Environmental Governance Pressure and the Co-Benefit of Carbon Emissions Reduction: Evidence from a Quasi-Natural Experiment on 2012 Air Standards

Abstract

Achieving carbon emission reduction synergy is vital for green economic transformation. This study examines whether environmental governance pressure promotes such synergy, simultaneously driving carbon reduction and pollution control. Leveraging the 2012 Ambient Air Quality Standard as a quasi-natural experiment, we employ a continuous difference-in-differences (DID) method on 250 prefecture-level cities from 2009 to 2022. Our findings reveal that increased environmental governance pressure significantly reduces both the total amount and intensity of carbon emissions, demonstrating a clear synergistic effect. This synergy is positively correlated with reductions in major air pollutants (e.g., SO₂ and NO_x), indicating that pressure curbs both the total amount and intensity of carbon emissions. Mechanistic analysis shows that this pressure primarily curtails carbon emissions by fostering green innovation and accelerating cleaner energy transitions, with no ‘green paradox’. It also promotes low-carbon industrial restructuring while reducing reliance on end-of-pipe pollution management. Heterogeneity analysis indicates stronger synergistic effects in regions with lower emission reduction costs (e.g., western China, less developed industrial bases). We recommend robust central government environmental regulation policies to amplify local governance pressure, strengthen carbon reduction synergy, and facilitate continuous green development.

1. Introduction

Achieving a significant reduction in air pollutants and carbon dioxide emissions is of the utmost importance for facilitating a sustainable and environmentally friendly societal transformation. The co-occurrence, shared origins, and processual links between air pollutants and carbon dioxide emissions [1,2] provide a theoretical foundation for local governments to integrate carbon emissions reduction into pollution prevention and control policies. Consequently, enhancing this synergistic effect has become a critical priority for the ongoing promotion of green development. Government agencies are increasingly focused on leveraging this synergy to foster continued green growth. Since 2022, the State Council and seven other departments, including the Ministry of Ecology and Environment, have issued the “14th Five-Year Plan” for comprehensive energy conservation and emission reduction, and the “Pollution Reduction and Carbon Synergy Program”.

These initiatives underscore the importance of leveraging existing ecological and environmental institutional frameworks to synergistically promote low-carbon development. They also encourage innovation in policy and measures, as well as the optimization of governance practices. Nevertheless, despite these efforts, the reality is that, while primary air pollutant emissions have declined, carbon dioxide emissions have continued to rise. Specifically, since 2015, emissions of sulfur dioxide, ammonia, and nitrogen compounds have decreased from approximately 18 million tons to 5 million tons and 13 million tons, respectively. In contrast, there has been an increase in the emissions of carbon dioxide. The annual average of these emissions has risen from 9.2 billion tons to approximately 10 billion tons, thus establishing the nation as the world’s largest carbon emitter (Figure 1). This discrepancy prompts critical inquiries: what are the factors hindering the anticipated synergistic reduction in regional CO₂ greenhouse gas emissions, despite the observed reduction in major air pollutants? The purpose of this study is to examine the reasons why the synergistic emission reduction effect of air pollutants and carbon dioxide is not significant.

Existing research identifies the green paradox effect and local governments’ preference for end-of-pipe governance as significant factors contributing to the limited synergistic effect of carbon emission reduction. First, some studies suggest that poorly designed environmental governance policies—for instance, those with short-term targets, those lacking regional coordination, or those that fail to align with carbon reduction goals—can induce a green paradox, exacerbating both air pollutant and carbon dioxide emissions [3], thereby impeding the synergistic management of these pollutants. This literature often examines the green paradox through the lens of horizontal competition among local governments regarding environmental regulation. It argues that, under China’s decentralized environmental governance system, incomplete implementation of local environmental policies, coupled with a “bottom-up competition” strategy in environmental regulation [4], leads to a time lag between policy announcement and implementation [5,6,7,8], ultimately fostering the green paradox [9]. Furthermore, some scholars investigate methods to enhance the synergistic effect of carbon emission reduction, finding that the failure of environmental policies to promote a low-carbon transition in regional energy consumption is central to the green paradox [5]. Conversely, the Porter effect, induced by environmental policies under certain conditions, can improve regional innovation [10] and significantly mitigate the green paradox [11]. Second, other scholars categorize environmental governance strategies at the industrial sector level into end-of-pipe and source-based approaches. End-of-pipe governance refers to treating pollutants at the end of the production process (e.g., flue gas desulfurization), while source-based governance emphasizes preventing pollution and carbon emissions at the source through methods including industrial restructuring and energy transition. They observe that government officials, under the pressures of an official promotion system [12,13], tend to favor end-of-pipe governance. While end-of-pipe governance effectively curtails regional air pollutant emissions, it struggles to achieve the synergistic control of air pollutants and carbon dioxide emissions [14] and may even, in some cases, exacerbate regional carbon emissions [15]. Moreover, studies at the industrial sector level indicate that, unlike end-of-pipe measures, source-based strategies, such as enhanced industrial restructuring, are more conducive to achieving synergistic control of both pollutants and CO₂ emissions [16,17].

In summary, the existing literature has analyzed the impact mechanisms of the synergistic management of air pollutants and CO₂, primarily focusing on the green paradox effect and end-of-pipe governance. However, there is a significant gap in research regarding the influence of environmental governance pressure mechanisms on this synergistic management, particularly from the perspective of local government pressure, which is a key driver of environmental policy implementation. In China, environmental governance mechanisms are characterized by both top-down and bottom-up pressure transfer paths. With the increasing emphasis on the Chinese modernization concept of harmonious coexistence between humanity and nature, environmental governance pressure has intensified. Local governments face both top-down pressure through environmental performance rankings, accountability measures, and official interviews, as well as bottom-up pressure from public opinion. As both the formulators and implementers of regional ecological and environmental policies, local governments’ environmental governance behavior is fundamentally driven by this environmental governance pressure [15,18]. This pressure can not only constrain local governments’ tendency towards “competition at the bottom” in environmental regulation and improve the efficiency of governance, but can also stimulate their motivation to engage in environmental governance and to transform their governance methods [19,20]. Environmental governance pressure, coupled with well-designed ecological and environmental policies, which leverage market-based instruments as the core and are complemented by command-and-control tools and public participation, can facilitate synergistic governance [11], achieving a combined policy effect greater than the sum of its parts (“1 + 1 > 2”). This synergistic effect is crucial for advancing environmental pollution prevention and control, as well as promoting the in-depth and sustained green transformation of the economy and society. Therefore, there is an urgent need to analyze the impact of environmental governance pressure on the synergistic effect of carbon emissions reduction, specifically from the perspective of local government pressure, and to delineate the underlying mechanisms of action.

This study makes several significant contributions to the existing literature on environmental governance and low-carbon development.

First, it innovatively shifts the research focus from environmental policies to the institutional environmental governance pressure on local governments as a critical, yet under-explored, driver of carbon emission reduction synergy. While previous studies often examine policy impacts, we highlight how this intrinsic governance pressure fundamentally shapes the co-benefits of carbon reduction and pollution control. By empirically validating the synergistic effect of this pressure, our work not only expands the theoretical framework of environmental governance but also offers crucial empirical insights for fostering sustainable green and low-carbon transformations.

Second, methodologically, this paper introduces a novel and more granular indicator for environmental governance pressure. Unlike binary proxies used in prior research, our indicator is objectively constructed from local government work reports using text analysis and word frequency, providing a continuous measure that captures the nuanced intensity of pressure across different regions and over time. Furthermore, by leveraging the 2012 Ambient Air Quality Standard as a quasi-natural experiment and employing a continuous difference-in-differences (DID) methodology, we establish a more robust causal relationship between governance pressure and regional carbon emissions, effectively addressing endogeneity concerns and the heterogeneity in pressure application.

Third, we systematically uncover the intricate causal mechanisms through which environmental governance pressure promotes carbon reduction synergy. Beyond merely identifying the effect, we empirically delineate the specific pathways: fostering green innovation, accelerating cleaner energy transitions (without evidence of a ‘green paradox’), promoting low-carbon industrial restructuring, and reducing local governments’ reliance on end-of-pipe pollution management. This detailed mechanistic analysis provides actionable insights for designing more effective and targeted environmental policies.

2. Institutional Background and Theoretical Analysis

2.1. Institutional Background

The central government has repeatedly underscored the pivotal function of environmental governance pressure in the context of regional environmental management. Since 2005, the central government has promulgated a series of policy documents aimed at enhancing environmental governance at the local level. These include The Decision on Strengthening Environmental Protection through the Implementation of Scientific Development, The Interim Provisions on Sanctions for Violations of Laws and Disciplines in Environmental Protection, and The Opinions on the Establishment of a Mechanism for Assessing and Evaluating Party and Government Leadership and Leadership Cadres for Promoting Scientific Development. These documents were designed to increase pressure on local governments to improve environmental governance through environmental protection legislation and specific policies. However, the manipulation of non-transparent regional environmental quality data by local officials [21] undermined the effectiveness of these assessment and promotion mechanisms in exerting effective pressure for environmental governance. This state of affairs persisted until 2012, when the former Ministry of Environmental Protection (MEP) and the State Administration of Quality Supervision, Inspection and Quarantine (AQSIQ) jointly promulgated The Ambient Air Quality Standards (2012). This marked a shift, initiating a period of gradually increasing pressure on local governments to manage air pollutants.

The Ambient Air Quality Standards (2012) stipulated a comprehensive implementation plan for the novel air quality standards in prefectural-level cities, which was executed in three phases. The initial phase, which commenced on 1 January 2013, encompassed 74 pilot cities in pivotal regions, including Beijing–Tianjin–Hebei, the Yangtze River Delta, and the Pearl River Delta, in addition to municipalities, provincial capitals, and cities with autonomous development agendas. The subsequent phase, with a deadline of 1 January 2014, encompassed 116 cities, with a particular focus on key cities and national environmental protection model cities. The third phase of the program encompassed the remaining 177 prefectural-level cities, with an implementation deadline of 1 November 2014. After the implementation of these novel standards, national ambient air monitoring stations initiated the consistent monitoring and publication of real-time urban air quality information, specifically the Air Quality Index (AQI)—a comprehensive indicator that synthesizes concentrations of major pollutants including PM_2.5, PM₁₀, SO₂, NO₂, O₃, and CO. This system was designed to prevent governmental interference at any level in the reported air quality monitoring results within their respective jurisdictions, thereby significantly reducing the potential for local officials to manipulate or conceal regional environmental data.

