The role of green fiscal stimulus policy in enhancing the synergistic performance of air pollution control and carbon mitigation: evidence from 281 cities in China

Abstract

Against escalating climate change and air pollution, economies need to explore viable policies for synergistic performance of air pollution control and carbon mitigation (SPPC). This under budget constraints and to achieve sustainable development. Existing research focuses mainly on market-based or command-based policies for SPPC, but the impact of green fiscal stimulus policies (GFSP) as incentive-based policies remains underexplored. This study used the energy saving and emission reduction fiscal pilot (ESER), a GFSP case, as a case study. Using non-radial directional distance function data envelopment analysis, it systematically measured urban SPPC efficiency based on data collected from 281 Chinese cities between 2007 and 2022. A staggered difference-in-difference method examines how GFSP influences SPPC. The results showed that GFSP significantly improved SPPC. This is because pilot cities enforced stricter environmental standards, signaling to enterprises the government’s commitment to environmental governance. When policy stability was assured, enterprises faced less climate policy uncertainty, which prompted greater investment in long-term clean technological innovation. As enterprises innovated, SPPC was realized. Meanwhile, affected by differences in incentive funding allocation, ESER had a more significant effect on SPPC in eastern, non-resource-based and non-industrial-based cities, as well as in cities with intense inter-jurisdictional competition. Moreover, the GFSP’s positive effects also spilled over to the non-targeted emission, methane; however, it did not have a significant impact on non-targeted pollutants. This study provides viable policy options for the realization of SPPC in other emerging economies with limited public budgets.

Introduction

Against the backdrop of escalating global warming and air pollution, environmental governance has become a major challenge for governments worldwide. According to the World Health Organization, air pollution causes 7 million global deaths annually, threatening human survival and development (Cai et al., 2018). Climate change has triggered frequent extreme weather and climate disasters, exacerbating the spread of over 200 infectious diseases and dozens of non-communicable diseases, endangering human existence (Mora et al., 2022). Countries have made efforts in carbon mitigation and air pollution control, such as the European Union’s carbon emission trading market, the United Kingdom’s clean air strategy, and Canada’s carbon tax policy (Tang, 2019; Bayer and Aklin, 2020). However, existing policies mainly target emissions or pollutants, and limited budgets must balance economic development with environmental governance. This has spurred a global search for policies to achieve synergistic management of emissions and pollutants under budget constraints, integrating economic and environmental goals. As a result, synergistic performance of air pollution control and carbon mitigation (SPPC) have gained increasing attention.

To date, studies on SPCC has primarily explored the roles of market- and command-based policies. For instance, command-based policies (e.g., emission standards and environmental regulations) can force localities to reduce carbon and pollutant emissions (Zhou et al., 2023). Another body of literature suggests that market policies enhance optimal resource allocation and environmental governance efficiency (Calel and Dechezleprêtre, 2016). Carbon trading market, for example, limit corporate emissions by allowing firms to trade carbon credits (Liu et al., 2015). While these policies target carbon emissions, industrial processes often generate other pollutants like SO₂ and NO₂ (Gao et al., 2022). Thus, corporate emission reduction measures simultaneously cut these pollutants, achieving SPPC. However, research on green fiscal stimulus policies (GFSP) as incentive-based tools for SPPC remains are relatively few.

GFSPs are important incentive-based policies that achieving SPPC through specific fiscal measures, such as green investment, taxes and subsidies. Unlike market-based or command-based policies, GFSP integrates their advantages to play a critical role in environmental governance (Zhao et al., 2023). This policy framework offers dual benefits: First, it guides capital toward low-carbon and pollution-reduction projects through market instruments such as tax breaks and targeted subsidies, motivating enterprises and the public to proactively cut emissions (Zhao et al., 2022). For example, Dutch government levies more than ten kinds of green taxes, including CO₂ tax, and vigorously invests in renewable energy projects, which not only reduces the dependence on fossil fuels and lowers carbon emissions from the source, but also reduces air pollution during production, thus realizing SPPC. Second, it combines incentives with penalties for high-pollution and high-emission activities. Fines and regulatory measures not only directly reduce emissions but also drive enterprises to invest in green research and development (R&D) (Zhang, 2021). For example, flue gas desulphurization technology and particulate matter control equipment can reduce pollutants and greenhouse gases, thus realizing SPPC.

