Research on the Impact of Carbon Emission Trading Policies on Urban Green Economic Efficiency—Based on Dual Macro and Micro Perspectives

Abstract

In the context of global climate change, carbon emission trading (CET) has become a critical tool for driving urban green economic transformation. Since 2011, China has launched CET pilot programs, supporting the achievement of the “dual carbon” goals. Studying the relationship between CET and urban green economic efficiency is essential for advancing urban green economic transitions. However, the existing research is limited by its single-perspective approach, insufficient exploration of mechanisms, and weak heterogeneity analysis, which restricts a comprehensivethe comprehensiveness of our understanding of policy effects. To address these gaps, this study is the first to integrate macro-regional data with micro-enterprise behavior, evaluating the impact of CET on urban green economic efficiency from a dual macro–micro perspective, thereby filling the research void in macro–micro data integration. At the macro level, this study employs panel data from 281 Chinese cities spanning 2007 to 2020, using fixed-effects and difference-in-differences (DID) models to assess the impact of CET on urban green economic efficiency. At the micro level, a game-theoretic pricing decision model is constructed to reveal behavioral differences among enterprises in complete and incomplete information markets and their indirect effects on green economic efficiency. The findings indicate that CET significantly enhances urban green economic efficiency, with technological innovation, green finance, and industrial structural upgrading serving as mediating mechanisms. Heterogeneity analysis shows that the effects are more pronounced in eastern, non-resource-based, small-to-medium-sized, and non-old industrial cities. The game-theoretic model further demonstrates that enterprises in complete information markets more effectively indirectly enhance green economic efficiency through CET mechanisms. By combining macro and micro perspectives, this study provides a new theoretical framework and practical insights for understanding the policy effects of CET. However, limitations such as data confined to Chinese pilots and model simplifications remain. Future research should expand data dimensions, allowing researchers to more comprehensively evaluate policy outcomes.

1. Introduction

Against the backdrop of increasingly severe global climate change, carbon emission trading (CET), as a market-based environmental regulatory tool [1], has gradually become an important means for countries to address climate change and promote green economic transformation. Since the signing of the Kyoto Protocol, international attention to carbon emission trading has continued to rise [2], with developed economies such as the European Union, the United States, and Japan successively establishing carbon emission trading markets [3]. As the world’s largest carbon emitter, China has explored a carbon market development path with Chinese characteristics by integrating its national conditions into the design and implementation of carbon emission trading policies, providing significant reference for the global community. Since 2013, China has launched carbon emission trading pilots in multiple provinces and cities and officially launched the national carbon emission trading market in 2021 [4], making it one of the largest carbon markets in the world. The carbon emission trading policy allocates carbon emission rights as a scarce resource through market mechanisms [5], aiming to incentivize enterprises to reduce carbon emissions through price signals and promote green and low-carbon development [6]. China’s unique design offers new insights for the global implementation of carbon emission trading.

Simultaneously, against this policy backdrop, Chinese President Xi Jinping announced at the General Debate of the 75th United Nations General Assembly in September 2020 that China would enhance its nationally determined contributions, adopt more robust policies and measures, and strive to peak carbon dioxide emissions by 2030 and achieve carbon neutrality by 2060. The proposal of these “dual carbon” goals marks China’s leading position in global climate governance and provides new strategic opportunities for the development of carbon emission trading policies. As a market-based environmental regulatory tool, carbon emission trading can mobilize the enthusiasm and initiative of various entities to protect the ecological environment, playing a critical role in achieving carbon peaking and neutrality goals at the urban level [7].

Moreover, carbon emission trading is not only an important policy tool for addressing climate change but also a key means of promoting urban green economic transformation [8]. Green economic efficiency, as a core indicator measuring the coordinated development of economic growth and resource–environmental sustainability, directly reflects the sustainability of urban economic development [9]. Improving green economic efficiency not only helps achieve carbon reduction goals but also promotes high-quality economic development [10,11]. Cities, as the primary carriers of economic activities, account for over 70% of global carbon emissions [12]. Therefore, the introduction of carbon emission trading policies provides a new pathway for cities to reduce pollution, cut carbon emissions, and expand green growth [5]. In China, the growth of urban green economic efficiency is not only related to the high-quality development of the national economy but also directly tied to the achievement of the “dual carbon” goals [13]. While advancing the construction of the carbon market, China has emphasized the integration of carbon emission trading policies with regional economic development and industrial transformation, providing strong support for the improvement of urban green economic efficiency.

Although carbon emission trading policy has been widely used around the world, it still faces many challenges in the policy design and implementation process. The carbon price in the carbon emission trading market fluctuates greatly, making it difficult for companies to carry out long-term technological innovation and emission reduction planning [14]. In its early stages, the European Union Emission trading System (EU ETS) failed to effectively incentivize companies to reduce emissions due to overly low carbon prices [3]. Carbon emission trading may also lead to the problem of “carbon leakage”, that is, high-emitting companies transfer production activities to areas with looser carbon emission controls, thus weakening the emission reduction effect of the policy [15]. The quota allocation method of carbon emission rights trading is also controversial. Free quota allocation may cause high-emitting companies to obtain too many carbon emission rights, thereby weakening the emission reduction effect of the policy [3]. In addition, the effect of carbon emission trading may vary significantly due to factors such as enterprise size, industry characteristics, and regional differences [16]. Zhang Yinghao et al. found that carbon emission trading has a significantly higher effect on improving green economic efficiency in the eastern region than in the central and western regions [5]. The shortcomings and challenges encountered in carbon emission trading policies indicate the urgency and importance of further research on carbon emission trading policies.

The existing literature has extensively examined the macro-level impacts, micro-level impacts, and influencing mechanisms of carbon emission trading policies, providing a robust theoretical foundation for this study to investigate the policy effects of carbon emission rights from both macro and micro perspectives.

At the macro level, the literature primarily focuses on the effects of carbon emission trading on economic growth, energy structural transformation, and regional carbon emission reduction outcomes. Sun and Huang (2020), through an analysis of China’s urbanization process, demonstrated that carbon emission trading policies significantly enhance regional carbon emission efficiency, particularly in highly urbanized areas [17]. Bayer and Aklin (2020) provided empirical evidence that the European Union Emission Trading System (EU ETS) effectively reduces carbon emissions even under low carbon prices, thereby validating the policy’s efficacy at the macro level [18]. Jiang et al. (2016), from the perspectives of policy design and implementation effectiveness, highlighted the potential of carbon emission trading policies in promoting clean energy adoption and reducing the dependence on fossil fuels [19]. Zhang SL et al. (2021) further explored the impact of carbon emission trading on regional green development and carbon equity, concluding that the policy plays a pivotal role in fostering regional coordinated development [20]. Oh D. H. et al. calculated the productivity index at the urban level, providing important references for the study of green economic efficiency at the urban level [21].

At the micro level, the literature predominantly investigates the effects of carbon emission trading on corporate behavior, technological innovation, and financial performance. Wang Wei dong et al. (2020) empirically established that carbon emission trading policies significantly stimulate corporate green R&D investment and technological innovation [22]. Liu Ye et al. (2023) extended this analysis by examining the policy’s impact on the quality of corporate export products, revealing that carbon emission trading enhances corporate environmental performance [23]. Liu Ying et al. (2024) demonstrated that while energy quota trading policies increase short-term operating costs for firms, they drive long-term improvements in financial performance through enhanced energy efficiency and technological innovation [24]. Cheng, K et al. (2024) investigated the role of digitalization in the green transformation of enterprises in resource-based cities, finding that the integration of carbon emission trading policies with digital technologies significantly boosts corporate green innovation capabilities [25]. Fang, L. et al. (2024) emphasized that carbon emission trading policies indirectly foster corporate innovation in green technologies by promoting urban green transformation [26].

The existing literature on both macro and micro levels offers valuable insights for this study to explore the impact of carbon emission trading policies on urban green economic efficiency from a dual macro–micro perspective. However, most studies adopt either a macro or micro perspective, lacking an integrated approach that bridges the two.

The existing literature has also preliminarily explored the mechanisms through which carbon emission trading policies exert their influence, primarily focusing on market mechanisms, technological innovation, and institutional design. Jung and Song (2023) analyzed the incentives of policies on corporate and market behavior from the perspectives of carbon price volatility and quota allocation [27]. Zhang et al. (2025) demonstrated that policies indirectly promote corporate green technological innovation by increasing carbon emission costs [28]. Additionally, the critical role of institutional design in policy implementation has been widely acknowledged. Cai et al. (2024) underscored that effective quota allocation mechanisms and robust regulatory frameworks are essential for the successful implementation of carbon emission trading policies [29]. While the existing literature has made significant strides in exploring the mechanisms of carbon emission trading policies, most studies focus on single mechanisms, lacking a comprehensive analysis of the synergistic effects of multiple mechanisms. This gap highlights the need for further research to integrate market mechanisms, technological innovation, and institutional design to provide a holistic understanding of the policy’s impact.

Through the review and combing of relevant literature, it can be seen that the existing literature has extensively discussed the relationship between carbon emission trading and green economic efficiency, but there are still the following shortcomings.

First, the research perspective is often singular, with existing studies predominantly adopting either a macro or micro viewpoint, lacking a comprehensive dual macro–micro analytical framework. Macro-level research primarily focuses on the impact of carbon emission trading on green economic efficiency at the regional or national level, while micro-level studies emphasize the critical factors influencing policy implementation effects at the corporate level. However, the effects of carbon emission trading are manifested both in the optimization of resource allocation at the macro level and in the behavioral adjustments of enterprises at the micro level. A single perspective is insufficient to fully unveil its impact mechanisms.

Second, the exploration of mechanisms remains superficial. Existing studies predominantly concentrate on the direct effects of carbon emission trading on green economic efficiency, with limited in-depth investigation into its intrinsic mechanisms. Although some studies have touched upon factors such as technological innovation, the pathways and specific impacts of these mechanisms have not been systematically analyzed. The lack of thorough mechanism research may constrain the understanding of the underlying logic of policy effects and undermine the scientific rigor and practical applicability of policy recommendations.

Third, the analysis of heterogeneity is inadequate. The existing literature offers a relatively weak examination of the heterogeneity in the effects of carbon emission trading policies. While some studies have noted potential differences in policy outcomes across regions or enterprises, discussions on key heterogeneity factors such as resource endowments, urban scale, and industrial foundations remain insufficient. Questions such as whether resource-rich cities and resource-scarce cities exhibit significant differences in policy responsiveness, or how large cities and small-to-medium cities perform differently in policy implementation, have not been fully addressed. This lack of in-depth heterogeneity analysis also hampers the refinement and targeting of policy design.

This study aims to explore the impact of carbon emission trading on urban green economic efficiency and its internal mechanism from a macro–micro dual perspective. The innovation of this study is mainly reflected in the following aspects: First, addressing the issue of the single perspective in existing research, this paper is the first to integrate macro-level regional data with micro-level corporate behavior, providing a comprehensive assessment of the impact of carbon emission trading on urban green economic efficiency from a dual macro–micro perspective. This approach fills the gap in the integration of macro and micro data in existing studies, fully captures the interactions between macro and micro levels, and offers a more holistic perspective for understanding policy effects. Second, to bridge the macro–micro perspectives, this paper is the first to combine the difference-in-differences (DID) method with game theory models. At the macro level, the DID method is employed to comprehensively evaluate the policy’s effects on urban outcomes. At the micro level, a game-theoretic pricing decision model for firms is constructed. By establishing game models under both complete and incomplete information, this paper analyzes firms’ decision-making behaviors in two different types of markets, revealing the indirect impact of these behaviors on urban green economic efficiency. This innovative approach provides a new methodology for integrating macro and micro data while offering a novel theoretical framework for understanding the micro-level mechanisms of policy effects. Third, addressing the insufficient exploration of mechanisms in existing research, this paper selects three key mechanisms—technological innovation, green finance, and industrial structure—to delve into the intrinsic pathways through which carbon emission trading policies influence urban green economic efficiency. Based on mediation effect tests, the paper systematically analyzes how carbon emission trading affects green economic efficiency through these three critical mediating variables. Fourth, to further address the lack of in-depth heterogeneity analysis in existing studies, this paper examines the regional heterogeneity, resource endowment heterogeneity, urban scale, and industrial foundation. By combining these factors with the policy context, it reveals the differential effects of carbon emission trading policies across different types of cities, providing targeted recommendations for further policy refinement.

2. Materials and Methods

2.1. Theoretical Analysis and Research Hypotheses

2.1.1. Carbon Emission Trading Policy and Urban Green Economic Efficiency

Carbon emission trading policies internalize environmental externalities through market mechanisms and encourage companies to optimize resource allocation, thereby improving green economic efficiency. Research shows that carbon trading policies can effectively reduce industry carbon emission intensity [30] and guide companies to adopt clean technologies through price signals [18]. The carbon emission performance of China’s pilot cities improved significantly after the implementation of the policy [31], indicating that the market mechanism’s role in promoting green efficiency is universal. In addition, the carbon trading policy strengthens the motivation of enterprises to reduce emissions through the quota allocation and trading mechanism and promotes green total factor productivity growth [31]. Empirical evidence from the European Union Emission Trading System (EU ETS) shows that corporate energy efficiency increases after the implementation of the policy [32], verifying the effectiveness of market-based environmental regulation. Judging from China’s practice, pilot policies have significantly reduced regional carbon emission intensity [19] and promoted the spatial spillover of low-carbon technological innovation [33], further supporting the improvement of green economic efficiency [34].