On one hand, the new standard, when viewed through the lens of China’s environmental governance model, characterized by a top-down approach, serves to heighten the pressure on local governments to enhance their environmental performance. This heightened pressure manifests in increased expectations for accountability and in the form of mandatory interviews. First, the implementation of the new standard provides real and effective environmental data to the higher government departments and the central environmental protection department. This enables the central and higher governments to optimize and innovate the ranking mechanism based on real environmental data. It also increases the strength of the ranking and assessment and enhances the ranking pressure faced by local governments. Second, the central government and higher-level governments can adjust and optimize the accountability mechanism to address the “bull nose” of the local government’s environmental governance. This allows for the more direct enforcement of environmental governance by the local government and enhances the accountability pressure on the local government’s environmental governance. The central environmental protection department has the capacity to enhance the interview mechanism by leveraging the public release of real-time environmental information. This approach enables the identification and mitigation of the most egregious environmental pollutions perpetrated by local government officials. Furthermore, the implementation of a precautionary approach, as outlined by the pertinent legislation, can serve to augment the environmental governance process, thereby enhancing the efficacy of the interview mechanism. On the other hand, according to the bottom-up logic of public opinion pressure, after the implementation of the new standards, it will be easier for the public to access the real air quality data of their cities and form public opinion pressure; at the same time, those who have suffered damage and civil environmental protection organizations can carry out environmental lawsuits and environmental protection activities based on the publicly available real-time environmental data, which will form public opinion pressure to promote the local government’s proactive environmental governance. The Ambient Air Quality Standards (2012) have imposed augmented pressure on regional governments to manage air pollutants. Given the existence of a correlation between regional air pollutants and greenhouse gases, such as carbon dioxide, which share characteristics of common origin, source, and process, this study focuses on analyzing and revealing the causal relationship between local governments’ environmental management pressure and the synergistic effect of carbon emissions reduction in the process of air pollution management.

2.2. Theoretical Analysis

2.2.1. Modeling

First, the economic mechanism and analytical logic of the construction of the relevant theoretical model are elaborated on the relationship between the pressure of environmental governance and the synergistic effect of carbon emissions reduction. We introduce the local government (F), as well as regional air pollutant emission reduction $Q_1$ and carbon dioxide emission reduction $Q_2$. Fossil energy combustion is not only the main source of greenhouse gases, such as carbon dioxide, but also the main source of air pollutants, such as SO₂, NO_x, and other particulate matter produced by combustion [1]. Air pollutants and carbon dioxide have the same root, the same source, and the same process, but the relationship between air pollutant emission reductions $Q_1$ and carbon dioxide emission reductions $Q_2$ is complex. On the one hand, if the local government adopts the environmental regulation “competition by the bottom” strategy in environmental governance, this will exacerbate the regional fossil energy consumption and lead to the green paradox effect, ultimately increasing the regional pollutants and carbon dioxide emissions, weakening the synergistic effect of carbon emissions reduction [5]. Concurrently, ecological environmental protection policies can promote green technology innovation through the Porter effect under specific conditions. This, in turn, can improve energy use efficiency, promote the synergistic emission reduction in air pollutants and carbon dioxide, and strengthen the carbon emissions reduction synergy effect [11]. The latter effect depends on the relative sizes of the green paradox effect and the Porter effect. On the other hand, from the perspective of local governments’ inclination towards end-of-pipe and source governance, the end-of-pipe approach can effectively curtail the emission of a specific air pollutant. However, it is incapable of synergistically regulating a range of pollutants and carbon dioxide. Furthermore, it can result in a decline in the reduction in pollutant emissions, while carbon dioxide emissions may not decline but rather increase in such circumstances [11]. The impact of environmental governance pressure on the synergistic control of air pollutants and carbon dioxide is significant, primarily through altering the local government’s environmental regulation “competition at the bottom” strategy and environmental governance [15,20]. The modeling analysis is executed in two sequential steps.

In the first step, the synergistic control of air pollutants and CO₂ is analyzed when the central policy is not effective at enhancing the environmental governance pressure on local governments. This suggests that local governments face lower emissions reduction targets, and their environmental governance decisions are characterized by the use of environmental regulations as a tool for fiscal competition, incomplete enforcement of environmental regulations, and the widespread use of the “competition at the bottom” strategy; at the same time, the preference for an end-of-pipe approach to environmental governance does not synergize with the suppression of regional air pollutant and carbon dioxide emissions and even leads to a green paradox effect that exacerbates air pollutant and carbon dioxide emissions.

In the second step, we analyze the synergistic management of air pollutants and carbon dioxide after the central government promulgates relevant policies that increase pressure on local governments to manage the environment. This suggests that local governments will face higher emission reduction targets and abandon the “race to the bottom” strategy of environmental regulation. They will have to increase the intensity of environmental regulation and accelerate the low-carbon transformation of energy sources while promoting regional green innovation through the Porter effect. Additionally, original end-of-pipe management cannot achieve higher emissions reduction targets; therefore, local governments must strengthen the green transformation of industrial structures and implement other source management methods. Environmental governance pressure will not expand the green paradox effect or intensify end-of-pipe governance. Ultimately, it can synergize the control of regional air pollutants and carbon dioxide emissions by accelerating the transition to low-carbon energy, promoting regional green innovation, and facilitating the green transformation of industrial structures.

The main modeling premise of this study is as follows: environmental governance pressure does not directly affect regional pollutant and carbon dioxide emissions. However, it changes the pollutant emission reduction target requirements faced by the local government—that is, it affects the critical values of pollutant emissions $\underset{\_}{Q}$ and $\overline{Q}$ of the jurisdictional area and ultimately affects the synergistic management of regional air pollution and carbon dioxide. Set $\overline{Q_1}>Q_1>\underset{¯}{Q_1}>0$, the meaning of the inequality is: when the pressure of environmental governance is small, the local government adopts the environmental regulation “competition by the bottom” strategy and, through the goal of governance, it can achieve the regional emission reduction target $\underset{¯}{Q_1}$. When the pressure of environmental governance is high, local governments abandon the “bottom-up competition” strategy of environmental regulation and adopt the source management approach to achieve emission reduction targets $\overline{Q_1}$. Environmental governance pressure determines the critical value of pollutant emissions in the area under the jurisdiction of the local government, the local government observes the critical value of pollutant emissions and makes the optimal decision according to its own utility function, and the optimal decision of the local government determines the synergistic effect of carbon emission reduction in environmental governance policies. In addition, local governments not only face an environmental governance performance assessment but also face the more important economic development performance assessment, so the local government’s objective function is as follows: if air pollutants in the jurisdiction are not lower than the emission reduction target, they must maximize the economic development results of the region.

When the local government faces weak environmental governance pressure (the regional emission reduction is not less than $\underset{¯}{Q_1}$), the objective function of the local government is as follows: if regional air pollutant emissions reduction is not less than $\underset{¯}{Q_1}$, they must maximize the regional economic development performance (equivalent to minimizing the pollutant abatement cost, unit abatement cost of $c$). Thus, the government’s behavior can be described as the choice to minimize the air pollutant abatement cost under certain constraints:

$${\min }_{\left\{Q_1\right\}}{Q}_1\times c$$

(1)

$$s.t.Q_1\ge \underset{¯}{Q_1}$$

(2)

When local governments face strong environmental governance pressure (jurisdictional emission reductions of not less than $\overline{Q_1}$), the local government objective function remains unchanged, but the jurisdictional air pollutant emission reductions in the constraints should be raised to not less than $\overline{Q_1}$:

$${\min }_{\left\{Q_1\right\}}{Q}_1\times c$$

(3)

$$s.t.Q_1\ge \overline{Q_1}$$

(4)

2.2.2. Model Solving

Before the implementation of Ambient Air Quality Standards (2012) in cities, local governments faced weak environmental governance pressure. The optimization Formula (2) only needs to satisfy $Q_1\ge \underset{¯}{Q_1}$. Since $\overline{Q_1}>Q_1>\underset{¯}{Q_1}>0$, regardless of whether the local government changes the “race to the bottom” strategy of environmental regulation and the preference for end-treatment, regional emission reductions can meet the constraint conditions, that is, regional emission reduction in pollutants and carbon dioxide can be $(Q_1,Q_2)=(\underset{¯}{Q_1},0)$ or $(Q_1,Q_2)=(\overline{Q_1},N{q}_2)$.

When the pollutant emission reduction in the prefecture-level city is $\underset{¯}{Q_1}$, the objective function is $\underset{¯}{Q_1}\times c$; when the regional emission reduction is $\overline{Q_1}$, the objective function is $\overline{Q_1}\times c$, since $\overline{Q_1}>\underset{¯}{Q_1}$, $\overline{Q_1}\times c>\underset{¯}{Q_1}\times c$, so the solution of the optimization problem (2) is $(Q_1^{*},Q_2^{*})=(\underset{¯}{Q_1},0)$. In this case, the local government adopts the environmental regulation “competition by the bottom” strategy and the end of the governance approach: that is, it can achieve the goal of pollutant emissions reduction.

After the implementation of the Ambient Air Quality Standards (2012) in cities, local governments faced strong environmental governance pressure. Optimization Equation (3) requires the satisfaction of $Q_1\ge \overline{Q_1}$. Because $\overline{Q_1}>Q_1>\underset{¯}{Q_1}$, the local government has to abandon the “race to the bottom” strategy of environmental regulation and change the regional environmental governance. In this scenario, the regional pollutant and carbon dioxide emission reduction is $(Q_1^{*},Q_2^{*})=(\overline{Q_1},N{q}_2)$, and the local government can realize the emission reduction target $\overline{Q_1}$ under the stronger environmental governance pressure by increasing the intensity of environmental regulation and strengthening the source management method.