China is both a globally significant emerging economy and the largest carbon emitter. Against this backdrop, the Chinese government has set ambitious targets to achieve carbon peak by 2030 and carbon neutrality by 2050. Between 2000 and 2017, the number of deaths due to air pollution in China increased by 36% from 0.7 million to nearly 1 million (Xu et al., 2023), accounting for over a quarter of the global total (Yu et al., 2025). In response, the Chinese government has issued an air quality continuous improvement action plan, which targets a 10% reduction in urban PM_2.5 concentrations by 2025 compared to 2020. This means that while maintaining economic growth, China needs to achieve environmental governance goals within just over a decade. Achieving these goals requires a feasible policy package to achieve SPPC under tight fiscal constraints. Unlike countries where the market plays a dominant role in environmental governance, government intervention plays a pivotal role in China’s environmental governance relying on its authoritarian system (Tu et al., 2024). This system endows the government with strong centralized decision-making and enforcement capabilities, enabling it to effectively constrain polluting behaviors through policy tools such as environmental inspections and emission standards, and coordinate cross-regional and cross-departmental comprehensive governance. Exploring the impact of GFSP on SPPC in China is crucial.

In 2011, the Chinese government proposed the energy saving and emission reduction fiscal pilot (ESER), a typical GFSP. This policy was implemented in three batches across 30 selected cities, as illustrated in Fig. 1. The first batch was launched in 2011, and comprised eight cities (Chongqing, Beijing, Guiyang, Hangzhou, Jilin, Changsha, Shenzhen, Xinyu); the second batch was implemented in 2013 and constituted 10 cities: Shaoguan, Dongguan, Tangshan, Jingmen, Nanping, Tsitsihar, Shijiazhuang, Tieling, Tongling, and Tongchuan; and the third batch was executed in 2014 and encompassed 12 cities: Xuzhou, Baotou, Liaocheng, Deyang, Hebi, Lanzhou, Linyi, Nanning, Meizhou, Tianjin, Urumqi, and Haidong. ESER has set energy conservation and pollutant emission decrease targets in domains such as renewable energy and green buildings. It explicitly requires a 10% reduction in energy consumption per unit of GDP and a 10% decrease in the aggregate discharge of key pollutants. It encourages local governments to take more active pollution and carbon reduction measures through means such as central comprehensive incentive funds and provincial supporting funds. Specifically, these measures include supporting enterprises in R&D and optimizing urban energy structures through the scaled utilization of renewable energy. The incentive funds are allocated by city type, with municipalities, provincial capitals, and state-planned cities receiving 500–600 million yuan annually, and other cities getting 400 million yuan. Furthermore, compared with other GFSPs, ESER is unique in that it not only includes an incentive mechanism for pilot cities but also an environmental protection assessment and constraint mechanism. The environmental performance evaluation directly influences the allocation of next-year incentive funds. Pilot cities failing the assessment face the risk of losing their pilot status and having funds recalled. Consequently, local governments often enforce stricter pollution standards to meet central assessments and avoid fund recovery.

Fig. 1

Energy saving and emission reduction fiscal pilot.

In this study, we explored the role of ESER in SPPC. The study’s contributions can be summarized into three main aspects: First, unlike studies focusing only on carbon emissions or air pollution (Rao et al., 2013; Zhang et al., 2020; Cai et al., 2023; Guo and Zhang, 2024), this study explores the role of GFSP in SPPC from the perspective of targeted synergistic governance, complementing studies on the environmental benefits of GFSP. Although a few studies have examined the effect of ESER on SPPC, most of them are based on provincial data (Zhu et al., 2024). However, because ESER was implemented at city level, the impact of ESER on SPPC can be more accurately estimated based on city-level data.

