)	(1.492)	(4.431)	(2.121)	(1.281)	(2.613)	(4.509)	(2.151)	(1.181)	(2.641)
R-squared	0.516	0.778	0.742	0.758	0.8902	0.9568	0.9673	0.9283	0.8871	0.9585	0.9585	0.9283
Time fixed effect	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Location fixed effect	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES
Control	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES	YES

Note: *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively. Values in parentheses are standard errors.

. In the main estimated coefficient M1DID, the CO₂ coefficient shows a significant positive effect, while the SO₂ coefficient exhibits a significant negative effect. This suggests that variations in power facility levels significantly influence CO₂ and SO₂ emissions, whereas no significant effects are observed on the other two pollutants. Specifically, higher levels of power facility correlate with increased CO₂ emissions but decreased SO₂ emissions. In the main estimated coefficient M2DID coefficient, only the SO₂ coefficient is significantly positive, indicating that variations in power generation across different regions significantly impact SO₂ emissions, while there is no significant impact on the remaining three pollutants, implying lower power generation leads to reduced SO₂ emissions. In the main estimated coefficient M3DID coefficient, the SO₂ and NO_x coefficients are significantly positive, indicating that differences in economic scale significantly influence SO₂ and NO_x emissions, with no significant impact on the remaining two pollutants, suggesting smaller economic scales lead to reduced SO₂ and NO_x emissions.

7. Conclusions

The synergistic promotion of pollution and carbon reduction is essential for China’s green transformation in the new development stage, making it crucial to investigate the synergistic effects of carbon trading and other market instruments. This study focuses on the emissions of air pollutants from China’s power industry across 30 provinces and cities from 2007 to 2021, utilizing the DID model to analyze the impact of the CETS on pollution and carbon reduction in this sector. Furthermore, we explored the pathways through which the CETS contributes to the synergy between pollution and carbon reduction in the power industry, leading to the following conclusions.

First, the implementation of the CETS has a significant synergistic effect on reducing pollution and carbon in China’s power sector. CO₂ emissions decreased by 75.3%, SO₂ emissions by 31.1%, NO_x emissions by 18.4%, and PM_2.5 emissions by 50.7%. After a series of robustness tests, this conclusion remains reliable. It can be seen that the CETS significantly reduces CO₂ and air pollutant emissions in the power industry, and the synergistic emission reduction effect of CO₂ and PM_2.5 is most obvious. This indicates that the implementation of the CETS has largely produced a synergistic effect of pollution and carbon reduction in China’s power industry.

Second, the implementation of the CETS will indirectly promote the synergistic effect of pollution and carbon reduction in China’s power sector. Based on previous studies, three pathways of CETS synergistic effects on pollution and carbon reduction in China’s power sector have been proposed and confirmed: reducing energy consumption, influencing power generation structure, and promoting technological innovation.

Third, variations in power facility level, power generation, and economic scale significantly influence CO₂ and SO₂ emissions, as well as SO₂ and NO_x emissions. Therefore, the outcomes of the CETS’s synergistic effect on pollution and carbon reduction in the power industry vary notably depending on these factors.

Based on these findings, we draw the following policy implications.

First, the industry coverage of carbon emission trading makets should be improved, more trading entities should be introduced to enhance trading activities, so that the synergistic effect of the CETS in reducing pollution and carbon emissions can be realized in more industries and in a wider area.

Second, green and low-carbon industries should be cultivated. The government should adjust its industrial structure transformation strategy to make greater use of green technologies. To this end, localities are actively building low-carbon agriculture, developing energy-saving industries, nurturing the service sector, and improving ecoefficiency by optimizing the industrial structure.

Third, with respect to achieving innovation-driven low-carbon development, technological innovation remains the fundamental driver of green economic development. Therefore, this study recommends that the government take measures to guide enterprises to adopt green technologies and change the crude development model. For example, technology subsidies and financial support can be provided to heavily polluting industries to reduce the negative externalities brought about by environmental regulations and to encourage enterprises to make technological improvements and innovations and take a long-term path to sustainable development.