2.2.3. Model Summary

Before the implementation of The Ambient Air Quality Standards (2012), local governments faced weak environmental governance pressures, and the reduction in air pollutants and carbon dioxide under optimal decision-making condition was $(Q_1^{*},Q_2^{*})=(\underset{¯}{Q_1},0)$. After the implementation of the new standards, local governments face stronger environmental governance pressure, and the optimal decision-making conditions for air pollutant and carbon dioxide emission reductions are $(Q_1^{*},Q_2^{*})=(\overline{Q_1},N{q}_2)$, indicating that the environmental governance pressure of local governments can synergize to promote regional carbon dioxide emission reduction. Accordingly, this study proposes Hypothesis 1 to be tested:

Hypothesis 1:

The implementation of the new air quality standards is expected to exert environmental governance pressure on local governments, thereby synergistically reducing regional carbon dioxide emissions, resulting in a significant carbon emission reduction effect.

The analysis also showed a premise for the carbon emission reduction effect: $\overline{Q_1}>\underset{¯}{Q_1}$, that is, environmental governance pressure can promote regional carbon dioxide emissions reduction through a synergistic effect while reducing air pollutants. The increase in regional air pollutant emissions reduction is the regulating variable generated by the carbon emissions reduction effect. With the increase in air pollutant emissions reduction $Q_1$, the pressure of environmental governance is more conducive to inhibiting regional carbon emissions. Based on the above analysis, this study proposes Hypothesis 2 to be tested:

Hypothesis 2:

Following the implementation of the new air quality standards, a greater reduction in regional pollutants will correlate with a stronger synergistic carbon emission reduction effect resulting from local government environmental governance pressures.

Based on the above analysis process, we can determine that local government environmental governance pressure mainly promotes regional CO₂ emissions reduction through two paths: On one hand, the environmental governance pressure prompts local governments to increase the intensity of regional environmental regulation, which restrains the “bottom-up competition” strategy of local governments’ environmental regulation and blocks the mechanism of the green paradox effect. The pressure on environmental governance, on the other hand, will change the intensity of the regional environmental regulation. On the other hand, the pressure of environmental governance will change the preference of local governments for end-of-pipe governance and ultimately inhibit regional carbon emissions through the low-carbon transformation of industrial structure and other source governance methods. Based on the above analysis, we propose Hypothesis 3 for testing:

Hypothesis 3:

Subsequent to the implementation of the new air quality standards, it is hypothesized that environmental governance pressure will primarily generate significant carbon emissions reduction synergies through channels, including enhancing regional green innovation levels, accelerating the green transformation of energy structures, and fostering the low-carbon transformation of industrial structures.

3. Research Design

3.1. Empirical Strategy

The Ambient Air Quality Standards (2012) and its implementation program were chosen as exogenous shocks based on the following considerations: ① The implementation of the new standards changed the environmental governance pressure faced by local governments. As mentioned in the institutional background section, the implementation of the new air quality standards enhances the ranking pressure, interview pressure, accountability pressure, and public opinion pressure faced by local governments. ② The implementation of the new standards will have an impact on regional carbon emissions mainly through synergistic effects. The implementation of the new standards will not directly affect regional carbon dioxide emissions but will reduce regional air pollutant emissions while lowering regional carbon emissions by increasing the pressure on local governments’ environmental governance; in line with the objective of this study, the causal relationship between the pressure on local governments’ environmental governance and the synergistic effect of carbon emission reduction is assessed. ③ Since 2012, all prefecture-level cities in China have gradually implemented the new air quality standards in three phases. After 2015, all prefecture-level cities in China had implemented the new standards. It is difficult to strictly divide the treatment and control groups in an empirical study. However, it is possible to use the heterogeneity of environmental governance pressures faced by local governments after the policy’s implementation to construct a sequential double-difference model and identify potential causality.

Based on the above discussion, accurately measuring local environmental governance pressure is key to identifying causality in continuous double-difference models. In existing studies, Zhang et al. [18] used dummy variables to measure regional environmental governance pressure based on the exogenous shock of implementing the 2012 Ambient Air Quality Standards. However, this method could not effectively identify the heterogeneous environmental governance pressure faced by local governments. Li et al. [22] employed the concept of ranking pressure to assess the environmental pollution control pressure faced by local governments. This approach was informed by the observation that an air quality ranking exists among prefecture-level cities within each province. Drawing on the measure of environmental governance ranking pressure and considering that ambient air quality standards were implemented in all prefecture-level cities across the country in 2015, this study takes the degree of change in the annual number of polluted days (AQI > 100) in cities in the province in 2022 in terms of their descending rankings relative to the 2015 descending rankings, to measure the heterogeneous environmental governance pressure on local governments. As demonstrated in Figure 2A, the findings reveal a broad distribution of ranking pressures on prefecture-level municipal governments. This distribution encompasses both municipalities that have ascended by 10 places in the rankings and those that have descended by the same amount. This observation suggests that local governments encounter heterogeneous environmental governance pressures following the implementation of the new standards.

Given that the utilization of environmental governance ranking pressure lacks efficacy in measuring accountability pressure, interview pressure, and public opinion pressure faced by local governments, it is imperative to develop a more precise metric for assessing environmental governance pressure. In light of the aforementioned considerations, we find that, in comparison with the previously revised air quality standards (2000), the Ambient Air Quality Standards (2012) primarily entail the recalibration of the statistical rigor of the original PM₁₀ particulate matter and the incorporation of PM_2.5 particulate matter as a contaminant. Meanwhile, we find that, in the 2009–2022 work reports of prefecture-level municipal governments, the keywords PM_2.5 and PM₁₀ began to appear after the implementation of the new standards in 2012. However, the above keywords did not appear in the work reports of prefecture-level municipal governments before 2012 (Figure 2B); the frequency of keywords PM_2.5 and PM₁₀ is, to some extent, a reflection of the environmental governance pressure on local governments due to the implementation of the new air quality standards. However, further evidence is necessary to substantiate this claim.

The frequency with which the keywords PM_2.5 and PM₁₀ appear in prefectural-level city government work reports is indicative of the environmental governance pressure on local authorities. Consequently, variations in the frequency of these keywords indicate shifts in ranking pressure related to local environmental governance. The present study offers further empirical support for the methodology employed to measure environmental governance pressure by analyzing the correlation between ranking pressure and lexical frequency change. As demonstrated in Figure 2C, keyword frequency exhibits a significant positive correlation with environmental governance ranking pressure. This indicates that higher keyword frequency in prefectural-level city government work reports corresponds to greater pressure on environmental pollution governance rankings. Consequently, the occurrence frequency of PM_2.5 and PM₁₀ keywords in these reports serves as an indicator of environmental governance ranking pressure. Furthermore, Figure 2D depicts the distribution of environmental governance pressures based on keyword frequency. This pattern aligns with the use of ranking pressure as a measurement metric, confirming significant differentiation in environmental governance pressures across local governments. Notably, while some prefectural-level cities show the complete absence of these keywords in their government work reports, others record occurrences exceeding 14 instances. The preceding analysis demonstrates that measuring environmental governance pressure through PM_2.5 and PM₁₀ keyword frequency in prefectural-level city government work reports is methodologically sound. Nevertheless, to address potential measurement bias from lexical frequency that may induce causal misattribution in extreme scenarios, this study employs keyword frequency for baseline estimations and utilizes ranking pressure in robustness tests. This dual-metric approach assesses the consistency of results across measurement methodologies.

3.2. Regression Model Setup

To empirically assess the relationship between environmental governance pressure and carbon emission reduction synergies, we construct a continuous difference-in-differences model:

$$C_{i,t}=\alpha +{\beta }_{1}ope{n}_{i,t}\times pressur{e}_i+{\beta }_{2}ope{n}_{i,t}+\theta X_{i,t}+{\eta }_i+{\mu }_t+{\epsilon }_{i,t}$$

(5)

In the model presented in Equation (5), the dependent variable is defined as C_i,t, which represents the total carbon dioxide emissions and carbon emission intensity of city i in year t. We posit that the carbon emission reduction synergy effect will affect the total regional carbon emissions and intensity. However, if the local government primarily reduces output to reduce regional pollutant and carbon dioxide emissions, the local government’s environmental governance pressure may have the exact opposite effect on the total amount of carbon emissions and intensity. Therefore, we simultaneously select the total amount of carbon emissions and the intensity of carbon emissions of the city as the explanatory variables.

The core explanatory variable $ope{n}_{i,t}\times pressur{e}_i$ is the interaction term between the new standard implementation variable and the environmental governance pressure exerted by local governments. The implementation of the new air quality standard in year t in prefecture-level city i is indicated by the variable $ope{n}_{i,t}$, with a value of 1 if the standard is implemented and 0 if it is not. The variable $pressur{e}_i$ is related to local government environmental governance pressure, reflecting the environmental governance pressure faced by prefecture-level cities i following the implementation of the policy. The coefficient of the core explanatory variable, denoted by ${\beta }_1$, is estimated to evaluate the change in carbon emissions of prefecture-level cities when environmental governance pressure increases by one unit following policy implementation in comparison with the period preceding policy implementation.

Variable $\alpha$ is defined as the intercept term of the empirical model. It is imperative to note that the variable $X_{i,t}$ constitutes a series of control variables, while $\theta$ represents the estimated coefficients of said control variables. In the context of this study, ${\mu }_t$ denotes city fixed effects, ${\mu }_t$ represents year fixed effects, and ${\epsilon }_{i,t}$ is the residual term.