In view of that, this article proposes the following:

Hypothesis 1. 

Carbon emission trading policy can improve the city’s green economic efficiency.

2.1.2. Carbon Emission Trading Policy, Technological Innovation, Industrial Structural Upgrading, and Green Finance

(1)

Technological innovation path

Carbon trading policies use price signals to force companies to increase investment in green R&D. Research shows that market-incentive environmental regulation significantly improves the level of low-carbon technological innovation of enterprises [35], and pilot policies promote green total factor productivity growth in high-carbon industries through the Porter effect [36]. Internationally, the implementation of EUETS has accelerated the diffusion of clean technologies [37], while China’s pilot policy encourages corporate patent innovation through carbon price fluctuations [11], and technological innovation is conducive to improving the city’s green economic efficiency [23].

(2)

Industrial structural upgrade path

Carbon trading promotes the transformation of energy-intensive industries through cost constraints. The market mechanism prompts enterprises to adjust the factor input structure [7,38], driving the industry to upgrade to knowledge-intensive and technology-intensive [9]. Pilot policies have significantly promoted the increase in the proportion of the service industry [39] and optimized the regional industrial structure through factor reallocation, which in turn affects the city’s green economic efficiency [40].

(3)

Green finance path

Green finance aims to provide financing support for low-carbon projects and promote the formation of green capital. In the context of carbon emission trading policies, carbon financial instruments such as carbon quota pledges and carbon bonds have broadened corporate financing channels [41], and the linkage between carbon price fluctuations and green credit [42] has strengthened the screening function of finance for green technologies. Research has confirmed that green finance improves urban green economic efficiency through the capital allocation effect [24], and there is a synergistic effect between carbon trading pilots and green investment [43].

Based on the above analysis, this article proposes the following:

Hypothesis 2. 

Carbon emission trading policy improves the city’s green economic efficiency through technological innovation, industrial structural upgrading, and green financial improvement.

2.1.3. Micro-Enterprise Decision-Making and Urban Green Economic Efficiency

Information transparency is a key prerequisite for the effective operation of the carbon market. In a complete information environment, companies can accurately assess emission reduction costs and transaction benefits [44,45] to optimize emission reduction decisions. In China’s pilot policies, the disclosure of quota allocation information reduces transaction costs [46], while the transparency of the carbon price formation mechanism reduces the risk of market manipulation [47]. In contrast, information asymmetry can easily lead to market failures, such as carbon price distortions [29] and quota arbitrage [26]. Existing studies have shown that companies in markets with complete information are more inclined to long-term green investments [36], while “free-riding” behavior exists in areas with incomplete information [48]. In addition, with complete information, the green efficiency gains of carbon trading policies can be further strengthened [37,49].

Based on the above analysis, this article proposes the following:

Hypothesis 3. 

Compared with the incomplete information market, enterprises in the complete information market can more effectively improve urban green economic efficiency indirectly through the carbon emission trading mechanism.

Based on the above analysis, the framework of this study was constructed as follows (Figure 1):

2.2. Data Source and Variable Description

2.2.1. Data Source

This article combines macro and micro perspectives to study the impact of carbon emission trading policies on urban green economic efficiency. For the macro level, this article selects panel data at the city level in China. The selection of the data interception time point is mainly based on two considerations: First, the national carbon emission trading market was officially launched in 2021. In order to avoid the institutional confounding effects of the comprehensive promotion period of the policy, the sample cut-off time point is set to 2020 before the policy diffusion; secondly, taking into account the time-lag characteristics of policy effects and to more comprehensively capture the potential impacts of the policy, this paper incorporates an extended sample spanning from 2007 to 2023 in the robustness tests. By comparing these results with the baseline regression, the study ensures the scientific rigor and reliability of the findings.

The macro data mainly come from the China City Statistical Yearbook, China Energy Yearbook, China Environment Yearbook, China Finance Yearbook, and the work reports of various city governments. Before conducting econometric regression, in order to avoid the impact of extreme values on the results, this study performs the following standardization processing on the original data: (1) eliminate urban samples with missing data rates > 15%; (2) perform bilateral 1% tail shrinkage processing on continuous variables; (3) use linear interpolation to fill non-continuous missing values. Finally, a balanced panel dataset covering 281 prefecture-level cities is obtained, with a total of 3934 valid observations.

At the micro level, this paper employs a game-theoretic evolutionary model, integrating the principle of utility maximization from microeconomics, to construct a decision-making and pricing model for firms under different scenarios of carbon emission trading markets. Through mathematical derivations, this paper simulates firms’ behaviors in carbon trading and quantitatively analyzes the revenue from buying and selling carbon allowances using function and parameter specifications. Finally, based on the evolutionary process and outcomes, a comparative static analysis is conducted for sensitivity analysis and discussion.

2.2.2. Variable Description

(1)

Explanatory variables: The core explanatory variable of this article is the carbon emission trading policy. Referring to the research of Dong Zhiqing and Wang Hui, the interaction term of policy and practice dummy variables is constructed posttreat, and its coefficient is the net effect of the carbon emission trading policy on urban green economic efficiency estimated by double differences [50]. Based on whether they are carbon emission trading pilot cities, this article takes the eight policy regions that have launched carbon emission trading pilots in China since 2011 as the experimental group, and cities that have not yet introduced carbon emission trading pilots as the control group, while the experimental group Policy is assigned a value of 1, and the control group Policy is assigned a value of 0. Construct a time dummy variable, with the year after the policy is implemented (Year) taking the value of 1 and the other years taking the value of 0; construct the interaction term between the policy dummy variable and the time dummy variable Policy_kt × Year_kt, i.e., posttreat [51].

(2)

Explained variable: The explained variable in this study is the green economic efficiency of the city. Referring to the research of Zhou Yunbo et al. [52], this study selects the green total factor productivity at the city level as an indicator to measure the green economic efficiency of the city.

This article uses the SBM-GML index method to measure green total factor productivity and draws on the practices of Li et al. [53] and Oh [21] to construct super-efficiency, which includes undesired outputs of SBM models.

First, the production possibility set is as follows:

$$\mathrm{P}\mathrm{P}\mathrm{S}=\left\{\begin{array}{c}\left(\mathrm{X},\overline{{\mathrm{Y}}^{\mathrm{g}}},\overline{{\mathrm{Y}}^{\mathrm{b}}}\right)|\mathrm{X}\ge {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{j}},\overline{{\mathrm{Y}}^{\mathrm{g}}}\le {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{j}},\\ \overline{{\mathrm{Y}}^{\mathrm{b}}}\ge {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{j}},\mathrm{L}\le \mathrm{e}\mathsf{\mu }\le \mathsf{\lambda },{\mathsf{\mu }}_{\mathrm{j}}\ge 0\end{array}\right\}$$

(1)

In Model (1), the (k)-th city includes (m) types of inputs, (s₁) types of expected outputs, and (s₂) types of undesirable outputs. X represents the (m)-dimensional input vector, (Yg) represents the (s₁)-dimensional expected output vector, (Y_b) represents the (s₂)-dimensional undesirable output vector, and (µ) represents the (L)-dimensional weight vector. Based on this, the following ultra-efficiency SBM model is constructed:

$$\mathsf{\eta }=\underset{\mathsf{\mu },\mathrm{x},{\mathrm{y}}^{\mathrm{g}},{\mathrm{y}}^{\mathrm{b}}}{\min }{\displaystyle \frac{\left({\mathrm{s}}_1+{\mathrm{s}}_2\right){\sum }_{\mathrm{i}=1}^{\mathrm{n}}\frac{\overline{{\mathrm{x}}_{\mathrm{t}}}}{{\mathrm{x}}_{\mathrm{i}\mathrm{o}}}}{\sum _{\mathrm{r}=1}^{{\mathrm{s}}_1}\frac{\overline{{\mathrm{y}}_{\mathrm{r}}^{\mathrm{g}}}}{{\mathrm{y}}_{{\mathrm{r}}_{\mathrm{o}}}^{\mathrm{g}}}+\sum _{\mathrm{k}=1}^{{\mathrm{s}}_2}\frac{\overline{{\mathrm{y}}_{\mathrm{k}}^{\mathrm{b}}}}{{\mathrm{y}}_{{\mathrm{k}}_{\mathrm{o}}}^{\mathrm{b}}}}}$$

(2)

S.T.

$$\mathrm{X}\ge {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{j}};\overline{{\mathrm{Y}}^{\mathrm{g}}}\le {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{j}}^{\mathrm{g}};\overline{{\mathrm{Y}}^{\mathrm{b}}}\ge {\sum }_{\mathrm{j}=1,\mathrm{j}\ne 0}^{\mathrm{L}}{\mathsf{\mu }}_{\mathrm{j}}{\mathrm{y}}_{\mathrm{j}}^{\mathrm{b}}$$

(3)

$$\overline{\mathrm{X}}\ge {\mathrm{x}}_{\mathrm{o}},\overline{{\mathrm{Y}}^{\mathrm{g}}}\le {\mathrm{y}}_{\mathrm{o}}^{\mathrm{g}};\overline{{\mathrm{Y}}^{\mathrm{g}}}\ge 0;\overline{{\mathrm{Y}}^{\mathrm{b}}}\ge 0,\mathrm{L}\le \mathrm{e}\mathsf{\mu }\le \mathsf{\lambda },{\mathsf{\mu }}_{\mathrm{j}}\ge 0$$

(4)

$$\overline{{\mathrm{x}}_{\mathrm{t}}}={\mathrm{x}}_{\mathrm{i}\mathrm{o}}+\mathrm{s}\left(\mathrm{i}=1,\cdots ,\mathrm{n}\right);\overline{{\mathrm{y}}_{\mathrm{r}}^{\mathrm{g}}}={\mathrm{y}}_{\mathrm{r}\mathrm{o}}^{\mathrm{g}}-{\mathrm{s}}^{\mathrm{g}}\left(\mathrm{r}=1,\cdots ,{\mathrm{s}}_1\right);\overline{{\mathrm{y}}_{\mathrm{k}}^{\mathrm{b}}}={\mathrm{y}}_{\mathrm{k}\mathrm{o}}^{\mathrm{b}}+{\mathrm{s}}^{\mathrm{b}}\left(\mathrm{k}=1,\cdots ,{\mathrm{s}}_2\right)$$

(5)

In Models (6) to (9), Xt, ${\mathrm{y}}_{\mathrm{r}}^{\mathrm{g}},$ and ${\mathrm{y}}_{\mathrm{k}}^{\mathrm{b}}$ represent the target values of inputs and outputs for the evaluated units, while ${\mathrm{X}}_{\mathrm{i}\mathrm{o}}$, ${\mathrm{y}}_{\mathrm{r}\mathrm{o}}^{\mathrm{g}},$ and ${\mathrm{y}}_{\mathrm{k}\mathrm{o}}^{\mathrm{b}}$ represent the original values.

The GML index can be decomposed into a technological progress index (GTC) and a technical efficiency index (GEF). Therefore, the GML index is constructed as follows:

$$\mathrm{G}\mathrm{M}\mathrm{L}=\mathrm{G}\mathrm{E}\mathrm{F}\times \mathrm{G}\mathrm{T}\mathrm{C}={\displaystyle \frac{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}}={\displaystyle \frac{{\mathrm{E}}^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}{{\mathrm{E}}^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}}\times {\displaystyle \frac{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}{{\mathrm{E}}^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}}\times {\displaystyle \frac{{\mathrm{E}}^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}}$$

(6)

$$\mathrm{G}\mathrm{E}\mathrm{F}={\displaystyle \frac{{\mathrm{E}}^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}{{\mathrm{E}}^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}}$$

(7)

$$\mathrm{G}\mathrm{T}\mathrm{C}={\displaystyle \frac{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}{{\mathrm{E}}^{\mathrm{t}+1}\left({\mathrm{x}}^{\mathrm{t}+1},{\mathrm{y}}^{\mathrm{t}+1}\right)}}\times {\displaystyle \frac{{\mathrm{E}}^{\mathrm{t}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}{{\mathrm{E}}^{\mathrm{g}}\left({\mathrm{x}}^{\mathrm{t}},{\mathrm{y}}^{\mathrm{t}}\right)}}$$

(8)

where ${\mathrm{x}}^{\mathrm{t}}, {\mathrm{y}}^{\mathrm{t}}$ represents the input–output value of period t, ${\mathrm{E}}^{\mathrm{g}}$ and ${\mathrm{E}}^{\mathrm{t}}$ represent the efficiency values of the global frontier and the frontier period t, ${\mathrm{x}}^{\mathrm{t}}$ and ${\mathrm{y}}^{\mathrm{t}}$ represent the input and output values for period t, and ${\mathrm{E}}^{\mathrm{g}}$ and ${\mathrm{E}}^{\mathrm{t}}$ represent the global frontier and the efficiency values for period t, respectively.

The input indicators in this study are categorized into three types: labor input, capital input, and energy input. Labor input is measured by the number of employed persons in urban districts, expressed in units of 10,000 people. Capital input is quantified by the built-up area of urban districts (in square kilometers) and the fixed capital stock for period t (in 10,000 CNY). Referring to the practice of Zhang Jun and Wang Jun [36,54], the perpetual disk is used. The deposit method uses 2006 as the base period to obtain the total energy consumption in the municipal area measured by energy input (10,000 tons of coal standard); the expected output indicator is measured by the actual GDP of prefecture-level cities (10,000 CNY) calculated at constant prices in 2006; and the undesired output indicator refers to the research of Zheng Jie et al. (2023) [40], using industrial SO₂ (tons), industrial wastewater emissions (10,000 tons), and industrial smoke and dust emissions (tons). We use the entropy method to fit the above indicators into comprehensive environmental pollution indicators. For missing data, we refer to the method of Tu Zhengge (2008) [38] and use the interpolation method to fill in the missing data.