3.3. Variable Definition

(1) Explained variables. Accurately measuring the CO₂ emissions of prefecture-level cities is the most important task in this empirical research, and, at this stage, academics mainly adopt two methods to measure the CO₂ emissions of prefecture-level cities. Shan et al. [23] measured the CO₂ emissions of prefecture-level cities in China at the prefecture level from 1997 to 2022. The measurement was based on an energy balance sheet covering 47 economic sectors, 17 fossil fuels, and 9 industrial products. The IPCC Regional Carbon Accounting Standard was used to develop the balance sheet. Chen et al. [24] estimated county-level carbon emissions during the same period by applying the particle swarm optimization–backpropagation (PSO-BP) algorithm to integrate DMSP/OLS and NPP/VIIRS satellite imagery, with county-level emissions aggregated to derive city-level values. This study utilizes the CO₂ emission data of prefecture-level cities, as reported by Shan et al. [23], with the following rationales in mind: first, the assessment of CO₂ emissions from prefecture-level cities based on energy consumption has garnered significant recognition among the scholarly community [25,26]. Second, the reliability of utilizing county-level carbon emission data to derive city-level emissions may be compromised by the phenomenon of spatial carbon leakage. The city’s carbon emissions data were obtained from the China Carbon Accounting Database (CEADS).

(2) Core explanatory variables. As outlined in the empirical strategy section, this study leverages the implementation of the Ambient Air Quality Standards (2012) as a quasi-natural experiment. It measures heterogeneous environmental governance pressures faced by local governments by quantifying the frequency of the keywords PM_2.5 and PM₁₀ in municipal government work reports. The specific methodology is as follows. First, we perform word segmentation on annual municipal government work reports and remove stop words. Second, we standardize different expressions—e.g., consolidating “fine particulate matter” and “particulate matter with a diameter less than 2.5 μm” into “PM_2.5” and “inhalable particulate matter” and “particulate matter with a diameter less than 10 μm” into “PM₁₀; this process is carried out to eliminate biases arising from varying terminology. Finally, the frequency of these keywords was counted to serve as an indicator of a city’s environmental governance pressure. For robustness testing, we further standardized word frequencies using Z-scores to mitigate the impact of variations in report length.

(3) Control variables. Drawing on the research methods of Zhang et al. [18], Yi et al. [27], and Guo et al. [28], we select the following control variables to characterize the economic and social characteristics of prefecture-level cities: ① Openness of the city to the outside world (degree): the share of foreign direct investment in the city as a percentage of GDP; ② Prefectural government size (gov): local government budget expenditures as a share of GDP; ③ Urban gross domestic product per capita and its square terms (gdp and gdps); ④ Level of urban human capital (human): ratio of the number of students enrolled in general higher education in the region to the total population of the region at the end of the year; ⑤ Total population of prefecture-level cities (pop); ⑥ Average annual urban rainfall (rain); ⑦ Average annual urban temperature (temp). Logarithms were taken for the total population of the prefecture-level city, the average annual rainfall of the prefecture-level city, and the average annual temperature of the prefecture-level city.

(4) Other variables. ① Urban green innovation level (GI): the number of green patent applications in prefecture-level cities. The higher the number of green patent applications, the higher the regional green innovation level. ② City environmental regulation intensity (ERS): based on the city’s industrial wastewater, industrial exhaust, and industrial sulfur dioxide emissions, drawing on the measurement methods of Yuan et al. [29] and Peng et al. [30]. The larger the ERS index, the higher the environmental regulation intensity. ③ The level of green transformation of the city’s energy structure (EGT): the proportion of energy consumption accounted for by coal. The smaller the proportion, the higher the level of the green and low-carbon energy consumption structure. ④ The level of green transformation of urban industrial structure (SC): the scale of high-carbon industries in prefecture-level cities. In the first step, the industries with the highest total carbon dioxide emissions, such as petroleum processing, coking and nuclear fuel processing, chemical raw material and chemical product manufacturing, ferrous metal smelting and calendaring, non-metallic mineral products, and non-ferrous metal smelting and calendaring, are selected as high-carbon industries based on the total amount of carbon emissions from the industries. In the second step, drawing on the measurement method of Shen et al. [31] and using the database of Chinese industrial enterprises in 2009–2015, the main business income at the enterprise level is first summed up to the industry level of prefecture-level cities, and finally summed up to the level of prefecture-level cities, to obtain the scale of high-carbon industries of the prefecture-level cities. The larger the scale of high-carbon industries of the prefecture-level cities, the lower the level of green transformation of the industrial structure. ⑤ End treatment (ET): city sulfur dioxide removal rate; the higher the SO₂ removal rate, the more cities use end-treatment to reduce pollutant emissions.

Due to missing data and unavailability, we exclude cities with missing data on CO₂ emissions, as well as cities under the jurisdiction of Xinjiang, Tibet, and Taiwan. The final research sample includes 250 prefecture-level cities from 2009 to 2022. The prefecture-level city panel data utilized in this study are primarily drawn from the 2009–2022 China Urban Statistical Yearbook, China Statistical Yearbook, and Regional Statistical Yearbook. The carbon emissions data of prefecture-level cities are chiefly sourced from the China Carbon Accounting Database (CEADS). The meteorological data of prefecture-level cities are obtained from the China Meteorological Data Network. The data concerning the size of high-carbon industries in prefecture-level cities are predominantly based on the 2009–2015 China Industrial Enterprise Database 2009–2015. The value variables are uniformly accounted for at constant prices, with 1978 as the base period. The descriptive statistics of the main variables are shown in

Table 1. Descriptive statistics.

Variable Symbol	Variable Name	Obs	Mean	SD	Min	Max
lnco2	Total carbon emissions	2095	3.3197	0.9051	0.2943	6.1263
lnco2_inten	Carbon intensity	2095	−2.3262	0.7509	−4.5557	−0.1643
open×pressure	Interaction term	2095	2.6081	4.1795	0.0000	25.0000
open	Policy implementation variables	2095	0.4463	0.4972	0.0000	1.0000
pressure	Pressure for environmental governance	2095	5.3599	4.4003	0.0000	25.0000
degree	Degree of openness to the outside world	2095	2.0107	1.8430	0.0000	19.7828
gov	Size of government	2095	16.9005	7.6264	4.3881	148.5164
gdp	GDP per capita	2095	0.8426	0.5515	0.1404	7.8640
gdps	GDP per capita squared	2095	1.0140	1.9519	0.0197	61.8420
human	Level of human capital	2095	2.0836	2.6168	0.0189	14.4580
lnpop	Total regional population	2095	5.9907	0.6238	3.9220	8.1345
lnrain	Average annual rainfall	2095	6.8630	0.5541	4.8354	7.9169
lntemp	Average annual temperature	2095	2.6666	0.4642	−1.4688	3.2464
ERS	Intensity of environmental regulation	2095	1.6040	1.4447	0.0230	12.6618
GI	Level of green innovation	2095	0.1965	0.4258	0.0000	3.9028
EGT	Energy green transition	2095	0.4398	0.1340	0.0121	0.7169
ET	End-of-end treatment	2095	−0.8235	1.0004	−11.2775	0
SC	Green transformation of industry	1418	6.2798	1.6672	−1.5934	10.6904

.

4. Empirical Analysis

4.1. Parallel Trend Test

The parallel trend of the control group and the treatment group is the premise for the DID model to effectively estimate the policy effect [32]. The reference year is defined as the year before the implementation of the policy. Dummy variables, designated as open_6, open_5, open_4, open_3, open_2, are constructed to characterize the initial six years, five years, four years, three years, and two years of the implementation of the new air quality standards, respectively. This analysis is conducted based on the time of implementation of the new air quality standards in different batches in prefecture-level cities. Concurrently, the dummy variables open1, open2, open3, open4, open5, and open6, respectively, are employed to delineate the one-year, two-year, three-year, four-year, five-year, and six-year periods following the implementation of the novel air quality standards. Concurrently, the interaction terms of the aforementioned virtual variables and environmental governance pressure (pressure) were generated to ascertain whether there were significant differences in the total amount and intensity of carbon emissions in prefecture-level cities under varying environmental governance pressures before the implementation of the novel ambient air quality standards.

The outcomes of the parallel trend test are presented in Figure 3A,B. The estimated coefficients of the dummy variables that characterize the period before the implementation of the new air quality standards did not achieve statistical significance at the 5% level. This finding suggests that the total amount and intensity of carbon dioxide emissions in regions experiencing varying environmental governance pressures did not exhibit significant differences before the implementation of the new air quality standards, thereby satisfying the assumption of parallel trend. Concurrently, the estimated coefficients of the dummy variables that characterize the post-implementation of the new air quality standards are predominantly negative, with the majority demonstrating statistical significance at the 5% level. This finding suggests that the pressure of environmental governance exerts a synergistic effect on carbon dioxide emissions reduction.

4.2. Analysis of Benchmark Regression Results

Table 2. Benchmark regression: environmental governance pressures and carbon reduction synergies.