Table 1. Green economic efficiency evaluation system.

Primary Indicator	Secondary Indicator	Tertiary Indicator	Unit
Input Indicators	Labor Input	Employed Population in Urban Districts	10,000 people
	Capital Input	Built-up Area of Urban Districts	Square kilometers
		Calculated Urban Capital Stock using the Perpetual Inventory Method (Deflated with 2006 as the Base Year)	10,000 CNY
	Energy Input	Total Energy Consumption in Urban Districts	10,000 tons of coal equivalent
Expected Output Indicators	Regional GDP	City GDP Calculated at Constant 2006 Prices	10,000 CNY
Unwanted Output Indicators	Three Urban Wastes	Industrial SO2 Emissions	Tons
		Industrial Wastewater Discharge	10,000 tons
		Industrial Smoke and Dust Emissions	Tons

shows the calculation indicators of green economy efficiency.

(3)

Control variables:

To avoid omitted variable bias, this study selects the following control variables:

① Innovation level (Aggregation): Regions with a high proportion of scientific researchers are more likely to improve green economic efficiency through technological innovation [55], and carbon emission trading policies can promote technological progress by forcing companies to increase R&D investment [56], so this variable needs to be controlled. ② Investment level (Invest): The scale of investment affects the path of capacity expansion and technology upgrading [57]. High investment may increase energy dependence or promote green substitution [58]. It may be an influencing variable of the explained variable and needs to be controlled. ③ Degree of openness to the outside world (Ope): The technological spillover effect of an open economy is significant [59]. A city that over-exports may cause the city to become a high-carbon city, which will have an impact on the explained variable, so this variable needs to be controlled [9]. ④ Economic structural level (Ind): The regional economic structural level can be measured by the secondary industry’s share of regional GDP [60]. However, in regions with higher economic structural levels, the emission reduction pressure and technological potential coexist [61], and their inhibitory effect on green efficiency needs to be controlled [62]. ⑤ Government intervention (Gov): Fiscal expenditure reflects the intensity of government intervention. The government supports low-carbon cities and enterprises through the implementation of environmental policies and green projects [63], which has a significant impact on the green economic efficiency of cities [64]. Its effect needs to be controlled to accurately evaluate the role of carbon emission trading policies. ⑥ Number of industrial enterprises above a certain scale (Structure): Cities with a large number of industrial enterprises tend to have high carbon emissions, and their energy consumption will be less than that of low-carbon cities. Therefore, this variable needs to be controlled to reduce the omitted variable bias problem (Fan et al., 2015) [65].

(4)

Mediating variables

Based on the theoretical analysis in the previous article, this study selects the technological innovation, industrial structure, and green finance as mediating variables. ① Technological innovation: According to the “Porter hypothesis” [66], carbon trading forces enterprises to increase their investment in low-carbon technology research and development through carbon price signals and compliance pressure, thus forming a long-term emission reduction cost advantage [67]. Furthermore, technological innovation is conducive to improving the total factor productivity of cities, which shows that technological innovation is irreplaceable in policy transmission [11,68]. ② Industrial structural upgrading: Industrial structural upgrading is the key path for carbon trading policies to optimize resource allocation [69]. Carbon trading policies can force capital to shift to low-carbon industries by increasing the marginal costs of high-energy-consuming industries [70], and industrial structural upgrading has a certain leverage effect on green efficiency [71,72]. ③ Green finance: The carbon emission trading policy has spawned new financial instruments, such as carbon finance and other green finance. Green finance can provide credit and other financial support to low-carbon enterprises and then affect the macro green economic efficiency from the micro level [28,42,43].

To more accurately assess the mechanistic effects of the three mediating variables, this study employs a multi-dimensional approach to measure the mechanistic variables. For technological innovation, the level of urban technological innovation is gauged using two indicators: science and technology expenditure and the number of patent applications. For industrial structural upgrading, this study utilizes two dimensions of indicators: the advancement of the industrial structure and the overall upgrading of industries.

For the calculation of green finance, there is no clear measurement method or unified measurement system in China. Most of the relevant research is based on measurement methods used by previous scholars. This study taking into account the need for comprehensiveness when measuring the development of green finance, so it examines the mechanism of green finance from three dimensions: green finance index, green funds, and green credit. Drawing on the research of Yan Tianshun et al. [26], the green finance index is constructed primarily from four dimensions—green credit, green securities, green insurance, and green investment—to establish China’s green finance indicator system. The entropy method, an objective weighting approach, is employed for measurement.

First, we perform data annotation processing:

For positive indicators,

$${\mathrm{y}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-\min \left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)}{\max \left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)-\min \left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)}}$$

(9)

For negative indicators,

$${\mathrm{y}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)-{\mathrm{x}}_{\mathrm{i}\mathrm{j}}}{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}\right)}}$$

(10)

where ${\mathrm{y}}_{\mathrm{i}\mathrm{j}}$ represents the jth indicator of the ith province. Translate the standardized values to avoid subsequent calculation errors.

Second, we determine the weight of the indicator. The formula is as follows:

$${\mathrm{p}}_{\mathrm{i}\mathrm{j}}={\displaystyle \frac{{\mathrm{y}}_{\mathrm{i}\mathrm{j}}}{\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}}}$$

(11)

where ${\mathrm{p}}_{\mathrm{i}\mathrm{j}}$ represents the weight of the ith province under the jth indicator.

The third step is to calculate the information entropy value of the jth indicator. The formula is as follows:

$${\mathrm{e}}_{\mathrm{j}}=-\mathrm{k}\left({\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}\mathrm{l}\mathrm{n}{\mathrm{p}}_{\mathrm{i}\mathrm{j}}\right),\mathrm{k}>\mathrm{0,0}<{\mathrm{e}}_{\mathrm{i}\mathrm{j}}<1$$

(12)

where k is generally a constant:

$$\mathrm{k}={\displaystyle \frac{1}{\ln \mathrm{n}}}$$

(13)

Next, we calculate the difference coefficient of the jth indicator and the weight of the indicator. The formula is as follows:

$${\mathrm{g}}_{\mathrm{i}}=1-{\mathrm{e}}_{\mathrm{j}}$$

(14)

$${\mathrm{w}}_{\mathrm{i}}={\displaystyle \frac{{\mathrm{g}}_{\mathrm{i}}}{\sum _{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{g}}_{\mathrm{i}}}}$$

(15)

Finally, we use the linear weighting method to multiply the translated standardized data obtained by each indicator by the required weight to obtain the value of the comprehensive indicator of the green finance development level. ${\mathrm{v}}_{\mathrm{i}}$ is the green finance development level index of region i. The formula is as follows:

$${\mathrm{v}}_{\mathrm{i}}={\sum }_{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{w}}_{\mathrm{j}}\times {\mathrm{x}}_{\mathrm{i}\mathrm{j}}$$

(16)

Table 2. Green finance indicator evaluation system.

Green Finance Indicator System
Secondary indicator	Tertiary indicator	Indicator description
Green credit	Proportion of new bank loans to listed environmental protection companies	The new bank loans of A-share listed environmental protection companies accounted for the loans of A-share listed companies to banks
	Proportion of interest expenses in high-energy-consuming industries	Interest expenses of six high-energy-consuming industries/total industrial interest
Green securities	Market value share of A-share listed environmental protection companies	Market value of listed environmental protection companies/total market value of A-share listed companies
	Proportion of A-share market capitalization of high-energy-consuming companies listed on A-shares	Market value of A-share listed high-energy-consuming companies/total market value of A-share listed companies
Green insurance	Scale of environmental pollution insurance	Agricultural insurance income/property insurance income
	Environmental pollution insurance compensation ratio	Agricultural insurance expenditure/agricultural insurance income
Green investment	Proportion of investment in environmental pollution control	Environmental pollution control investment/GDP
	Proportion of fiscal environmental protection expenditure	Fiscal environmental protection expenditure/total fiscal expenditure

shows the indicator system selected for the calculation of the Green Finance Index:

In summary, the scalar selection in this article is shown in

Table 3. Variable description.

Type	Definition	Symbol	Measurement
Explained variable	Green economic efficiency	Efficiency	Super-efficiency SBM model
Explanatory variables	Carbon emission rights trading	Posttreat	Intersection of the policy implementation dummy variable and time dummy variable
Control variables	Innovation level	Aggregation	Percentage of scientific researchers among the employees at the end of the year
	Investment level	Invest	Logarithm of fixed-asset investment at the end of the year
	Degree of opening up	Ope	Proportion of total imports and exports in GDP
	Economic structural level	Ind	Proportion of added value of the secondary industry in GDP
	Government intervention level	Gov	Proportion of fiscal expenditure in GDP
	Industrial economic scale	Number	Logarithm of the number of industrial enterprises above the designated size
Intermediary variables	Technological innovation	Innovation	Logarithm of the number of patent applications
		Science	Proportion of technological expenditure in GDP
	Upgrade of industrial structure	Advanced Structure	Proportion of added value of the tertiary industry in GDP
		Overall Upgrade	Proportion of added value of the primary industry to GDP multiplied by 1+ the proportion of the secondary industry to GDP multiplied by 2+ the proportion of the tertiary industry to GDP multiplied by 3
	Green finance	Green Fund	Total market value of green funds is greater than the total market value of all funds
		Green Equity	Carbon trading, energy use rights trading, and emission trading account for the total amount of equity market transactions
		Green Index	Entropy method calculation

:

2.3. Model Selection and Model Construction

2.3.1. Model Selection

Before constructing the econometric model, this study takes the first batch of carbon emission trading pilot policies in 2011 as the research object and uses the independent-samples T-test method to conduct a preliminary analysis of the policy effects. The empirical results in

Table 4. Independent-samples T test.

Variable	Before Policy Implementation			After the Policy is Implemented
	Intervention Group	Control Group	T Value	Intervention Group	Control Group	T Value
Efficiency	0.280	0.322	−5.322	0.333	0.340	−1.569
Observations	1490			2682

show that before the implementation of the policy, for the intervention group and its counterpart, there is a significant difference in the green economic efficiency of the control group (T value = −5.322, p < 0.01); after the implementation of the policy, the efficiency of the intervention group significantly increases to 0.333, and the significance of the difference with the control group is significantly reduced (T value = −1.569). This shows that the carbon emission trading policy effectively narrows the efficiency gap between pilot cities and non-pilot cities, which verifies the effectiveness of policy intervention, and we can build a model to study the specific causal effects and paths between the two.

In order to determine the applicable form of the panel data model, this study uses the Hausman test to distinguish between fixed-effects and random-effects models. As shown in

Table 5. Hausman test.

Variables	FE
	(1)
Posttreat	0.0177 ***
	(0.00395)
Controls	Yes
Constant	0.0896 ***
	(0.0323)
Observations	3934
Number of city	281
Hausman	169.6
p-value	0.000

Note: Values in parentheses are standard errors, *** p < 0.01.

, the test statistic value is 169.6 (p < 0.001), indicating that the fixed-effects model has better estimation consistency. Based on that, this article uses the fixed-effects model for quantitative empirical testing.

2.3.2. Model Establishment

(1)

Baseline regression model

When evaluating the policy effect, this study refers to the research of Liu et al. [24], adopts a difference-in-differences (DID) model framework, and selects a two-way fixed-effects model of the city and time. It not only compares the changes before and after the policy implementation between the treatment group and the control group but also further introduces multi-period considerations in the time dimension to more accurately capture the dynamic changes in the policy impact. Specifically, this article refers to the research methods of He Lingyun and Wei Dongming [37], and the model design is as follows:

$$\begin{gathered} {\mathrm{Efficiency}}_{\mathrm{k}\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{t}\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}}_{\mathrm{k}\mathrm{t}}+{{\mathsf{\alpha }}_2\mathrm{I}\mathrm{n}\mathrm{v}\mathrm{e}\mathrm{s}\mathrm{t}}_{\mathrm{k}\mathrm{t}}+{\mathsf{\alpha }}_3{\mathrm{I}\mathrm{n}\mathrm{d}}_{\mathrm{k}\mathrm{t}}+{{\mathsf{\alpha }}_4\mathrm{O}\mathrm{p}\mathrm{e}}_{\mathrm{k}\mathrm{t}} \\ +{{\mathsf{\alpha }}_5\mathrm{A}\mathrm{g}\mathrm{g}\mathrm{r}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}}_{\mathrm{k}\mathrm{t}}+{{\mathsf{\alpha }}_6\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}}_{\mathrm{k}\mathrm{t}}+{\mathsf{\zeta }}_{\mathrm{k}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{k}\mathrm{t}} \end{gathered}$$

(17)

In model (17), Efficiency represent the green efficiency level of the city, posttreat is the interaction term of the policy dummy variable and the time dummy variable, namely Policykt × Yearkt, Invest, Ind, Ope, Aggregation, and Number are the control variables of the model, k represents the region, t represents the year, ζ_k represents the region fixed effects, δ_t represents the time fixed effects, and ε_kt is the random error term.

(2)

Evolutionary game model

In order to explore the indirect impact of micro-enterprise decision-making on urban green economic efficiency, this study refers to the research of Chen Hongzhuan et al. and Hu Yufeng et al. [73,74], and it divides the carbon market into a quasi-market and market mechanism regulation market, similar to the complete information market and incomplete information market. The evolutionary game model is adopted, combined with the enterprise profit maximization model, the consumer utility maximization model, and the enterprise carbon rights trading game pricing model. By analyzing the micro-enterprise game-pricing decision-making behavior, the different decisions of enterprises in different markets are analyzed to determine the impact on the green economic efficiency of the city.

(3)

Mechanism test model

In order to explore the transmission path of the carbon emission trading market affecting urban green economic efficiency, this article refers to the intermediary effect assessed with testing method of Wen Zhonglin and Ye Baojuan (2014) [39], using the stepwise regression method to build a mechanism testing model. The models are shown in models (18), (19), and (20):

$${\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}}_{\mathrm{k}\mathrm{t}}={\mathsf{\eta }}_0+{\mathsf{\eta }}_1{\mathrm{d}\mathrm{i}\mathrm{d}}_{\mathrm{k}\mathrm{t}}+{{\mathsf{\eta }}_2\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{k}\mathrm{t}}+{\mathsf{\zeta }}_{\mathrm{k}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{k}\mathrm{t}}$$

(18)

$$\mathrm{M}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}={\mathsf{\varphi }}_0+{\mathsf{\varphi }}_1{\mathrm{d}\mathrm{i}\mathrm{d}}_{\mathrm{k}\mathrm{t}}+{{\mathsf{\varphi }}_2\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{k}\mathrm{t}}+{\mathsf{\zeta }}_{\mathrm{k}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{k}\mathrm{t}}$$

(19)

$${\mathrm{E}\mathrm{f}\mathrm{f}\mathrm{i}\mathrm{c}\mathrm{i}\mathrm{e}\mathrm{n}\mathrm{c}\mathrm{y}}_{\mathrm{k}\mathrm{t}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{d}\mathrm{i}\mathrm{d}}_{\mathrm{k}\mathrm{t}}+{{{{\mathsf{\gamma }}_2\mathrm{M}\mathrm{e}\mathrm{d}\mathrm{i}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}}_{\mathrm{k}\mathrm{t}}+\mathsf{\gamma }}_3\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}}_{\mathrm{k}\mathrm{t}}+{\mathsf{\zeta }}_{\mathrm{k}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\theta }}_{\mathrm{k}\mathrm{t}}$$

(20)

Among them, Mediator is the mediating variable. According to the mediation effect test of the stepwise regression method, in models (18), (19), and (20), if η₁ and Φ₁ are significant, while γ₁ is not significant and γ₂ is significant, this indicates a complete mediating effect. If γ₁ is significant, it indicates a partial mediating effect.

3. Results

3.1. Empirical Results at the Macro Level

3.1.1. Descriptive Statistics and Collinearity Diagnosis

First, based on the data processing, the descriptive statistics of each variable in this study are shown in

Table 6. Descriptive statistics.

Variable	N	Mean	p50	SD	Min	Max
Efficiency	3934	0.317	0.296	0.111	0.103	0.818
Posttreat	3934	0.281	0	0.450	0	1
Aggregation	3934	1.581	1.230	1.158	0.119	6.821
Invest	3934	16.06	16.08	1.101	11.24	18.43
Ope	3934	0.237	0.0900	0.392	0	2.438
Ind	3934	47.12	47.49	10.87	11.70	73.71
Gov	3934	18.89	16.25	10.38	4.262	140.4
Number	3934	6.570	6.558	1.112	2.944	9.018
Stucture	3934	2.277	2.269	0.144	1.831	2.663
GreenFinance	3934	0.308	0.322	0.100	0.0580	0.507
Innovation	3934	1.128	0.300	2.124	0	11.41

.

Secondly, before the formal baseline regression is carried out, in order to test whether there is multicollinearity among the variables, this study conducted a collinearity test on each variable. The test results are shown in

Table 7. Collinearity diagnosis.

Variable	VIF	1/VIF
Number	2.850	0.350
Invest	2.150	0.466
Gov	1.890	0.530
Ind	1.430	0.698
Ope	1.230	0.812
Aggregation	1.180	0.850
Posttreat	1.060	0.944
Mean	VIF	1.680

. According to the test results in Table 7, the variance inflation factor (VIF value) of each variable is less than 5, indicating that there is no serious multicollinearity.

3.1.2. Baseline Regression

This study conducts baseline regression without controlling fixed effects. Columns (1), (3), and (5) of

Table 8. Baseline regression results.

Variable	Baseline Regression Level		Add Some Control Variables		Add All Control Variables
	OLS	FE	OLS	FE	OLS	FE
	(1)	(2)	(3)	(4)	(5)	(6)
Posttreat	0.029 ***	0.060 ***	0.019 ***	0.018 ***	0.012 ***	0.018 ***
	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)	(0.004)
Aggregation			0.002	0.013 ***	−0.008 ***	0.010 ***
			(0.001)	(0.002)	(0.001)	(0.002)
Invest					0.022 ***	0.020 ***
					(0.002)	(0.002)
Ope			0.007	−0.114 ***	0.015 ***	−0.085 ***
			(0.005)	(0.007)	(0.005)	(0.007)
Ind					−0.003 ***	−0.001 ***
					(0.000)	(0.000)
Gov					−0.002 ***	0.000
					(0.000)	(0.000)
Number			0.031 ***	0.010 **	0.011 ***	−0.011 **
			(0.002)	(0.004)	(0.002)	(0.005)
Constant	0.309 ***	0.300 ***	0.101 ***	0.247 ***	0.068 **	0.090 ***
	(0.002)	(0.001)	(0.011)	(0.028)	(0.029)	(0.032)
Year	No	Yes	No	Yes	No	Yes
City	No	Yes	No	Yes	No	Yes
N	3934	3934	3934	3934	3934	3934
R2	0.231	0.269	0.318	0.441	0.393	0.480

Note: Values in parentheses are standard errors, ** p < 0.05, *** p < 0.01.

use OLS regression to represent the baseline regression results. These results show that carbon emission trading policies can positively promote the city’s green economic efficiency. After controlling for city and year effects, the regression results in columns (2), (4), and (6) of the table further consolidate this conclusion, and both results are significant at the 1% significance level, verifying Hypothesis 1. The regression results in column (6) show that the implementation of carbon emission trading policies can increase urban green economic efficiency by an average of 0.018 units. From the control variable level, urban innovation, the investment level, and government intervention have a positive effect on urban green economic efficiency, while the level of opening up, the economic structure level, and the number of industrial enterprises have a negative effect on urban green economic efficiency. This may be because when the city’s economic volume is greater, its energy consumption is greater, and its green level is lower [64].

3.1.3. Robustness Test

(1)

Parallel trend test

Figure 2 shows the dynamic effect of urban green economic efficiency before and after the implementation of the policy. Taking the year before the implementation of the policy as the base period, before the implementation of the policy, the effect value remained relatively stable, and the error lines crossed, which was consistent with the parallel trend hypothesis. After the implementation of the policy, the effect value showed a steady upward trend, and the error lines gradually separated, indicating that the policy may have had a significant positive impact. This result verifies the effectiveness of the carbon emission trading policy.

(2)

Data after joining the national carbon emission trading market

Taking into account the time-lag characteristics of policy effects, this article adds an expanded sample from 2007 to 2023 and compares it with the baseline regression. According to the regression results in column (1) of

Table 9. Robustness test results.

Variable	Robustness Check
	Until 2023	Return of Tail	Replace the Explained Variable	Add Control Variables
	(1)	(2)	(3)	(4)
Posttreat	0.025 ***	0.019 ***	0.027 ***	0.012 ***
	−0.005	−0.003	−0.005	−0.004
Aggregation	0.017 ***	0.008 ***	0.013 ***	0.011 ***
	−0.003	−0.002	−0.003	−0.002
Invest	0.015 ***	0.014 ***	0.012 ***	0.006 **
	−0.002	−0.002	−0.003	−0.003
Ope	−0.111 ***	−0.099 ***	−0.120 ***	−0.081 ***
	−0.009	−0.008	−0.009	−0.007
Ind	−0.001 ***	0	0	0
	0	0	0	0
Gov	0	0.001 ***	0	0
	0	0	0	0
Number	0.001	−0.004	−0.012 *	−0.012 **
	−0.006	−0.004	−0.006	−0.005
Internet				0.022 ***
				−0.003
Constant	0.130 ***	0.111 ***	0.412 ***	0.014
	−0.037	−0.028	−0.042	−0.033
Year	Yes	Yes	Yes	Yes
City	Yes	Yes	Yes	Yes
N	4761	3934	3934	3934
R2	0.468	0.496	0.423	0.427
Variable	Lag by one period		Lag by two periods
	OLS	FE	OLS	FE
	(6)	(7)	(8)	(9)
Pre1	0.005 ***	0.003 ***
	−0.001	−0.001
Pre2			0.006 ***	0.003 ***
			−0.001	−0.001
Aggregation	−0.005 ***	0.010 ***	−0.004 ***	0.010 ***
	−0.002	−0.003	−0.002	−0.003
Invest	0.018 ***	0.025 ***	0.017 ***	0.025 ***
	−0.002	−0.002	−0.002	−0.002
Ope	0.019 ***	−0.040 ***	0.020 ***	−0.040 ***
	−0.004	−0.006	−0.004	−0.006
Ind	−0.003 ***	0	−0.003 ***	0
	0	0	0	0
Gov	−0.002 ***	0	−0.002***	0
	0	0	0	0
Number	0.011 ***	−0.016 ***	0.012 ***	−0.016 ***
	−0.003	−0.006	−0.003	−0.006
Constant	0.124 ***	0.031	0.136 ***	0.033
	−0.034	−0.038	−0.034	−0.038
City	Yes	Yes	Yes	Yes
Year	Yes	Yes	Yes	Yes
N	3934	3934	3934	3934
R2	0.373	0.427	0.375	0.411

Note: Values in parentheses are standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

, the impact of carbon emission trading policy on urban green economic efficiency is still positively significant at the 1% significance level. Compared with the baseline regression results in column (5) and column (6) of Table 8, by 2023, the implementation of carbon emission trading policy can increase urban green economic efficiency by an average of 0.034 units, making the policy effect more significant. This may be because over time, policy implementation in China has improved, market mechanisms have matured, and carbon price signals have become more stable and effective, creating better conditions for enhancing urban green economic efficiency.

(3)

Tail-off return

This article performs 1% tailing processing on the data for econometric regression. In column (2) of Table 9, this article changes the previous data-processing method and uses 1% tailing regression to eliminate extreme values. The regression results in Table 9 (2) show that, under truncated regression, the coefficient of posttreat is 0.019 and is statistically significant at the 1% level. Compared with the baseline regression results in column (5) and the regression results in column (6) of Table 8, the coefficient values exhibit no substantial changes, further validating the positive impact of carbon emission trading on urban green economic efficiency.

(4)

Replace the interpreted variable

The accounting index of urban green economic efficiency used in the benchmark regression of this study is calculated using the super-efficiency SBM model. Column (3) of Table 9 shows the calculated value of urban green economic efficiency obtained by using the DEA-CCR model and the input and output indicators in Table 1, and using this calculated value to replace the results of the SBM model. Table 9 (3) shows that after replacing the explained variable indicator, the coefficient of posttreat is 0.027 and is statistically significant at the 1% level. Due to differences in measurement methods, the coefficient value is larger compared to the regression results in columns (5) and (6) of Table 8. Nevertheless, it can still be concluded that the carbon emission trading policy has a positive impact on urban green economic efficiency, thus verifying the robustness of the baseline regression results.

(5)

Add new control variables

This study refers to the research of Li M and Zhang Q. On the basis of the original control variables, a new control variable, the level of urban Internet (Internet), is added to re-conduct the benchmark regression [75]. The level of urban Internet may be related to the economic level of the city, so it is necessary to control this variable. This study uses the logarithm of the number of international Internet users to calculate that level [48]. Table 9 (4) shows that after incorporating the new control variables, the coefficient of posttreat is 0.012. Compared with the baseline regression results in column (5) and the regression results in column (6) of Table 8, the coefficient value exhibits no significant changes, indicating that the conclusions drawn from the baseline regression results remain valid. Furthermore, the level of Internet development in cities has a positive impact on urban green economic efficiency, with each unit increase in Internet leading to an average increase of 0.022 units in efficiency.

(6)

Lagged explanatory variable

To account for potential time-lag effects in the implementation of carbon emission trading policies, this study conducts robustness checks using a lagged period treatment method. Specifically, we reconstruct the parameters of the explanatory variables by lagging the policy implementation timeline by one period (1 year) and two periods (2 years), respectively. The baseline regression analysis is re-conducted through the difference-in-differences (DID) method, with the regression results presented in columns (6) to (9) of Table 9.

In Table 9, Pre1 denotes the 1-period lagged policy (Pre1) and Pre2 represents the 2-period lagged policy. According to the lagged effect regression results in Table 9, after implementing a 1-year policy lag, the carbon emission trading policy exhibits statistically significant positive effects on urban green economic efficiency at the 1% significance level. The OLS regression results indicate that the policy implementation increases urban green economic efficiency by an average of 0.005 units, while the FE regression shows an average increase of 0.003 units. With a 2-period policy lag, as evidenced by the baseline regression results in columns (8)~(9) of Table 9, the policy maintains its statistically significant positive impact at the 1% level. The OLS regression estimates a 0.006-unit enhancement in urban green economic efficiency, whereas the FE regression demonstrates a 0.003-unit improvement. Although the coefficient magnitudes are smaller than those in the baseline regression results of Table 8, the persistently significant positive effects confirm that the carbon emission trading policy generates stable and progressive enhancements in urban green economic efficiency across temporal dimensions. This pattern robustly validates the consistency of baseline regression outcomes, demonstrating the policy’s temporal resilience in promoting sustainable urban development.