Variable	Full Sample		Excluding Carbon Trading Pilot Cities
	(1)	(2)	(3)	(4)
	Carbon Emissions	Carbon Intensity	Carbon Emissions	Carbon Intensity
open×pressure	−0.0074 **	−0.0044	−0.0122 ***	−0.0100 **
	(0.0031)	(0.0033)	(0.0037)	(0.0039)
open	0.0635 ***	0.0994 ***	0.0980 ***	0.1330 ***
	(0.0227)	(0.0247)	(0.0276)	(0.0296)
degree	−0.0054	−0.0131 **	−0.0025	−0.0082
	(0.0058)	(0.0059)	(0.0054)	(0.0057)
gov	−0.0002	0.0039 *	0.0002	0.0027
	(0.0012)	(0.0022)	(0.0011)	(0.0017)
gdp	0.1351 *	−0.5172 ***	0.1176	−1.0630 ***
	(0.0703)	(0.0954)	(0.1327)	(0.1726)
gdps	−0.0157 **	0.0539 ***	−0.0043	0.2021 ***
	(0.0074)	(0.0127)	(0.0296)	(0.0499)
human	−0.0225	−0.0233	−0.0238	−0.0231 *
	(0.0143)	(0.0146)	(0.0146)	(0.0138)
lnpop	0.1153	−0.2190	0.1539	−0.3592
	(0.1494)	(0.1770)	(0.2012)	(0.2215)
lnrain	0.0042	0.0180	0.0269	0.0372
	(0.0283)	(0.0284)	(0.0310)	(0.0307)
lntemp	−0.0226	0.0696 **	−0.0259	0.0371
	(0.0260)	(0.0304)	(0.0306)	(0.0317)
constant	2.6156 ***	−0.9648	2.2620 *	0.2266
	(0.9066)	(1.0687)	(1.2364)	(1.3484)
Year effect	YES	YES	YES	YES
City effect	YES	YES	YES	YES
N	2095	2095	1658	1658
R2	0.9591	0.9358	0.9572	0.9331

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively; if not otherwise specified, the same applies below.

presents the estimation results for Model (3). The coefficients for open×pressure in Columns (1) and (2) are −0.0074 and −0.0044, respectively. These estimates indicate that, relative to the pre-implementation period, each unit increase in environmental governance pressure after the new standard’s implementation reduces urban CO₂ emissions by 0.74% and lowers CO₂ emission intensity by 0.44%. However, the estimated coefficient for open×pressure is small and fails to pass statistical significance testing. This is possibly because of the interference from carbon trading pilot policies: the gradually implemented carbon trading market pilot policies during the same period, as a hybrid regulation combining market flexibility with the rigidity of aggregate control, established stronger and more direct emissions reduction constraints for pilot regions through legally binding carbon emission caps and carbon price signals. The rigid constraints imposed by this policy, combined with the administrative rigidity of environmental governance pressure, created a cumulative effect in pilot regions.

This may have entirely dominated emissions reduction behaviors, masking the net effect of open×pressure as a standalone policy and leading to underestimation bias in the estimation results [33,34,35]. After excluding municipalities directly under the central government (Beijing, Shanghai, Tianjin, Chongqing) and prefectural-level cities in Hubei and Guangdong provinces from the estimations in Columns (3) and (4), the coefficients for open×pressure increase to −0.0122 and −0.01, respectively. These estimates are statistically significant at the 5% level, consistent with theoretical expectations. The empirical results confirm that environmental governance pressure significantly reduces both total regional carbon emissions and emission intensity, thereby supporting Hypothesis 1.

4.3. Robustness Test

4.3.1. Exclusion of Ex/Ante Policy Effects

A potential alternative explanation for our findings is that pre-existing policies, rather than the new standards, may have driven the observed reductions in CO₂ emissions and intensity. To address this concern, we conduct a falsification test using cities from 2009 to 2012 (pre-implementation period) as a placebo sample. We artificially assign 2010, 2011, and 2012 as implementation years for the first, second, and third batches of cities, respectively. As

Table 3. Time placebo test.

Variable	Full Sample		Excluding Carbon Trading Pilot Cities
	(1)	(2)	(3)	(4)
	Total Carbon Emissions	Carbon Intensity	Total Carbon Emissions	Carbon Intensity
open×pressure_P	−0.0020	−0.0008	−0.0036	−0.0005
	(0.0024)	(0.0024)	(0.0025)	(0.0025)
open_P	0.0032	−0.0002	0.0147	0.0018
	(0.0185)	(0.0190)	(0.0187)	(0.0188)
constant	1.9127 **	−2.2054 ***	0.0611	−1.8024
	(0.8806)	(0.7861)	(1.7279)	(1.6115)
Control variable	YES	YES	YES	YES
Year effect	YES	YES	YES	YES
City effect	YES	YES	YES	YES
N	898	898	722	722
R2	0.9842	0.9739	0.9845	0.9742

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② **, and *** indicate significance at the 5%, and 1% levels, respectively.

shows, the placebo test yields statistically insignificant estimates, confirming that the emission reductions are indeed caused by increased environmental governance pressure following the new standards’ implementation, not by prior policies.

4.3.2. Exclusion of Other City-Level-Specific Policies of the Same Period

Concurrent policy interventions during the sample period, including carbon trading pilots, low-carbon city initiatives, and innovative city pilots, may confound the estimated effects of open×pressure. Specifically, the National Development and Reform Commission launched low-carbon provincial pilots in six provinces and 81 cities across three batches (2010, 2012, 2017), while ministries implemented innovative city pilots in 59 cities during four phases (2008, 2010, 2016, 2018).

Table 4. Excluding other policy effects during the sample observation period.

	(1)	(2)	(3)	(4)	(5)	(6)
	Carbon Emissions	Carbon Intensity	Carbon Emission	Carbon Intensity	Carbon Emission	Carbon Intensity
open×pressure	−0.0109 **	−0.0096 *	−0.0103 **	−0.0093 *	−0.0105 **	−0.0094 *
	(0.0052)	(0.0049)	(0.0051)	(0.0048)	(0.0052)	(0.0049)
open	0.0772 **	0.1050 ***	0.0674 *	0.0986 ***	0.0682 *	0.0995 ***
	(0.0344)	(0.0336)	(0.0349)	(0.0343)	(0.0348)	(0.0340)
constant	2.9646 **	−0.4457	2.8654 **	−0.4707	2.8908 **	−0.4616
	(1.3046)	(1.3342)	(1.3433)	(1.3621)	(1.3277)	(1.3466)
Innovation city	NO	NO	YES	YES	YES	YES
Low-carbon cities	NO	NO	NO	NO	YES	YES
Control variable	YES	YES	YES	YES	YES	YES
Year effect	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES
Year × province	YES	YES	YES	YES	YES	YES
N	1658	1658	1658	1658	1658	1658
R2	0.9654	0.9477	0.9656	0.9479	0.9656	0.9479

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

presents regression results controlling for these concurrent policies. The open×pressure coefficients remain significantly negative at the 10% level, consistent with benchmark estimates. This confirms that local governments’ environmental governance pressure independently reduces carbon emissions, unaffected by contemporaneous policy interventions.

4.3.3. Potential Selectivity in Policy Implementation

The causal identification in our continuous difference-in-differences framework relies on the assumption of random assignment in policy implementation. However, the Ambient Air Quality Standards (2012) explicitly mandated phased implementation: provincial capitals, direct-controlled municipalities, separately planned cities, and major urban agglomerations (Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei) were required to adopt the standards by 2013, while key and model environmental protection cities followed by 2014. To address potential selection bias, we employ the methodological approach of Zhang et al. [18], constructing city-type fixed effects based on the administrative hierarchy (provincial capitals, direct-controlled municipalities) and regional characteristics (major urban agglomerations, environmental policy cities). We systematically control for these factors through interaction terms with linear time trends in Model (3). As

Table 5. Robustness tests: further treatment of potential selectivity in policy implementation.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	Carbon Emissions	Carbon Intensity	Carbon Emissions	Carbon Intensity	Carbon Emissions	Carbon Intensity
open×pressure	−0.0072 **	−0.0041	−0.0098 ***	−0.0078 **	−0.0095 **	−0.0078 **
	(0.0031)	(0.0033)	(0.0036)	(0.0037)	(0.0037)	(0.0038)
open	0.0591 **	0.0821 ***	0.0707 ***	0.0768 ***	0.0620 *	0.0750 *
	(0.0230)	(0.0244)	(0.0259)	(0.0268)	(0.0369)	(0.0417)
constant	2.4762 ***	−1.0370	2.7873 ***	−0.5629	2.6500 **	−0.3082
	(0.9002)	(1.0885)	(0.9873)	(1.1825)	(1.0456)	(1.1911)
Provincial city	YES	YES	YES	YES	YES	YES
Municipalities	YES	YES	YES	YES	YES	YES
PRD Cities	NO	NO	YES	YES	YES	YES
Yangtze River Delta	NO	NO	YES	YES	YES	YES
Beijing–Tianjin–Hebei cities	NO	NO	YES	YES	YES	YES
Planned cities	NO	NO	NO	NO	YES	YES
Environmentally friendly city	NO	NO	NO	NO	YES	YES
Control variable	YES	YES	YES	YES	YES	YES
Year effect	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES
N	2095	2095	2095	2095	2095	2095
R2	0.9594	0.9362	0.9603	0.9378	0.9606	0.9385

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

demonstrates, the environmental governance pressure (open×pressure) coefficients maintain their significantly negative signs (p < 0.05) across all specifications. This robustness check confirms that our benchmark results are not substantially confounded by non-random policy assignment.

4.3.4. Replacement of Core Explanatory Variables

In the baseline regression, environmental governance pressure (pressure) is measured by the frequency of PM_2.5 and PM₁₀ keywords in government work reports. To test the robustness of the results, we adopt two alternative measures of environmental governance pressure. First, following Li Wenjing, we construct the urban environmental governance ranking pressure (rank). Specifically, cities within each province are ranked based on the annual number of days with AQI > 100, and we compare the relative ranking changes between 2015 and 2022. If a city’s ranking deteriorated in 2022 compared with 2015, the rank is coded as 1; otherwise, it is coded as 0. Thus, rank is a dummy variable reflecting changes in relative pollution rankings and capturing the urban environmental governance ranking pressure within provinces. Second, to address potential concerns about variations in report length, we apply Z-score standardization to the keyword frequencies, generating a new indicator (pressure1) to ensure comparability across cities and years. These two alternative indicators are then incorporated into the regression. Columns (1) and (2) in

Table 6. Robustness tests: replacing core explanatory variables.

	(1)	(2)	(3)	(4)
	Total Carbon Emissions	Carbon Intensity	Total Carbon Emissions	Carbon Intensity
open×rank	−0.0515 ***	−0.0509 **
	(0.0196)	(0.0201)
open×pressure1			−0.0324 *** (0.0075)	−0.0194 ** (0.0082)
open	0.0464 **	0.0855 ***	0.0240	0.0758 ***
	(0.0231)	(0.0234)	(0.0165)	(0.0178)
constant	2.7832 ***	−1.6921 *	2.6156 ***	−0.9648
	(0.8207)	(0.9363)	(0.7157)	(0.8386)
Control variable	YES	YES	YES	YES
Year effect	YES	YES	YES	YES
City effect	YES	YES	YES	YES
N	2018	2018	2093	2093
R2	0.9645	0.9497	0.9591	0.9358

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

present the results based on ranking pressure, while Columns (3) and (4) present those based on standardized keyword frequency. Regardless of the replacement method, the estimated coefficients of the core explanatory variables open×rank and open×pressure1 remained negative and statistically significant at least at the 5% level, indicating that the baseline regression results are not substantially affected by measurement bias.