(7)

Endogeneity test

This article refers to the research of Zhou Chang et al. and uses the method of propensity score matching (PSM) combined with difference-in-differences (DID) to overcome the endogeneity problem caused by sample selection bias [31]. In non-randomized experiments, by matching samples of the treatment group and the control group, the two groups are comparable in terms of observable covariates, thereby reducing selection bias. In terms of matching methods, this study refers to the approach of Zhou Di et al. and uses radius matching, nearest neighbor matching, and kernel matching for baseline regression [13]. Through these three matching methods, it is ensured that the matched samples have a good balance in the values of covariates, so as to more accurately estimate the effect of the policy.

Columns (1) to (3) of

Table 10. Endogeneity test results.

Variable	PSM-DID
	Radius Matching	Nearest Neighbor Matching	Kernel Matching
	(1)	(2)	(3)
Posttreat	0.012 ***	0.012 ***	0.012 ***
	(0.004)	(0.004)	(0.004)
Aggregation	−0.007 ***	−0.008 ***	−0.008 ***
	(0.002)	(0.001)	(0.002)
Invest	0.022 ***	0.022 ***	0.022 ***
	(0.002)	(0.002)	(0.002)
Ope	0.016 ***	0.015 ***	0.015 ***
	(0.005)	(0.005)	(0.005)
Ind	−0.003 ***	−0.003 ***	−0.003 ***
	(0.000)	(0.000)	(0.000)
Gov	−0.002 ***	−0.002 ***	−0.002 ***
	(0.000)	(0.000)	(0.000)
Number	0.011 ***	0.011 ***	0.011 ***
	(0.002)	(0.002)	(0.002)
Constant	0.067 **	0.068 **	0.066 **
	(0.029)	(0.029)	(0.029)
Year	Yes	Yes	Yes
City	Yes	Yes	Yes
N	3934.000	3934.000	3924.000
R2	0.590	0.593	0.592

Note: Values in parentheses are standard errors, ** p < 0.05, *** p < 0.01.

show the radius matching results, nearest neighbor matching results, and kernel matching results, respectively. According to the regression results, the carbon emission trading system still shows positive significance for urban green economic efficiency at the 1% significance level, verifying the baseline regression results in Table 8.

3.1.4. Heterogeneity Analysis

We conducted a heterogeneous analysis of cities with different regional locations, resource endowments, city sizes, and industrial bases to observe the effects of policy implementation in each region. Regarding regional heterogeneity, China’s regional development has significant “east–central–west” gradient differences, and regions differ significantly in marketization level, environmental regulation intensity, and green technological diffusion capacity [75,76]; regarding resource endowment heterogeneity, there are systematic differences in economic structure, energy dependence, and environmental governance costs between resource-based cities and non-resource-based cities [77], as resource-based cities have long relied on high-carbon industries, have strong industrial structural rigidity, and have greater resistance to transformation under policy impacts [59]; for urban size heterogeneity, the agglomeration effect and economies of scale of megacities may amplify policy effects, while small and medium-sized cities are limited by their technology absorptive capacity and financial resources, and there is a threshold effect in policy response [78]; and regarding the heterogeneity of the industrial base, there are significant differences in carbon emission structure and emission reduction potential between industry-led cities and service-led cities [79]. The former has a high proportion of heavy industry and has great space for emission reduction but high costs; while the latter has a high proportion of low-carbon service industries and may have lower policy marginal benefits [78].

Table 11. Heterogeneity analysis I.

Variable	Regional Heterogeneity
	East	Middle	West	Northeast
	(1)	(2)	(3)	(4)
Posttreat	0.031 ***	0.002	0.014 **	−0.025
	(0.009)	(0.007)	(0.007)	(0.019)
Aggregation	0.037 ***	0.006	0.001	−0.016
	(0.006)	(0.005)	(0.003)	(0.011)
Invest	0.031 ***	0.034 ***	0.035 ***	−0.003
	(0.006)	(0.004)	(0.003)	(0.008)
Ope	−0.030 ***	−0.056 ***	−0.015 *	−0.435 ***
	(0.009)	(0.021)	(0.008)	(0.053)
Ind	−0.003 ***	0.000	0.002 ***	−0.001
	(0.001)	(0.000)	(0.000)	(0.001)
Gov	−0.001 **	−0.002 **	−0.001 ***	0.004 ***
	(0.001)	(0.001)	(0.000)	(0.001)
Number	−0.022	−0.001	−0.011	0.044 **
	(0.014)	(0.010)	(0.008)	(0.019)
Constant	0.134	−0.226 ***	−0.272 ***	0.163 *
	(0.140)	(0.067)	(0.051)	(0.098)
Year	Yes	Yes	Yes	Yes
City	Yes	Yes	Yes	Yes
N	1190.000	1106.000	1162.000	448.000
R2	0.327	0.180	0.145	0.223

Note: Values in parentheses are standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

presents baseline regressions on cities in different regions based on their geographical locations. The coefficients of the variables are significantly positive in the eastern and western regions, at 0.031 and 0.014, respectively, indicating that the carbon emission trading policy has a positive effect on enhancing green economic efficiency in these areas. Considering China’s regional characteristics, the eastern region, with its advanced economic development and industrial structure, provides more mature conditions for policy implementation, thus yielding significant results. Although the western region is relatively less developed, China’s recent Western Development Strategy and green finance policies have laid a solid foundation for the implementation of the carbon emission trading policy [11]. The coefficients are close to zero in the central region and negative in the northeastern region, suggesting that the policy’s impact is not significant. This may be related to the fact that most pilot cities are located in the eastern region, as well as the high proportion of traditional heavy industries and economic transformation challenges in the central and northeastern regions, which make green transformation relatively difficult [27]. The regional heterogeneity results show that the effects of the carbon emission trading policy vary significantly across different regions. Policymakers need to optimize the regional implementation of the policy by considering factors such as regional economic development levels, industrial structural characteristics, and resource endowments to maximize policy effectiveness.

Table 12. Heterogeneity analysis II.

Variable	Whether It Is Resource-Based		Internal Division of Resources
	Resource-Based	Non-Resource-Based	Growth-Based	Maturity-Based	Decline-Based	Regeneration-Based
	(1)	(2)	(3)	(4)	(5)	(6)
Posttreat	0.011	0.025 ***	0.008	−0.002	0.026 ***	−0.001
	(0.007)	(0.006)	(0.030)	(0.011)	(0.009)	(0.013)
Aggregation	−0.002	0.015 ***	−0.017	−0.001	0.010	−0.017 ***
	(0.005)	(0.003)	(0.025)	(0.007)	(0.008)	(0.006)
Invest	0.008 **	0.033 ***	0.033 **	0.008	−0.001	0.026 ***
	(0.003)	(0.003)	(0.014)	(0.005)	(0.004)	(0.006)
Ope	−0.285 ***	−0.034 ***	−0.489	−0.357 ***	−0.138 ***	−0.068
	(0.030)	(0.006)	(0.312)	(0.043)	(0.034)	(0.059)
Ind	0.000	−0.002 ***	0.002	0.000	0.001 ***	−0.000
	(0.000)	(0.000)	(0.001)	(0.000)	(0.000)	(0.001)
Gov	0.001 ***	−0.000	−0.002	0.000	0.005 ***	0.003 ***
	(0.000)	(0.000)	(0.001)	(0.001)	(0.001)	(0.001)
Number	−0.015 **	−0.015 *	−0.035	−0.002	−0.006	−0.024 *
	(0.008)	(0.008)	(0.026)	(0.011)	(0.011)	(0.012)
Constant	0.242 ***	−0.033	0.000	0.183 **	0.143 **	0.037
	(0.056)	(0.052)	(0.247)	(0.085)	(0.071)	(0.101)
Year	Yes	Yes	Yes	Yes	Yes	Yes
City	Yes	Yes	Yes	Yes	Yes	Yes
N	1582	2324	196	854	322	210
R2	0.122	0.369	0.177	0.155	0.382	0.393

Note: Values in parentheses are standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

presents grouped regressions based on resource endowment differences and further classifies resource-based cities internally. The regression results in Table 12 show that in resource-based cities, the coefficient of posttreat is 0.011 but not significant, indicating that the policy has a limited impact on the green economic efficiency of these cities. This may be because resource-based cities rely heavily on traditional resource industries, with a single industrial structure, making them less adaptable to the carbon emission trading policy [80]. In contrast, in non-resource-based cities, the coefficient of the posttreat variable is 0.025, significant at the 1% level, indicating that the policy has a significantly positive impact on the green economic efficiency of non-resource-based cities. Additionally, the first batch of carbon emission trading pilot cities in China was mostly non-resource-based, suggesting better policy implementation effects. Further analysis of cities categorized by internal resource types shows that the coefficient of the posttreat variable in declining cities is 0.026, also significant at the 1% level, while its impact is not significant in regenerating cities. This indicates that the carbon emission trading policy is particularly effective in promoting green economic efficiency in non-resource-based and declining cities, as declining cities are more likely to achieve green transformation under policy incentives. Policymakers should fully consider the resource types and development stages of cities and formulate differentiated policy support measures for resource-based and regenerating cities. For example, China has strengthened technical assistance and fiscal subsidies for regenerating cities to promote their green economic efficiency.

Table 13. Heterogeneity analysis III.

Variable	City Size		Is It an Old Industrial City?
	Big	Medium-Sized and Small	Yes	No
	(1)	(2)	(3)	(4)
Posttreat	0.018 **	0.021 ***	0.009	0.023 ***
	(0.007)	(0.006)	(0.007)	(0.006)
Aggregation	0.022 ***	−0.001	−0.003	0.021 ***
	(0.004)	(0.004)	(0.003)	(0.004)
Invest	0.032 ***	0.021 ***	0.018 ***	0.028 ***
	(0.004)	(0.003)	(0.003)	(0.003)
Ope	−0.040 ***	−0.039 ***	−0.287 ***	−0.039 ***
	(0.008)	(0.008)	(0.027)	(0.006)
Ind	−0.003 ***	0.000	0.001 **	−0.001 ***
	(0.000)	(0.000)	(0.000)	(0.000)
Gov	−0.000	0.001 ***	0.002 ***	−0.000
	(0.000)	(0.000)	(0.000)	(0.000)
Number	−0.028 ***	−0.008	−0.023 ***	−0.012
	(0.010)	(0.007)	(0.008)	(0.008)
Constant	0.117	−0.002	0.107 **	−0.010
	(0.077)	(0.042)	(0.047)	(0.054)
Year	Yes	Yes	Yes	Yes
City	Yes	Yes	Yes	Yes
N	1400.000	2506.000	1302.000	2604.000
R2	0.368	0.276	0.234	0.328

Note: Values in parentheses are standard errors, ** p < 0.05, *** p < 0.01.

(1)~(2) presents grouped regressions based on city size. In terms of city size, the coefficients of the posttreat variable in large cities and small-to-medium cities are 0.018 and 0.021, respectively, both significant at the 1% level, indicating that the policy has a positive impact on the green economic efficiency of both types of cities, with a slightly greater effect on small-to-medium cities. This may be because small-to-medium cities are more flexible in economic restructuring and green technological application. In contrast, although large cities have larger economic scales, their complex economic structures and higher carbon emission baselines may somewhat weaken the policy’s effectiveness [81].

Table 13 (3)~(4) presents grouped regressions based on industrial foundations. In terms of industrial foundations, the coefficients of the posttreat variable in old industrial cities and non-old industrial cities are 0.009 and 0.023, respectively, with the latter significant at the 1% level. This indicates that the policy has a more pronounced effect on enhancing green economic efficiency in non-old industrial cities. Old industrial cities, with their high proportion of traditional industries, face greater difficulties in transformation and require long-term government support and incentives, resulting in relatively limited current policy effects. In contrast, non-old industrial cities, with fewer traditional industries, have more flexible industrial structures and are more likely to achieve green transformation through policy incentives [82].

The heterogeneity test results provide scientific evidence for policymakers to fully consider the economic development levels, resource endowments, industrial structures, and city sizes of different cities when formulating or adjusting the carbon emission trading system, thereby maximizing policy effectiveness. At the same time, by understanding which types of cities benefit more from the policy, policymakers can adjust and optimize the policy in a targeted manner, achieving a win–win situation for both environmental and economic benefits.

3.1.5. Mechanism Inspection

In order to identify the causal path of carbon emission trading policy promoting urban green economic efficiency at the macro level, this study selects technological innovation, industrial structural upgrading, and green finance as mediating variables and conducts stepwise regression based on the mediating effect test method of Wen Zhonglin and Ye Baojuan [39]. Since the test results of model (18) have been listed in column (6) of the benchmark regression in Table 8,

Table 14. Mechanistic test results.