5. Further Analysis

5.1. Synergy Analysis

The empirical results demonstrate that environmental governance pressure significantly reduces both the total volume and intensity of carbon emissions in prefectural-level cities. To elucidate the underlying mechanisms of this synergistic effect, we employ an interaction term model for further analysis. Given that sulfur dioxide (SO₂) and nitrogen oxides (NO_x) represent major air pollutants, we examine their synergistic emission reduction effects through first-difference transformations. Specifically, we calculate urban emission reductions for SO₂ and NO_x by taking the negative values of their first differences and then construct interaction terms between these measures and environmental governance pressure (open×pressure). The negative and statistically significant coefficients (p < 0.10) for the interaction terms in

Table 7. Interaction term test for synergies: sulfur dioxide and ammonia nitride reductions.

Variable	(1)	(2)	(3)	(4)
	Total Carbon Emissions	Carbon Intensity	Total Carbon Emissions	Carbon Intensity
open×pressure×SO2	−0.0009 **	−0.0007 *
	(0.0004)	(0.0004)
SO2	0.0002	0.0007
	(0.0009)	(0.0009)
open×pressure×NOx			−0.0020 **	−0.0016 *
			(0.0009)	(0.0010)
NOx			0.0001	0.0009
			(0.0013)	(0.0014)
open×pressure	−0.0071 ***	−0.0047 **	−0.0049 *	−0.0029
	(0.0021)	(0.0022)	(0.0026)	(0.0027)
open	0.0475 **	0.0739 ***	0.0335	0.0626 ***
	(0.0214)	(0.0225)	(0.0233)	(0.0242)
constant	0.4399	−1.4950	0.4197	−1.5221 *
	(0.8734)	(0.9143)	(0.8768)	(0.9195)
Control variable	YES	YES	YES	YES
Year effect	YES	YES	YES	YES
City effect	YES	YES	YES	YES
N	1240	1240	1229	1229
R2	0.9721	0.9553	0.9718	0.9550

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

demonstrate that greater reductions in regional air pollutants are associated with stronger carbon emission reduction synergies. These results indicate that environmental governance pressure reduces both total carbon emissions and emission intensity, principally through this synergistic mechanism, thereby confirming Hypothesis 2.

To further verify the synergistic emission reduction effects, we conduct subgroup analyses by stratifying the sample based on mean values of CO₂ and NO_x emission reductions. The baseline model (3) is estimated separately for high-reduction and low-reduction subgroups.

Table 8. Synergy group regression test: sulfur dioxide and ammonia nitride reductions.

Variable	Sulfur Dioxide Emission Reductions				Ammonia Nitride Emission Reductions
	High		Low		High		Low
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	Emissions	Intensity	Emissions	Intensity	Emissions	Intensity	Emissions	Intensity
open×pressure	−0.0102 **	−0.0091 **	−0.0027	0.0012	−0.0103 **	−0.0092 **	−0.0029	0.0005
	(0.0041)	(0.0042)	(0.0032)	(0.0031)	(0.0041)	(0.0041)	(0.0033)	(0.0031)
open	0.0556	0.1333 ***	0.0159	0.0130	0.0553	0.1380 ***	0.0150	0.0155
	(0.0368)	(0.0428)	(0.0254)	(0.0269)	(0.0377)	(0.0441)	(0.0254)	(0.0259)
constant	2.1247 **	−1.0753	−0.0797	−3.8235 *	2.0800 **	−1.1346	−0.0988	−3.8453 *
	(0.8580)	(1.2341)	(1.4183)	(1.4275)	(0.8623)	(1.2286)	(1.4242)	(1.4281)
Control	YES	YES	YES	YES	YES	YES	YES	YES
year effect	YES	YES	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES	YES	YES
N	1080	1080	992	992	1080	1080	993	993
R2	0.9515	0.9293	0.9744	0.9603	0.9529	0.9306	0.9730	0.9598

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

presents the subgroup estimation results. In regions with above-average SO₂ and NO_x emission reductions, the environmental governance pressure (open×pressure) coefficients are negative and statistically significant (p < 0.05). By contrast, the coefficients are insignificant in below-average reduction regions. These results demonstrate that the synergistic carbon reduction effect is particularly pronounced in high-emission-reduction areas, corroborating both the interaction term model findings and Hypothesis 2.

5.2. Mechanism Analysis

As theoretically established, environmental governance pressure facilitates carbon emission reduction synergies through dual mechanisms. On one hand, it mitigates intergovernmental “race to the bottom” competition in environmental regulation, thereby suppressing the green paradox effect while stimulating regional green innovation and energy structure transformation. On the other hand, it shifts local governments’ governance preferences from end-of-pipe solutions to source control strategies, particularly through industrial structure decarbonization. Based on the preceding definitions of Environmental Regulation (ERS), Green Energy Transition (EGT), Green Innovation (GI), End-of-Pipe Treatment (ET), and Source Control (SC), we analyze the synergistic carbon reduction effect of government environmental governance pressure using the three-step mediation effect method. The mechanism analysis of the green paradox and green technology innovation is shown in

Table 9. Mechanism analysis: green paradox and green technology innovation.

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	ERS	EGT	Emissions	Intensity	GI	Emissions	Intensity
open×pressure	0.0218 **	−0.0030 ***	−0.0084 ***	−0.0068 *	0.0030 **	−0.0097 ***	−0.0092 **
	(0.0104)	(0.0004)	(0.0035)	(0.0037)	(0.0012)	(0.0037)	(0.0039)
open	−0.1959 **	0.0285 ***	0.0757 ***	0.1026 ***	−0.0605 ***	0.0798 ***	0.1161 ***
	(0.0883)	(0.0044)	(0.0272)	(0.0287)	(0.0133)	(0.0288)	(0.0318)
EGT			0.7822 ***	1.0691 ***
			(0.2738)	(0.2870)
GI						−0.0735 **	−0.0627 **
						(0.0308)	(0.0252)
constant	−0.5511	0.5720 ***	1.8146	−0.3849	−1.8044 ***	2.7341 *	0.7018
	(2.5682)	(0.1219)	(1.2965)	(1.3919)	(0.5623)	(1.4468)	(1.6174)
Control variable	YES	YES	YES	YES	YES	YES	YES
Year effect	YES	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES	YES
N	2095	2095	2095	2095	2095	2095	2095
R2	0.8224	0.9239	0.9581	0.9353	0.8813	0.9588	0.9370

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

.

From the perspective of the mechanism through which the green paradox arises, under the incentive mechanism of ‘Chinese-style’ decentralization, horizontal competition among local governments triggers a ‘race to the bottom’ effect in environmental regulation competition, thereby leading to an increase in fossil energy consumption, which is the mechanism behind the green paradox [5]. In Column (1), the dependent variable is regional environmental regulation. The estimated coefficient of open×pressure is significantly positive at the 5% level, indicating that stronger environmental governance pressure leads local governments to adopt stricter environmental regulations. Therefore, environmental governance pressure can break the “race to the bottom” effect in local government environmental regulation competition, thus not leading to a green paradox. In Column (2), the dependent variable is the proportion of regional fossil energy consumption. The estimated coefficient of open×pressure is significantly negative at the 1% level, indicating that stronger environmental governance pressure leads to a lower proportion of fossil energy consumption, further demonstrating that environmental governance pressure does not lead to a green paradox. The estimation results in Column (1) and Column (2) collectively indicate that environmental governance pressure does not lead to a green paradox.

Columns (2), (3), and (4) of Table 9 further analyze the mechanism of the green energy transition. In Column (2), the estimated coefficient of open×pressure indicates that environmental governance pressure significantly promotes the green transformation of the energy consumption structure. In Columns (3) and (4), the estimated coefficient of EGT is significantly positive, meaning that greater fossil energy consumption leads to higher regional carbon dioxide emissions and higher carbon emission intensity, which is consistent with expectations. The estimated coefficients of open×pressure are −0.0084 and −0.0068, respectively, and their absolute values are smaller than the estimated coefficients in the baseline regression results (−0.0122, −0.0100). This passes the three-step mediation effect test, indicating that environmental governance pressure can reduce total carbon emissions and carbon emission intensity by promoting the green transformation of the regional energy consumption structure.

Columns (5), (6), and (7) of Table 9 further analyze the mechanism of green technology innovation. In Column (5), the estimated coefficient of open×pressure indicates that environmental governance pressure is conducive to promoting regional green technology innovation, and environmental governance pressure can stimulate the Porter effect. In Columns (6) and (7), the estimated coefficient of GI is significantly negative, meaning that green technology innovation can reduce carbon dioxide emissions and carbon emission intensity, which is consistent with expectations. The estimated coefficients of open×pressure are −0.0097 and −0.0092, respectively, and their absolute values are smaller than the estimated coefficients in the baseline regression results (−0.0122, −0.0100). This passes the three-step mediation effect test, indicating that environmental governance pressure can reduce total carbon emissions and carbon emission intensity by promoting regional green technology innovation.

Columns (1), (2), and (3) of

Table 10. Mechanism analysis: “End-of-End Treatment” or “Source Control”.

Variable	(1)	(2)	(3)	(4)	(5)	(6)
	ET	Emissions	Intensity	SC	Emissions	Intensity
open×pressure	−0.0197 **	−0.0109 ***	−0.0079 *	−0.0133 *	−0.0082 *	−0.0061 **
	(0.0084)	(0.0032)	(0.0042)	(0.0074)	(0.0037)	(0.0034)
open	−0.0191	0.0624 **	0.0818 ***	0.0734	0.0611 *	0.0755 **
	(0.0836)	(0.0275)	(0.0279)	(0.0980)	(0.0336)	(0.0323)
ET		0.0121 ***	0.0141 *
		(0.0043)	(0.0078)
SC					0.0183 *	0.0115 **
					(0.0099)	(0.0049)
constant	−0.8280	−0.2548	−2.4798 *	11.0007 ***	−0.3633	−2.6268 *
	(2.4839)	(1.4331)	(1.4045)	(2.5410)	(1.5483)	(1.4679)
Control variable	YES	YES	YES	YES	YES	YES
Year effect	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES
N	2095	2095	2095	2095	2095	2095
R2	0.6496	0.9744	0.9595	0.8760	0.9754	0.9616

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

present the empirical results analyzing the mechanism of end-of-pipe treatment. In Column (1), the estimated coefficient of open×pressure is significantly negative at the 5% level, indicating that environmental governance pressure helps to reduce local governments’ preference for end-of-pipe treatment. In the estimation results of Columns (2) and (3), the estimated coefficient of ET is significantly positive at the 10% level, suggesting that the end-of-pipe treatment of pollutants, conversely, increases total carbon dioxide emissions and emission intensity. The estimated coefficients of open×pressure are −0.0109 and −0.0079, respectively, and their absolute values are smaller than those in the baseline regression results (−0.0122, −0.0100). The above empirical estimation results indicate that environmental governance pressure can reduce total regional carbon emissions and carbon emission intensity by weakening local governments’ preference for end-of-pipe treatment.

Columns (4), (5), and (6) of Table 10 present the empirical results analyzing the mechanism of source control (or industrial structure adjustment). In Column (4), the estimated coefficient of open×pressure is significantly negative at the 10% level, indicating that environmental governance pressure can reduce the proportion of high-carbon industries in the region and promote green industrial transformation. In the estimation results of Columns (5) and (6), the estimated coefficient for the proportion of high-carbon industries (if ‘ET’ is used to represent this in the table) is significantly positive at the 10% level, indicating that a larger proportion of regional high-carbon industries leads to higher total carbon dioxide emissions and emission intensity, which is consistent with theory and expectations. The estimated coefficients of open×pressure are −0.0082 and −0.0061, respectively, and their absolute values are smaller than those in the baseline regression results (−0.0122, −0.0100). The above empirical estimation results indicate that environmental governance pressure can reduce total regional carbon emissions and carbon emission intensity by strengthening source control.

The above empirical results indicate that environmental governance pressure can influence regional total carbon dioxide emissions and emission intensity through channels such as promoting the green transformation of the energy consumption structure, stimulating the Porter effect, weakening the preference for end-of-pipe treatment, and strengthening source control. The relative strengths of these mechanisms are further analyzed below.

The mediation effects were tested using the Bootstrap method proposed by Preacher et al. [36], and the results are shown in

Table 11. Bootstrap-based mediation effect testing.

Path of Causality and Consequence	Indirect Effects	95% Confidence Interval
		Lower Limit of the Confidence Interval	Upper Limit of the Confidence Interval
open×pressure→EGT→CE	−0.0034	−0.0055	−0.0013
open×pressure→GI→CE	−0.0026	−0.0040	−0.0012
open×pressure→ET→CE	−0.0015	−0.0022	−0.0008
open×pressure→SC→CE	−0.0037	−0.0065	−0.0009
open×pressure→EGT→CI	−0.0028	−0.0046	−0.0010
open×pressure→GI→CI	−0.0007	−0.0013	−0.0001
open×pressure→ET→CI	−0.0023	−0.0027	−0.0019
open×pressure→SC→CI	−0.0036	−0.0057	−0.0015

. The mediation effects of environmental governance pressure on regional total carbon emissions through channels such as EGT, GI, ET, and SC are, respectively, −0.0034, −0.0026, −0.0015, and −0.0037, all with 95% confidence intervals not including zero. The relative strengths of the four mediation effects are: SC > EGT > GI > ET. The mediation effects of environmental governance pressure on regional carbon emission intensity through channels such as EGT, GI, ET, and SC are, respectively, −0.0028, −0.0007, −0.0023, and −0.0036, all with 95% confidence intervals not including zero. The relative strengths of the four mediation effects are: SC > EGT > ET > GI. Green energy transition and source control are the main mechanisms for generating synergistic carbon reduction effects, while the mediation effects of end-of-pipe treatment and green innovation are relatively weak.

Source control consistently shows the strongest mediation effect for both total emissions and emission intensity. Economically, this signifies that environmental governance pressure most effectively drives carbon reduction by compelling fundamental industrial restructuring, promoting cleaner production technologies, and enhancing resource efficiency at the source of pollution. This aligns with China’s long-term strategy of shifting from extensive growth to high-quality, green development, which prioritizes “remedying the root cause” over “treating the symptom.” Local governments, under pressure, are encouraged to eliminate outdated, high-carbon production capacity and foster new, green industries, directly leading to more profound and sustainable emission reductions.

The substantial mediation effect of green energy transition underscores its critical role in China’s decarbonization efforts. Environmental governance pressure directly stimulates investments in renewable energy sources (e.g., solar, wind), accelerates the phasing out of fossil fuels, and promotes energy efficiency improvements across sectors. From an economic perspective, this transition involves massive infrastructure development, technological advancements, and the creation of new green energy industries. While requiring substantial upfront capital, it offers long-term economic benefits such as enhanced energy security, reduced reliance on imported fossil fuels, and the potential for new economic growth engines. The strong effect indicates that explicit pressure from environmental governance can effectively overcome initial economic barriers and drive large-scale capital deployment towards a low-carbon energy system.

While crucial for long-term sustainability, green innovation exhibits a relatively weaker immediate mediation effect compared to SC and EGT, especially for carbon emission intensity. This is largely due to the inherent economic characteristics of innovation, involving high R&D costs, commercialization risks, and a significant time lag for widespread adoption to yield substantial emission reductions. Similarly, End-of-Pipe Treatment consistently shows the weakest mediation effect. This suggests that sustained environmental governance pressure decisively discourages reliance on costly and often less efficient “end-of-pipe” solutions. Instead, it propels local governments towards more economically rational and environmentally effective preventative, source-oriented strategies, reflecting a maturation in governance approaches that prioritize fundamental changes over symptomatic treatments.

5.3. Analysis of the Spatial Spillover Effect

Environmental governance pressure can also, through spillover effects, influence the energy structure, level of green innovation, preference for end-of-pipe treatment, and industrial structure of geographically adjacent regions [32]. The experiences, technologies, and policy models developed by cities with higher environmental governance pressure in promoting cleaner energy structures and greener industrial structures may be learned from and adopted by adjacent cities, fostering regional demonstrations of energy transition. Concurrently, cities’ investments and achievements in green technology R&D will diffuse to neighboring cities through channels such as talent flow, technological cooperation, and literature sharing, thereby enhancing the overall green innovation capacity of the region. Therefore, this study subsequently constructs a spatial geographical matrix and, based on spatial econometric models, further analyzes the spatial spillover effects of environmental governance pressure.

A prerequisite for using a spatial difference-in-differences (DID) model is that the dependent variable exhibits spatial correlation. Thus, this study first employs a geographical-distance-based spatial weight matrix to calculate Moran’s I index for both total carbon emissions and the carbon emissions intensity of prefecture-level cities, to determine their level of spatial correlation. The results are presented in Figure 4. Moran’s I indices calculated under the geographical-distance-based spatial weight matrix are all significantly positive throughout the study period. Therefore, the total carbon emissions and carbon emission intensity of prefecture-level cities exhibit significant positive spatial dependence, justifying the use of spatial econometric models for the subsequent analysis.

Subsequently, LM, LR, and Wald tests were performed on the Spatial Error Difference-in-Differences (SEM-DID) model and the Spatial Lag Difference-in-Differences (SLM-DID) model. As shown in

Table 12. Results of spatial econometric model selection.

Spatial Econometric Model	LM Test		LR Test	Wald Test
	LM	Robust LM
SLM-DID	4.5862 ***	0.2745 ***	18.8562 ***	20.8746 ***
SEM-DID	9.8521 ***	7.8651 ***	25.2478 ***	26.5421 ***

Note: *** indicates significance at the 1% level.

, the results of all three tests indicate that the spatial Durbin difference-in-differences (SDM-DID) model is optimal. Therefore, we select the SDM-DID for regression analysis.

Table 13. An analysis of the spatial spillover effect.

Variable	Carbon Emissions			Carbon Intensity
	SEM-DID	Effect Decomposition		SEM-DID	Effect Decomposition
		Direct Effects	Indirect Effects		Direct Effects	Indirect Effects
open×pressure	−0.0063 *** (0.0015)	−0.0075 *** (0.0013)	−0.0042 *** (0.0007)	−0.0045** (0.0022)	−0.0054 ** (0.0027)	−0.0036 ** (0.0018)
W×open×pressure	−0.0042 *** (0.0008)			−0.0038 *** (0.0007)
$\rho$	−0.0025 ** (0.0012)			−0.0036 *** (0.0017)
Control variable	YES	YES	YES	YES	YES	YES
Year effect	YES	YES	YES	YES	YES	YES
City effect	YES	YES	YES	YES	YES	YES
R2	0.7524			0.7985

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② **, and *** indicate significance at the 5%, and 1% levels, respectively.

reports the results of the spillover effect analysis. Under the geographical spatial weight matrix, the coefficients of open×pressure are both significantly negative, indicating that environmental governance pressure is beneficial in reducing total regional carbon emissions and carbon emissions intensity. This is consistent with the benchmark regression results. From the perspective of spillover effects, the estimated coefficient for W×open×pressure is significantly negative, suggesting that environmental governance pressure significantly reduces carbon emissions and intensity in geographically adjacent cities, implying a negative spatial spillover effect.

5.4. Heterogeneity Analysis: Emission Reduction Costs

Regional disparities in the industrial foundation and initial CO₂ emission levels lead to varying costs of CO₂ emission reduction across regions. Facing these different carbon reduction costs, local governments’ differentiated decision-making also contributes to the heterogeneity in the synergistic effect of carbon emissions reduction.