Variable		Technological Innovation
		Innovation		Efficiency			Science		Efficiency
		(1)		(2)			(3)		(4)
Posttreat		0.127 ***		0.017 ***			0.051 ***		0.022 ***
		−0.029		−0.004			−0.011		−0.005
Aggregation		0.043 ***		0.010 ***			−0.001		0.011 ***
		−0.016		−0.002			−0.006		−0.003
Invest		−0.046 ***		0.020 ***			0.043 ***		0.026 ***
		−0.015		−0.002			−0.005		−0.002
Ope		−0.678 ***		−0.078 ***			−0.121 ***		−0.043 ***
		−0.05		−0.008			−0.013		−0.006
Ind		−0.013 ***		−0.000 *			−0.002 ***		−0.001 **
		−0.001		0			−0.001		0
Gov		−0.001		0			0.002 **		0
		−0.001		0			−0.001		0
Number		0.081 **		−0.012 **			0.088 ***		−0.015 **
		−0.036		−0.005			−0.013		−0.006
Innovation				0.009 ***
				−0.002
Science									0.005 ***
									0
Constant		1.972 ***		0.071 **			−0.917 ***		0.019
		−0.229		−0.033			−0.085		−0.039
Year		Yes		Yes			Yes		Yes
City		Yes		Yes			Yes		Yes
N		3934		3934			3934		3934
R2		0.398		0.484			0.37		0.337
Variable		Upgrade of Industrial Structure
		Advanced Structure		Efficiency			Overall Upgrade		Efficiency
		(5)		(6)			(7)		(8)
Posttreat		0.010 ***		0.016 ***			0.021 **		0.021 ***
		−0.003		−0.004			−0.01		−0.005
Aggregation		0		0.011 ***			−0.015 **		0.012 ***
		−0.002		−0.002			−0.006		−0.003
Invest		0.039 ***		0.012 ***			0.057 ***		0.023 ***
		−0.001		−0.002			−0.005		−0.002
Ope		−0.023 ***		−0.082 ***			−0.035 ***		−0.040 ***
		−0.005		−0.007			−0.013		−0.006
Ind		−0.007 ***		0.001 ***			−0.041 ***		0.001 ***
		0		0			−0.001		0
Gov		0		0			0.005 ***		0
		0		0			−0.001		0
Number		0.024 ***		−0.015 ***			−0.061 ***		−0.013 **
		−0.003		−0.005			−0.013		−0.006
Structure				0.198 ***
				−0.023
Upgrade									0.047 ***
									−0.008
Constant		1.818 ***		−0.268 ***			2.296 ***		−0.083 **
		−0.022		−0.053			−0.082		−0.042
Year		Yes		Yes			Yes		Yes
City		Yes		Yes			Yes		Yes
N		3934		3934			3934		3934
R2		0.662		0.496			0.764		0.445
Variable	Green Finance
	Green Index		Efficiency		Green Fund	Efficiency		Green Equity		Efficiency
	(9)		(10)		(11)	(12)		(13)		(14)
Posttreat	0.018 ***		0.012 ***		0.002 ***	0.021 ***		0.001 **		0.021 ***
	(0.002)		(0.004)		(0.001)	(0.005)		(0.000)		(0.005)
Aggregation	−0.001		0.011***		0.000	0.011***		0.000		0.011 ***
	(0.001)		(0.002)		(0.000)	(0.003)		(0.000)		(0.003)
Invest	0.028 ***		0.011 ***		0.004 ***	0.023 ***		0.002 ***		0.024 ***
	(0.001)		(0.002)		(0.000)	(0.002)		(0.000)		(0.002)
Ope	−0.034 ***		−0.073 ***		−0.005 ***	−0.039 ***		−0.003 ***		−0.040 ***
	(0.003)		(0.007)		(0.001)	(0.006)		(0.001)		(0.006)
Ind	−0.002 ***		0.000		−0.000 ***	−0.000		−0.000 ***		−0.000 *
	(0.000)		(0.000)		(0.000)	(0.000)		(0.000)		(0.000)
Gov	0.001 ***		−0.000		0.000 ***	0.000		0.000 ***		0.000
	(0.000)		(0.000)		(0.000)	(0.000)		(0.000)		(0.000)
Number	−0.004 *		−0.009 *		−0.001	−0.015 ***		−0.001		−0.015 ***
	(0.002)		(0.005)		(0.001)	(0.005)		(0.001)		(0.006)
Green Index			0.337 ***
			(0.037)
Green Fund						0.536 ***
						(0.120)
Green Equity										0.736 ***
										(0.160)
Constant	−0.024 *		0.093 ***		0.002	0.023		−0.001		0.024
	(0.014)		(0.032)		(0.005)	(0.038)		(0.004)		(0.038)
Year	Yes		Yes		Yes	Yes		Yes		Yes
City	Yes		Yes		Yes	Yes		Yes		Yes
N	3934		3934		3934	3934		3934		3934
R2	0.615		0.498		0.404	0.441		0.413		0.441

Note: Values in parentheses are standard errors, * p < 0.1, ** p < 0.05, *** p < 0.01.

only shows the regression results of models (19) and (20).

1

Mediating Role of technological innovation

The implementation of the carbon emission trading policy can incentivize enterprises through carbon price signals, prompting them to increase investment in green technology research and development [83,84] to reduce carbon emission costs or obtain additional benefits from carbon quota trading. This policy-driven technological innovation not only enhances the technical efficiency of enterprises but also fosters technological advancement across the entire industry, thereby exerting a significant positive impact on urban-level green economic efficiency [85,86]. The results of the mechanism test in Table 14 (1)~(4) validate this assertion.

This study measures technological innovation from the perspective of technological expenditure and the number of patent applications at the city level. Firstly, according to the mechanism test results in Table 14 (1)~(2), the coefficient of the posttreat variable in the technological innovation model is 0.127, and it is significant at the 1% significance level, indicating that the implementation of carbon emission trading policy has a positive impact on technological innovation efficiency. At the same time, for posttreat in column (2), the coefficient is 0.017, which is significant at the 1% significance level. This is consistent with the baseline regression results in Table 8 (6), and the difference is small, indicating that technological innovation plays a partial intermediary role between carbon emission trading policy and urban green economic efficiency.

Secondly, as shown in Table 14 (3)~(4), in column (3), the coefficient of posttreat is 0.051, which is significant at the 1% level, indicating that the carbon emission trading policy can enhance the level of technological innovation in cities. In column (4), the coefficient of posttreat is also 0.051, significant at the 1% level, suggesting that for every unit increase in posttreat, efficiency increases by 0.022 units. It is consistent with the baseline regression results in Table 8 (6), and the difference is significant. This further validates that technological innovation plays a partial mediating role in the relationship between the carbon emission trading policy and urban economic efficiency. Technological innovation is a key factor in promoting urban green development in the context of carbon emission trading policy [49,84].

2

Mediating Role of Industrial Structural upgrading

The implementation of the carbon emission trading policy exerts cost pressure on high-emission enterprises through the quota allocation mechanism, while simultaneously creating market advantages for low-carbon and environmentally friendly enterprises [87]. This policy effect drives the reallocation of resources from energy-intensive and high-emission traditional industries to low-carbon and efficient emerging industries, thereby promoting the optimization and upgrading of the industrial structure [88].

This study uses industrial structural progress and overall industrial upgrading to measure the upgrading of the industrial structure at the urban level. Firstly, as shown in columns (5) to (8) of Table 14, the coefficient of the posttreat in column (3) is 0.010 and is significant at the 1% significance level, indicating that the carbon emission trading policy has a positive impact on industrial structural upgrading. In column (4), the posttreat coefficient is 0.016, which is significant at the 1% significance level. It is consistent with the baseline regression results in Table 8 (6), and the difference is small. This shows that industrial structural upgrading plays a partial intermediary role between carbon emission trading policy and urban green economic efficiency.

Secondly, as shown in Table 14 (7)~(8), in column (7), the coefficient of posttreat is 0.021, which is significant at the 1% level, indicating that the carbon emission trading policy can promote overall industrial progress in cities. In column (4), the coefficient of posttreat is also 0.021, significant at the 1% level, suggesting that for every unit increase in posttreat, efficiency increases by 0.021 units. Compared to the baseline regression in Table 8 (6), the policy effect is more pronounced, further validating that industrial structural upgrading plays a partial mediating role in the relationship between the carbon emission trading policy and urban green economic efficiency. Industrial structural upgrading also plays an important role as an intermediary variable between carbon emission trading policy and urban green economic efficiency [89,90].

3

Mediating Role of green finance

The implementation of the carbon emission trading policy guides financial institutions to allocate more funds to low-carbon and environmentally friendly sectors through the carbon market mechanism, thereby driving the development of green finance [41]. Simultaneously, by channeling capital towards green industries, green finance can reduce carbon emissions while promoting economic growth, achieving a synergistic enhancement of green development and economic efficiency [26], as demonstrated in columns (9) to (14) of Table 14.

To verify the mediating role of green finance, this paper examines three dimensions of green finance, using the green finance index calculated by the entropy method, green funds, and green equities, to measure the mediating effect of green finance. Firstly, the coefficient of the posttreat in column (9) is 0.018, which indicates that the carbon emission trading policy has a positive impact on green financial efficiency at the 1% significance level. The coefficient of the posttreat variables in column (10) is 0.012, which is significant at the 1% significance level, consistent with the baseline regression results in Table 8 (6). Although there is a slight deviation from the 0.018 coefficient value in Table 8 (6), it does not affect the causal effect between variables, indicating that green finance serves as the third intermediary variable, and its intermediary effect has also been verified.

Secondly, as shown in Table 14 (11)~(12), in column (11), the coefficient of posttreat is 0.002, which is significant at the 1% level, indicating that the carbon emission trading policy can enhance the value of green funds at the city level. In column (4), the coefficient of posttreat is 0.021, significant at the 1% level, suggesting that for every unit increase in posttreat, efficiency increases by 0.021 units. Compared to the baseline regression in Table 8 (6), the policy effect is more pronounced, confirming that green finance plays a partial mediating role in the relationship between the carbon emission trading policy and urban green economic efficiency.

Furthermore, as shown in Table 14 (13)~(14), in column (13), the coefficient of the posttreat variable is 0.001 and is statistically significant at the 1% level. This result indicates that the implementation of the carbon emission trading policy can significantly improve the maturity of the green equity market at the city level. In column (14), the coefficient of posttreat is 0.021, significant at the 1% level, suggesting that for every unit increase in posttreat, the efficiency increases by 0.021 units. Compared to the baseline regression in Table 8 (6), the policy effect is more significant, once again verifying that green finance plays a partial mediating role in the relationship between the carbon emission trading policy and urban green economic efficiency.

Finally, the coefficients of the mediator variable are all positive at the 1% significance level, further verifying the rationality and reliability of the mediating effect.

The mechanism test results of this study show that the carbon emission trading policy improves the efficiency of the city’s green economy by promoting technological innovation, upgrading the industrial structure and developing green finance, verifying hypothesis 2. While implementing the carbon emission trading policy, policymakers and implementers should focus on the incentives for technological innovation, the optimization of the industrial structure, and the construction of the green financial system to achieve the green and sustainable development of the city’s economy [61].

3.2. Empirical Results at the Micro Level

3.2.1. Micro Game Evolution Model

A previous article revealed the impact of a carbon emission trading policy on urban green economic efficiency and its impact mechanism from a macro level, and existing studies have shown that carbon emission trading also has a promoting effect on enterprises [18,65,91]. As enterprises are the main participants in the carbon market, their decision-making behavior determines the effectiveness of carbon emission trading policies to a certain extent, thereby affecting the industrial chain where the enterprises are located and the level of green economic efficiency of the city [25]. This article will start from the decision-making behavior of enterprises at the micro level. Combining the principles of microeconomic effect maximization and profit maximization, a previous study explores the impact of the micro-enterprise level on urban green economic efficiency based on enterprise decision-making in different types of carbon emission rights markets [27].

First, we build a carbon rights trading game pricing model for enterprises in a quasi-market. The core trading methods of the quasi-market carbon emission trading market rely on administrative regulation, regional consultation methods, etc., and comprehensively consider the overall interests of both parties in carbon rights transactions to promote carbon rights trading, thereby effectively promoting the optimal allocation of carbon emissions rights [73,92].

Its assumptions and model deductions are as follows:

(1)

Basic assumptions

Assume that the information of both parties in the carbon rights transaction is complete, that is, the seller of the carbon rights transaction is clear about the current market demand for carbon rights, and the seller of the carbon rights can set the price according to the price that maximizes benefits. Assume that the relationship between carbon emissions Q and the current output of the enterprise Y is Y = F(Q), which is a strictly increasing convex function.

(2)

Carbon rights buyer

Assuming that the output price of the carbon emission rights of the carbon rights buyer in production is R₁, and the market carbon rights trading volume selected when the carbon rights trading price is P is Q₁ (P), then its income I₁ is as follows:

$${\mathrm{I}}_1={\mathrm{R}}_1·{\mathrm{F}}_1\left({\mathrm{Q}}_1\right(\mathrm{P}\left)\right)-\mathrm{P}·{\mathrm{Q}}_1\left(\mathrm{P}\right)$$

(21)

The first half of the Formula (21) represents the output profit of the enterprise that purchases carbon rights, and the second half represents the cost that the enterprise has to pay to purchase carbon rights.