When carbon reduction costs are low, local governments under environmental governance pressure are more inclined to promote the green transformation of the energy structure, incentivize green technological innovation in enterprises, and abandon end-of-pipe treatment in favor of source control, among other such measures. These approaches not only benefit environmental protection but also have a relatively minor impact on economic development, allowing principal officials to maximize their promotion prospects. Consequently, the synergistic effect of environmental governance pressure on carbon emissions reduction is stronger.

Conversely, when carbon reduction costs are high, local governments facing environmental governance pressure find that adjusting industrial and energy structures or abandoning end-of-pipe treatment would hurt rapid regional economic development. This situation is detrimental to principal officials maximizing their promotion prospects, leading to a weaker synergistic effect of environmental governance pressure on carbon emission reduction.

When the initial regional CO₂ emission level is higher, and the regional industrial development level is higher, the difficulty of pollutant reduction in the region increases, leading to higher abatement costs and, consequently, a weaker synergistic effect of carbon emissions reduction. We measured the initial regional CO₂ emission level (IPL) using the average CO₂ emission levels of cities from 1997 to 2008. The regional industrial development level (S) was measured according to the average proportion of the secondary industry in cities during the sample observation period. Based on the lower quartile, the full sample was divided into sub-samples with lower and higher abatement costs, and the empirical results were re-estimated for these different sub-samples.

Columns (1), (2), (3), and (4) of

Table 14. Heterogeneity analysis: emission reduction costs.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
	IPL < 7.54	IPL < 7.54	IPL > 7.54	IPL > 7.54	S < 43.204	S < 43.204	S > 43.204	S > 43.204
	Emissions	Intensity	Emissions	Intensity	Emissions	Intensity	Emissions	Intensity
open×pressure	−0.0253 *	−0.0219 *	−0.0097 **	−0.0094 **	−0.0152 *	−0.0129 *	−0.0122 ***	−0.0104 **
	(0.0128)	(0.0124)	(0.0042)	(0.0044)	(0.0076)	(0.0074)	(0.0044)	(0.0047)
open	0.1752 **	0.1977 ***	0.0808 **	0.1302 ***	0.0764	0.1205 **	0.1023 ***	0.1331 ***
	(0.0684)	(0.0673)	(0.0323)	(0.0351)	(0.0540)	(0.0519)	(0.0337)	(0.0368)
constant	2.2976	0.0307	1.3572	−0.6794	1.7183	−2.0134	2.7692	0.9011
	(1.9743)	(2.0211)	(1.3113)	(1.2653)	(1.1026)	(1.2696)	(1.7401)	(1.7767)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Year effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
City effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	403	403	1255	1255	404	404	1254	1254
R2	0.9328	0.9240	0.9461	0.9428	0.9714	0.9545	0.9506	0.9274

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

report the heterogeneity analysis results based on the initial CO₂ emission level (IPL). Compared to the sub-sample with higher initial CO₂ emissions, the absolute value of the estimated coefficients for the sub-sample with lower initial CO₂ emissions is larger. This indicates that the lower the abatement cost, the stronger the synergistic effect of environmental governance pressure on carbon emissions reduction.

Columns (5), (6), (7), and (8) of Table 14 present the heterogeneity analysis results based on the industrial development level (S). Similarly, compared to the sub-sample with a higher industrial development level, the absolute value of the estimated coefficients for the sub-sample with a lower industrial development level is larger. This further confirms the conclusion that the lower the abatement cost, the stronger the synergistic effect of environmental governance pressure on carbon emissions reduction.

Furthermore, China’s Central and Eastern regions, characterized by high levels of industrial development and higher initial pollutant emission levels, face higher abatement costs, resulting in a weaker synergistic effect of environmental governance pressure on carbon emissions reduction. Conversely, Western regions, with lower industrial development levels and lower initial pollutant emission levels, experience lower abatement costs, leading to a stronger synergistic effect of environmental governance pressure on carbon emission reduction.

Table 15. Heterogeneity analysis: Eastern region, Central region, and Western region.

	Eastern China		Central China		Western China
	(1)	(2)	(3)	(4)	(5)	(6)
	Emissions	Intensity	Emissions	Intensity	Emissions	Intensity
open×pressure	−0.0064	−0.0061	−0.0088	−0.0069	−0.0364 ***	−0.0348 ***
	(0.0067)	(0.0064)	(0.0062)	(0.0061)	(0.0073)	(0.0082)
open	0.0771	0.1092	0.0905 *	0.1002 **	0.1716 ***	0.2293 ***
	(0.0691)	(0.0658)	(0.0483)	(0.0497)	(0.0493)	(0.0526)
constant	1.8587	−1.8748	3.6599 ***	2.2285 *	−6.2758	−10.4046 **
	(1.7362)	(1.6883)	(1.3743)	(1.2004)	(3.9215)	(4.0175)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes
Year effect	Yes	Yes	Yes	Yes	Yes	Yes
City effect	Yes	Yes	Yes	Yes	Yes	Yes
N	544	544	700	700	414	414
R2	0.9550	0.9315	0.9489	0.9413	0.9615	0.9351

Note: ① Standard errors clustered to the prefecture level are reported in parentheses; ② *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

further reports the heterogeneity analysis results based on these different regional divisions. Columns (1) and (2) correspond to the Eastern region sub-sample, while Columns (3) and (4) correspond to the Central region sub-sample. For these regions, the estimated coefficients of open×pressure are negative but not statistically significant, which suggests that, due to higher abatement costs, the synergistic effect of environmental governance pressure on carbon emissions reduction is not significant. In contrast, Columns (5) and (6) correspond to the Western region sub-sample, where the estimated coefficient of open×pressure is significantly negative at the 1% level. This indicates that the lower the abatement cost, the more significant the synergistic effect of environmental governance pressure on carbon emission reduction.

6. Conclusions and Policy Recommendations

In the context of carbon emissions reduction, the synergistic effect of environmental governance pressure is crucial to the construction of ecological civilization in the new era. This study reveals the carbon emissions reduction synergistic effect of the local government environmental governance pressure mechanism using the successive double-difference method based on the implementation of Ambient Air Quality Standards (2012) as an exogenous shock, in addition to using the word frequency of local government work reports to measure the environmental governance pressure of the local government. We found that local government environmental governance pressure has a significant carbon emissions reduction synergistic effect, and environmental governance pressure not only synergistically suppresses the total carbon emissions of prefecture-level cities but also synergistically reduces the regional carbon emission intensity. The above conclusions remain robust after a series of robustness tests. The synergy test showed that the larger the emission reduction in pollutants such as sulfur dioxide and ammonia nitrogen compounds, the more favorable the local government environmental governance pressure is to synergistically inhibit the total amount and intensity of urban carbon emissions, indicating that local government environmental governance pressure inhibits the total amount and intensity of regional carbon emissions, mainly through synergistic effects. The mechanism test shows that the pressure of local government environmental governance constrains the environmental regulation “competition at the bottom” behavior of regional local governments; the increase in environmental regulation intensity does not expand the green paradox and intensify the end-of-pipe governance but rather suppresses the regional carbon emission total and intensity through the channels of accelerating the green transformation of the energy structure, promoting the low-carbon transformation of the industrial structure, and promoting the regional green innovation. Thus, the total amount and intensity of regional carbon emissions are suppressed through accelerating the green transformation of energy structure, promoting the low-carbon transformation of industrial structure, and promoting regional green innovation. Moreover, the heterogeneity analysis indicates that the synergistic effect of environmental governance pressure on carbon emission reductions is more pronounced in Western regions with lower abatement costs, cities with lower initial emission levels, and areas with weaker industrial foundations. The findings of this study have clear policy implications:

①.

Strengthen the pressure mechanism of local government environmental governance through both top-down and bottom-up approaches. Central and higher-level governments should increase top-down pressure on local governments by introducing real-time data-driven environmental performance evaluations and strengthening accountability mechanisms and formal interviews for leading officials in non-compliant regions. Concurrently, they should improve environmental information disclosure platforms, broaden public oversight channels, and encourage environmental NGOs to file public interest lawsuits in accordance with the law. This will generate effective bottom-up public pressure, thereby enhancing the “two-way” environmental governance pressure faced by local governments.

②.

Adjust and optimize local government performance evaluation metrics to enhance synergistic effects in carbon reduction. Incorporate pollutant reduction targets for sulfur dioxide, ammonia nitrogen compounds, and other pollutants, alongside carbon dioxide emission totals and intensity metrics, into the indicator system. Assign differentiated weightings to these metrics to incentivize local governments to pursue coordinated advancement in pollution control and carbon reduction.

③.

Guiding enterprises to carry out more green technology innovation activities by policy means, and strengthening the mechanism for generating synergistic carbon emission reduction effects. Local governments not only need to promote the green technological innovation of enterprises through environmental governance policies, but must also provide compensation for the green technological innovation activities of enterprises through financial and tax policies, guiding enterprises to carry out more green innovation activities.

④.

The mechanism of blocking the environmental governance behavior of local government through the pressure environmental management system leads to the green paradox effect. On the one hand, the central government should establish scientifically sound and reasonably differentiated regional environmental standards to avoid the transfer of high-carbon industries caused by one-size-fits-all policies. At the same time, it should promote market-based tools such as energy rights and carbon emission rights trading to encourage industries to shift from passive emissions reduction to proactive transformation.

⑤.

Local governments need to shift their environmental management philosophy, promoting a transition from end-of-pipe pollution control to source-based prevention. They should be encouraged to establish timetables for phasing out high-carbon industries and formulating plans for clean energy substitution. By establishing green transition funds for industrial restructuring and providing expedited approval processes for low-carbon projects, local governments can reduce the institutional costs of source-based pollution control and enhance their long-term synergistic effects in reducing pollution and carbon emissions. To provide a clearer understanding of the logical connections among the policy implications, Figure 5 illustrates the proposed framework of policy recommendations derived from our findings.