The maximization condition is as follows:

$${\mathrm{I}}_1^{\prime }=0$$

That is,

$${\mathrm{F}}_1^{\prime }({\mathrm{Q}}_1(\mathrm{P}))={\displaystyle \frac{\mathrm{P}}{{\mathrm{R}}_1}}$$

$$\therefore {\mathrm{Q}}_1(\mathrm{P})={\mathrm{F}}_1^{\prime -1}\left({\displaystyle \frac{\mathrm{P}}{{\mathrm{R}}_1}}\right)$$

(3)

Carbon rights seller

Suppose the enterprise cost function of the carbon rights seller is C₂(Q₂(P)), with C₂ including the emissions reduction cost of the enterprise to introduce advanced technology to reduce carbon emissions, labor costs, etc. The seller hopes to set the transaction price of carbon rights at a level that maximizes profits. Then, its benefit I₂ is as follows:

$${\mathrm{I}}_1={\mathrm{R}}_2·{\mathrm{F}}_2\left({\mathrm{Q}}_2\right(\mathrm{P}\left)\right)+\mathrm{P}·{\mathrm{Q}}_1\left(\mathrm{P}\right)-{\mathrm{C}}_2\left({\mathrm{Q}}_2\right(\mathrm{P}\left)\right)$$

(22)

The maximization condition is the following:

$${\mathrm{I}}_2^{\prime }=0$$

That is,

$${\mathrm{F}}_2^{\prime }({\mathrm{Q}}_2(\mathrm{P}))={\displaystyle \frac{{\mathrm{C}}_2-\mathrm{P}}{{\mathrm{R}}_2}}$$

$$\therefore {\mathrm{Q}}_2(\mathrm{P})={\mathrm{F}}_2^{\prime -1}\left({\displaystyle \frac{{\mathrm{C}}_2-\mathrm{P}}{{\mathrm{R}}_2}}\right)$$

When there are multiple carbon rights buyers in the market, the total demand for carbon rights by each enterprise constitutes the total demand for carbon emission rights in the carbon trading market. Under the ideal state of microeconomic decision-making, the carbon rights trading price should be equal to the marginal cost price of carbon rights trading, and the Nash equilibrium solution under the complete information market is obtained (${\mathsf{\epsilon }}_1$, ${\mathsf{\epsilon }}_2$):

$${\mathsf{\epsilon }}_1=-{\displaystyle \frac{{\mathrm{A}}_1(\mathrm{M}+{\mathrm{A}}_2\mathsf{\epsilon }(1+{\mathsf{\epsilon }}_2\left)\right)}{{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_1)({\mathrm{A}}_1\mathsf{\epsilon }-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_1\left)\right)}}$$

$${\mathsf{\epsilon }}_2=-{\displaystyle \frac{{\mathrm{A}}_2(\mathrm{M}+{\mathrm{A}}_1\mathsf{\epsilon }(1+{\mathsf{\epsilon }}_1\left)\right)}{{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_2)({\mathrm{A}}_1\mathsf{\epsilon }-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2\left)\right)}}$$

Secondly, we construct a pricing model for carbon rights trading under the market mechanism, where there are many carbon rights demanders and sellers in the carbon rights trading market. All parties conduct a series of carbon rights trading price games from their own interests [93]. Finally, based on the demand and supply of carbon rights in the trading market, the carbon rights trading price is determined by the comprehensive trading market mechanism on the basis of the full game between the two parties [57,94].

The model assumptions and deductions are as follows:

(1)

Basic assumptions

The carbon rights trading demanders and sellers cannot fully understand each other’s situation, that is, the information in the market is incomplete. Assume that the buyer of carbon rights trading has a preliminary valuation of the carbon rights in the transaction as A₁, and the seller of carbon rights trading has a preliminary valuation of the carbon rights unit sales volume in the transaction as A₂; then, the two parties will bargain around their respective preliminary valuations.

In the carbon trading market, buyers and sellers face imperfect information, as they cannot fully or clearly understand each other’s circumstances. It is assumed that the buyer initially estimates the quantity of carbon rights to be traded as A₁, while the seller estimates the unit price of the carbon rights as A₂. The two parties will then engage in a negotiation process based on their respective initial estimates.

② Let the parameter Ψ₁ (Ψ₁ ∈ [−1, 0]) represent the bidding strategy of the carbon rights buyer. The bidding strategy is assumed to be P₁ = A₁(1 + ε₁). In the carbon rights trading price game, the carbon rights buyer tends to lower the trading price and will not accept a transaction price higher than A1.

③ Let the parameter Ψ₂ (Ψ₂ ∈ [−1, 1]) represent the quoting strategy of the carbon rights seller. The quoting strategy is assumed to be P₂ = A₂(1 + ε₂).

In the negotiation process, the carbon rights seller wants P₂ and A₂, but due to national policies and considerations regarding the buyer’s use of the carbon rights and overall social benefits, the seller sometimes has to make concessions in terms of profit and may accept a transaction price lower than A₂. When ε₂ = −1, it represents a transaction with no payment (i.e., a free transaction).

④ To simplify the calculation, it is assumed that the multiplier effects resulting from the carbon rights transaction between both parties are not considered.

(2)

Pricing model and solution

The process of the pricing game between two parties in carbon rights trading is similar to a process of seeking a solution to maximize mutual utility [95]. Based on the solution of microeconomic utility maximization, this study analyzes the game process between the two parties in the carbon rights trading market.

If P₁ is less than P₂, the carbon rights buyer cannot accept P₂, resulting in a failure to reach an agreement on the price, meaning the negotiation fails, and both parties’ utility is zero.

If P₁ is more than P₂, both parties can reach a mutually agreed price:

$$\mathrm{P}={\mathrm{P}}_2+\mathsf{\epsilon }({\mathrm{P}}_1-{\mathrm{P}}_2)={\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)+\mathsf{\epsilon }\left[{\mathrm{A}}_1\right(1+{\mathsf{\epsilon }}_1)-{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_2\left)\right]$$

(23)

where ε is the negotiation price parameter, ε ∈ [0, 1], and its specific value is ultimately determined by the negotiation between the two parties. The expected payoff functions for both parties, U₁ and U₂, are, respectively, as follows:

$${\mathrm{U}}_1={\mathrm{A}}_1-\mathrm{P}$$

(24)

$${\mathrm{U}}_2=\mathrm{P}-{\mathrm{A}}_2$$

(25)

In accordance with the principle of profit maximization, the seller of carbon rights seeks a final transaction price, P, that exceeds their own valuation, A₂, while the buyer aims for a final negotiated price, P, that is lower than their own valuation, A₁. Consequently, both parties endeavor to maximize their respective utilities. Based on these assumptions, the following equation can be derived:

The expected payoff function U₁* of the carbon rights buyer is as follows:

$$\begin{gathered} {\mathrm{U}}_1^{\mathrm{*}}=({\mathrm{A}}_1-\mathrm{P})\times \mathrm{P}({\mathrm{P}}_1\ge {\mathrm{P}}_2) \\ =\{{\mathrm{A}}_1-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)-\mathsf{\epsilon }[{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)\left]\right\}\times \mathrm{P}({\mathsf{\epsilon }}_2\le {\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)/{\mathrm{A}}_2-1) \\ =0.5\{{\mathrm{A}}_1-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)-\mathsf{\epsilon }[{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)\left]\right\}\times \left[{\mathrm{A}}_1\right(1+{\mathsf{\epsilon }}_1)/{\mathrm{A}}_2] \end{gathered}$$

(26)

The expected utility function U₂* of the carbon rights seller is as follows:

$$\begin{gathered} {\mathrm{U}}_2^{\mathrm{*}}=(\mathrm{M}+\mathrm{P}-{\mathrm{A}}_2)\times \mathrm{P}({\mathrm{P}}_1\ge {\mathrm{P}}_2) \\ =\{\mathrm{M}+{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)+\mathsf{\epsilon }[{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)]-{\mathrm{A}}_2\}\times \mathrm{P}({\mathsf{\epsilon }}_2\ge {\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)/{\mathrm{A}}_1-1) \\ =[1-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)/{\mathrm{A}}_1]\times \{\mathrm{M}+{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)+\mathsf{\epsilon }[{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)-{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)]-{\mathrm{A}}_2\} \end{gathered}$$

(27)

where M represents the indirect benefits brought by the carbon rights transaction, such as carbon emission rewards granted by local governments, reduced financing costs due to improved ESG ratings, etc. Generally, M << A₁, M << A₂.

It is assumed that there is a Nash equilibrium strategic price (P₁, P₂), where (ε₁, ε₂) satisfy the following two functions: ε₁ ∈ $\underset{{\mathsf{\epsilon }}_1}{\max }{\mathrm{U}}_1^{\mathrm{*}}$, ε₂ ∈ $\underset{{\mathsf{\epsilon }}_2}{\max }{\mathrm{U}}_2^{\mathrm{*}}$. Then, (P₁, P₂) under (ε₁, ε₂) is the equilibrium solution for the price negotiation.

$${\displaystyle \frac{\partial {\mathrm{U}}_1^{\mathrm{*}}}{\partial {\mathsf{\epsilon }}_1}}={\displaystyle \frac{{\mathrm{A}}_1}{{2\mathrm{A}}_1}}[{\mathrm{A}}_1-{\mathrm{A}}_2-{\mathrm{A}}_2{\mathsf{\epsilon }}_2-\mathsf{\epsilon }{\mathrm{A}}_1(1+{\mathsf{\epsilon }}_1)+\mathsf{\epsilon }{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2)]+{\displaystyle \frac{1}{2}}[{\mathrm{A}}_1/{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_1)](-\mathsf{\epsilon }{\mathrm{A}}_1)=0$$

(28)

Solving that gives

$${\mathsf{\epsilon }}_1={\displaystyle \frac{1-2\mathsf{\epsilon }}{2\mathsf{\epsilon }}}-{\displaystyle \frac{{\mathrm{A}}_2(1+{\mathsf{\epsilon }}_2-\mathsf{\epsilon }·{\mathsf{\epsilon }}_2-0.5\mathsf{\epsilon })}{2{\mathrm{A}}_1\mathsf{\epsilon }}}$$

(29)

and similarly,

$${\mathsf{\epsilon }}_2={\displaystyle \frac{\mathrm{M}}{2{\mathrm{A}}_2(\mathsf{\epsilon }-1)}}+{\displaystyle \frac{{\mathrm{A}}_1(2\mathsf{\epsilon }-1+\mathsf{\epsilon }·{\mathsf{\epsilon }}_1)}{2{\mathrm{A}}_2(\mathsf{\epsilon }-1)}}+{\displaystyle \frac{1-2\mathsf{\epsilon }}{2(\mathsf{\epsilon }-1)}}$$

(30)

To further simplify the analysis, assuming A₁ = A₂ = A, we then derive ε₁ and ε₂ as follows:

$${\mathsf{\epsilon }}_1=-1+{\displaystyle \frac{\mathrm{M}}{3\mathrm{A}·\mathsf{\epsilon }}}$$

(31)

$${\mathsf{\epsilon }}_2={\displaystyle \frac{2\mathrm{M}}{3\mathrm{A}(\mathsf{\epsilon }-1)}}-{\displaystyle \frac{\mathsf{\epsilon }}{2(\mathsf{\epsilon }-1)}}$$

(32)

From this parameter, the Nash equilibrium solution of the final price can be obtained (P₁, P₂).

3.2.2. Sensitivity Analysis

(1)

From the perspective of information transparency, when under complete information, the game equilibrium solution is stable at P = (C₂ + M)/(1 + ε₁ − ε₂), avoiding pricing distortions caused by information bias. Enterprises can accurately identify the marginal emission reduction cost difference, carbon price P. Closely following the true emission reduction cost C₂ forms an effective price signal; under incomplete information, the buyer overestimates the emission reduction cost (A₁ < C₂) or the seller underestimates the transaction income (A₂ < M), leading to the loss of potential transactions and the equilibrium price deviating from the Pareto optimal, which may cause some companies with high emission reduction costs to exit the market due to low pricing [89].

By comparing corporate decisions under complete information and incomplete information, we can see that increased information opacity directly compresses the bargaining space, allowing carbon prices to truly reflect environmental costs and driving companies to achieve emission reduction resource allocation through transactions rather than confrontational negotiations.

(2)

From the perspective of marginal emission reduction costs C, under complete information, the decline in C₂ is immediately transmitted to carbon prices P (because P is linearly related to C₂), stimulating buyers to increase purchases, sellers to obtain clear price signals, and planning of emission reduction technology research and development in advance; under incomplete information, C₂ changes are partially absorbed by information friction, and price adjustments lag, resulting in “emission reduction dividend dissipation”. Companies are unable to predict long-term price trends and tend to engage in short-term arbitrage rather than technological investment [96].

Through the sensitivity analysis of emission reduction costs, it can be seen that the complete information market has the “amplifier effect” of cost transmission, allowing the green benefits of technological progress to spread rapidly through the price mechanism, indicating that information transparency enhances the effectiveness of emission reduction incentives.

(3)

From the perspective of the Nash equilibrium price, in a complete information market, information symmetry forces ε₁ and ε₂ to approach 0 (bargaining power is eliminated by market supply and demand), and the equilibrium price is determined endogenously by the real emission reduction cost and external benefits M. For example, government supervision in quota trading in the steel industry makes ε₂ ≤ 0.1. In an incomplete information market, ε₁ and ε₂ become the dominant pricing factors. For example, new energy companies gain higher bargaining power due to the opacity of technical information. Price deviation causes carbon trading to deviate from the essence of emission reduction, resulting in speculative behavior, such as carbon quota hoarding [92].

Through sensitivity analysis of Nash equilibrium prices, it is found that complete information weakens the influence of negotiation elasticity parameters and makes carbon prices return to the essential attributes of environmental goods, indicating that eliminating bargaining distortions is a key mechanism to improve green efficiency.

Indirect benefits M play a leverage role in negotiations between the two parties. Under complete information, the increase in M is fully internalized as the company’s expected benefits. The buyer can clearly predict the trend of M changes, which can help it form a stable long-term purchase plan. Under incomplete information, it is difficult for companies to quantify the actual value of M, resulting in “policy arbitrage” phenomena, such as false reporting of the direction of green credit use. In addition, the increase in M is used by some companies to increase bargaining power rather than actual emission reduction, resulting in “incentive leakage” behavior [59], which damages the city’s economic benefits and then its green economic benefits.

Through the four-fold sensitivity deduction, it can be seen that under the condition of complete information, the carbon price formation mechanism is highly sensitive to changes in core parameters such as C₂ and M and is rigid to non-market factors such as ε₁ and ε₂. This characteristic enables the carbon trading market to have both environmental cost reflection efficiency and anti-interference ability. The interaction of each variable verifies the theoretical core of Hypothesis 3—information transparency systematically improves the transaction efficiency of enterprises through three paths: ① compressing bargaining space, ② accelerating technological diffusion, and ③ optimizing policy transmission. It increases the enthusiasm of enterprises to participate in carbon emission trading, maximizes the effectiveness of carbon emission trading policies at the micro level, and improves green economic efficiency. This analysis provides a theoretical basis for the construction of policies such as mandatory information disclosure and unified carbon asset assessment standards.

In summary, the sensitivity analysis of micro level enterprise decision-making is shown in

Table 15. Sensitivity analysis.

Key Variable	Path of Action	Complete Information Market Performance	Market Comparison with Imperfect Information
Information transparency	Information transparency↑ → Information error↓ → Transaction efficiency↑	The carbon price P is close to the real emission reduction cost C2, and the game equilibrium is stable	Valuation deviation leads to loss of transactions and price deviates from Pareto optimality
Marginal emission reduction cost C2	C2↓ → Technological upgrade power↑ → Supply Q↑	C2 changes are completely transmitted to P, stimulating buyers to increase purchases and technical pre-research	Price adjustment lags behind, and companies tend to take short-term arbitrage
Buyer’s ability to lower price ε1	ε1↑ → P↓ → Seller’s incentive to reduce emissions is damaged	The bargaining power is eliminated by the market (ε1 tends towards 0), and the price is determined endogenously by C2 + M	ε1 dominates pricing, lowering P and causing high-cost enterprises to exit
Seller’s ability to raise prices ε2	ε2↑ → P↑ → Buyer’s costs surge	Government regulation restricts ε2 (such as ε2 ≤ 0.1), and prices return to environmental attributes	ε2 over-valuation leads to speculation (such as quota hoarding)
Indirect benefits M	M↑ → Reservation price↑ → Market activity↑	M enhancement is fully internalized into expected returns, driving long-term transactions (such as ESG policy cases)	M valuation ambiguity leads to “incentive leakage” (such as false reporting of green credit use)

Note: ‘↑’ represents an increase, increase, or increase in a variable or indicator, ‘↓’ represents a decrease, decrease, or decrease in a variable or indicator, and ‘→’ represents a causal relationship or direction of influence, where changes in the preceding variable lead to changes in the following variable.

:

3.2.3. Analysis of Trading Efficiency in Two Types of Markets

According to the concept of Pareto improvement, in a scenario where complete information is available, such as in a quasi-market, there is no price adjustment that can benefit one party without harming the other. Consequently, in the presence of complete information, the carbon credit trading market is considered Pareto optimal. In this context, firms are more likely to invest in improving production technologies, thereby achieving higher green economic efficiency and enhancing their overall green economic performance.

In contrast, in situations of incomplete information, the final carbon credit trading price represents an equilibrium outcome resulting from negotiations in which both parties make concessions. This leaves room for potential Pareto improvements, indicating that a carbon credit trading market with incomplete information is Pareto inefficient.

According to economic principles, under conditions of complete information, the equilibrium is achieved when MC = MR = P*. Under conditions of complete information, equilibrium is achieved when the marginal cost (MC) equals the marginal revenue (MR), which in turn equals the price (P*). However, under incomplete information, the price curve, denoted as P(y), shifts downward due to competitive bargaining among firms, as shown in Figure 3 for comparative static analysis. This process continues until the price curve tangentially touches the average cost (AC) curve, resulting in a price P* under incomplete information. As depicted in the graph, the price P* under complete information is lower than P** under incomplete information. This discrepancy arises from the transaction costs imposed by information barriers in the latter scenario. The cost of purchasing carbon credits at a price above P* represents an opportunity cost for the buyer—funds that could otherwise be invested in technological improvements. Therefore, regardless of whether the final transaction price is higher or lower, the deviation of P** from P* results in increased transaction costs. These higher costs lead to a loss of potential investment in technological advancements, thereby reducing the incentive for firms to adopt emission reduction technologies. Consequently, this undermines efforts to improve overall green energy efficiency.

In summary, the efficiency of corporate decision-making strategies in different markets can affect the enthusiasm of enterprises to participate in the carbon market and improve technology, and can have a certain impact on the policy effect of carbon emission trading policies. The existence of transaction costs damages the opportunity cost of enterprises to improve production technology, inhibits enterprises from improving green total factor productivity, and thus affects the green economic efficiency of the city where the enterprise is located, further verifying Hypothesis 3.

4. Discussion

As global climate change becomes increasingly severe, the green and low-carbon transformation of cities is urgently needed. The carbon emission trading policy, as a market-based environmental regulatory tool, offers a sustainable green option for cities worldwide.

In this context, existing research has conducted in-depth discussions on the carbon emission trading policy. At the macro level, Jung and Song (2023) used international practical experience as an example and found that carbon emission trading policies can promote the green economic efficiency of cities [27]. Zhang et al. (2025) used China’s practical experience as an example to explore the relationship between carbon emission trading policies and the green development efficiency of cities. The study found that carbon emission trading policies can improve the green development efficiency of cities and promote the carbon balance between regions [28]. Chen et al. (2021) conducted a natural experiment based on China’s carbon emission trading policy to explore the positive impact of the carbon emission trading policy on urban energy efficiency [97]. This study also explores the impact of the carbon emission trading policy on urban green economic efficiency from a macro perspective. The empirical results show that the carbon emission trading policy can, on average, increase urban green economic efficiency by 0.018 units, validating the findings of existing studies.

Around the micro level, Xi et al. (2024) used China’s carbon emission trading policy as an example and found that the policy promotes green technological innovation in enterprises [98]. Jia et al. (2022) explored the carbon emission reduction pressure and green technological innovation of enterprises. The study found that enterprises can actively engage in technological innovation while reducing carbon emissions, further demonstrating the incentive effect of carbon emission trading policies on micro-enterprises [99]. Guo et al. (2023) found that carbon emission trading policies can promote energy conservation and emission reduction in enterprises and affect their financial performance [100]. At the micro level, this study employs a game pricing model. The research results indicate that reasonable policy interventions can incentivize enterprises to achieve green transformation, thereby positively influencing urban green economic efficiency, further underscoring the importance of the carbon emission trading policy.

Building on the existing literature, this study makes innovations in the research perspective, methodology, and content. In terms of the research perspective, existing studies predominantly adopt a single-dimensional approach, focusing either on the macro or micro level, lacking a comprehensive macro–micro dual perspective. This study, however, adopts a dual macro–micro perspective, demonstrating the promoting effect of the carbon emission trading policy on urban green economic efficiency. In terms of research methodology, while existing studies often rely on a single measurement method, this study combines the difference-in-differences (DID) model and evolutionary game theory. The DID regression results reveal that the carbon emission trading policy can increase urban green economic efficiency by an average of 0.018 units. The evolutionary game model results show that information transparency, emission reduction costs, and indirect benefits during transactions significantly influence enterprises’ trading behavior in the carbon market. Under complete information, enterprises can reduce transaction costs and enhance emission reduction efficiency, thereby affecting urban green economic efficiency at the micro level. In terms of research content, while the existing literature has explored the mechanisms through which the carbon emission trading policy exerts macro-level impacts, such as enhancing urban green innovation capabilities and promoting carbon emission reduction, this study further investigates these mechanisms. In addition to technological innovation, this study introduces industrial structural upgrading and green finance as mediating variables, measuring these mechanisms across multiple dimensions. The mediation effect tests reveal that the carbon emission trading policy enhances urban green economic efficiency through technological innovation, industrial structural upgrading, and green finance. All three mechanisms show significant positive effects on urban green economic efficiency at the 1% significance level. Specifically, a one-unit increase in technological innovation raises urban green economic efficiency by an average of 0.005 to 0.009 units; a one-unit improvement in industrial structure increases it by 0.047 to 0.198 units; and a one-unit increase in green finance boosts it by 0.337 to 0.736 units. Additionally, this study conducts heterogeneity analysis based on the regional location, resource endowment, and industrial base. The findings indicate that the carbon emission trading policy has a more pronounced impact on eastern, non-resource-based, resource-declining, small and medium-sized cities, as well as cities without an old industrial base.

In summary, while validating the positive impact of the carbon emission trading policy on urban green economic efficiency, this study makes significant breakthroughs in research perspective, methodology, and content compared to the existing literature. It addresses the limitations of current studies and provides insights and references for future multi-perspective analyses of the carbon emission trading policy.

Finally, although this study offers important theoretical contributions to the analysis of the carbon emission trading policy’s impact on urban green economic efficiency, it has certain limitations. First, the research is primarily based on data from pilot cities in China. Future studies should consider making international comparisons to validate the policy’s applicability across different countries and regions. Second, while the evolutionary game model used in this study provides useful insights into enterprise behavior, it simplifies the complex decision-making processes of enterprises in the market. Future research should employ more sophisticated models to account for irrational behaviors and their impact on policy effectiveness. Third, while analyzing the policy’s impact on cities, this study primarily focuses on intra-city dynamics, with insufficient consideration of inter-city and inter-regional linkages. Future research should delve deeper into the interconnected effects of the carbon emission trading policy across cities and regions.

5. Conclusion and Policy Recommendations

This study analyzes the impact of carbon emission trading policies on urban green economic efficiency from a dual macro–micro perspective and draws the following key conclusions. First, the carbon emission trading policy has significantly enhanced urban green economic efficiency. After policy implementation, green economic efficiency increased by an average of 0.018 units. This improvement is more pronounced in eastern cities, non-resource-based cities, small-to-medium-sized cities, and non-old industrial cities, while the effect is weaker in central–western regions, resource-based cities, large cities, and old industrial cities. Specifically, for eastern cities, policy implementation increased green economic efficiency by an average of 0.031 units; for non-resource cities, it increased by 0.025 units; for small-to-medium-sized cities, it increased by 0.021 units; and for non-old industrial cities, it increased by 0.023 units. Therefore, in future policy promotion, regional disparities must be considered, and flexible policy measures should be adopted to maximize policy effectiveness.

Second, in the process of enhancing urban green economic efficiency, intermediary pathways such as technological innovation, industrial structural upgrading, and green finance play a critical role. Policymakers should focus on these key variables during policy implementation to achieve better execution outcomes.

Finally, at the micro level, this study reveals how firms’ decision-making behaviors in the carbon market influence green economic efficiency. The study finds that, compared to incomplete information markets, firms in complete information markets can more effectively enhance urban green economic efficiency indirectly through carbon emission trading mechanisms.

By integrating macro and micro perspectives, this study theoretically and empirically demonstrates the promoting effect of carbon emission trading policies on urban green economic efficiency. Based on the findings, the following policy recommendations are proposed:

(1)

Promote regional differentiated policy design

Since the effects of carbon emission trading policies in different regions vary significantly, it is recommended that governments of various countries fully consider the differences in economic development levels, resource endowments, and industrial structures in various regions when formulating and implementing policies. For eastern and non-resource-based cities, the depth and breadth of the carbon market can be increased, green technology innovation and green financial development can be further promoted, and market-based means can be used to guide funds to green industries. For the central and western regions and resource-based cities, policies should focus on transitional measures, promote technology guidance, and offer green investment support to avoid negative impacts caused by a single policy.

(2)

Strengthen green financial support

Green finance plays an important role in improving the efficiency of the green economy. Governments should further promote the development of the green financial market, especially in the fields of green bonds and green funds. Through policies such as fiscal subsidies and tax incentives, financial institutions are encouraged to provide low-interest loans for green projects to help enterprises and local governments better implement carbon reduction targets. In addition, the government should strengthen supervision of the green financial market to ensure that funds flow to green projects and avoid the emergence of “greenwashing”.

(3)

Improve corporate carbon emission management capabilities

The behavior of enterprises in the carbon emission rights trading market directly affects the policy effect. Therefore, governments should strengthen the construction of corporate carbon emission management capabilities, help enterprises establish a sound carbon emission monitoring and reporting system, help enterprises understand their own carbon emission levels, and formulate reasonable emission reduction measures. The government can also encourage enterprises to increase investment in green technological innovation and green production through technical support and financial subsidies, thereby improving the overall green economic efficiency.

(4)

Strengthening cooperation between the government and enterprises to jointly promote green transformation

Cooperation between government and business is critical to driving the transition to a green economy. It is recommended that the government and enterprises establish more cooperation platforms to jointly formulate green development strategies and promote green infrastructural construction and green technological application. The government can encourage enterprises to participate in green transformation through policy guidance and incentive measures to improve their competitiveness in the green economy.

To sum up, the successful implementation of carbon emission trading policy relies on the comprehensive cooperation of regional differentiated policies, green financial support, corporate behavior guidance, technological innovation promotion, industrial structural adjustment, international cooperation, and cooperation between the government and enterprises. Through these measures, we can effectively improve the efficiency of the green economy, promote the long-term implementation of carbon emission trading policies, and promote the process of China’s green and low-carbon development.