Second, previous studies have emphasized the significance of technological advancement and environmental regulation in the impact of ESER on SPPC (Blackman et al., 2018; Bi et al., 2024). However, while ESER imposes stricter emission standards on pilot cities, the actual environmental investments of enterprises depend on their judgment of the governance attitude of the government, which has been over-looked. When policy signals are ambiguous or uncertain, enterprises may adopt short-term compliance measures instead of long-term technological investment, severely undermining ESER’s effectiveness. Ignoring this mechanism could lead to overestimating the real effect of ESER and failure to explain why the same policy yields significantly different governance outcomes in various pilot cities. This study explored how climate policy uncertainty influences the impact of GFSP on SPPC, providing evidence to accurately evaluate policy effects and understand effect disparities.

Third, this study found that ESER had no significant spillover effect on non-target pollutants, but exhibited significant spillover effects on the non-target emission methane. And after methane was included, some cities realized the optimum SPPC efficiency value of 1. Before ESER implementation, methane emissions grew at an annual rate of 2.63%, which dropped to 0.89% post-policy. Existing research has often overlooked these spillover effects. Although a few studies explored ESER’s spillover effects on non-target pollutants and found no significant impact, they overlooked the policy’s effect on non-target emissions (Wang and Qiu, 2021; Fan and Liang, 2023), thus underestimating its environmental benefits. This finding provides evidence for optimizing policy design to integrate emissions like methane into synergistic management.

Literature review and research hypotheses

Environmental effect of green fiscal stimulate policy

GFSPs are important incentive-based policies, referring to measures such as taxation and subsidies (Zhao et al., 2023) used by governments to achieve green development. Common GFSPs include environmental taxes, fiscal subsidies and green investments (Baranzini et al., 2017). As GFSP combine the advantages of market-based and command-based policies, numerous studies have explored their environmental effects. Existing research can be broadly categorized into two types:

One strand of literature examines the emission and pollutant reduction effects of GFSPs. According to the externality theory, GFSPs can internalize external environmental costs by increasing the emission costs of firms through taxation (Becker et al., 2013), and they can also promote energy efficiency and emission reductions (Bi et al., 2024) by directing and increasing investment in innovation, which can effectively reduce CO₂ (Mehta and Prajapati 2024), NO_x (Bonilla et al., 2018) and PM_2.5 (Mardones and Cabello, 2019). Another strand of literature reports that GFSPs also have a positive effect on improving air quality (Greenstone and Hanna, 2014). For example, Germany’s low-carbon zone policy was effective in reducing PM₁₀ levels, leading to improved local air quality and lower infant mortality (Gehrsitz, 2017). As an important form of GFSP, fiscal subsidies also play an important role in improving air quality. Wang and Ma (2024) found that the new energy vehicle subsidy policy significantly reduced air pollution. In the long run, fiscal subsidies also supported R&D. In addition, many studies have explained the mechanism by which GFSPs improve air quality from the perspective of industrial structure upgrading (Wang and Qiu, 2021; Liu and Li, 2022).

While GFSPs are widely recognized as an effective means of reducing carbon emission and air pollution, their potential unintended consequences, the so-called “green paradox”, have also garnered academic attention. This paradox suggests that, rather than improving the environment as expected, GFSPs may exacerbate environmental problems owing to imperfect policy design; this may lead to an increase in pollution emissions, and governments at all levels may incur high expenditure and energy wastage owing to the duplication of mechanisms (Byrom et al., 2020), which in turn affects the effectiveness of emission reductions. Moreover, fiscal decentralization cannot be disregarded, and Zhang et al. (2017) found that a single environmental policy may reduce carbon emissions; however, the policy lag caused by fiscal decentralization promotes carbon emissions, leading to the “green paradox”. In addition, factors such as the asymmetry between environmental information disclosure, public demand, and the inspection system (Wang et al., 2023) impact the policy effect of GFSPs, and cause increased pollution in response.

Synergistic performance of air pollution control and carbon mitigation

Although SPPC has gradually garnered attention, many scholars have tended to limit their attention to a single environmental indicator when evaluating environmental policies, such as focusing only on decreasing emissions or pollutants, while neglecting the intrinsic connection and mutual influence between them (Moz-Christofoletti and Pereda, 2021). In fact, air pollution and carbon emissions share common origins and process characteristics (Gao et al., 2022), and both often arise from common activities, including fossil fuel combustion and industrial production. These activities not only consume a large amount of energy, but they also release large amounts of emissions and hazardous substances, which have a serious adverse effect on the environment. Therefore, combating carbon emission and air pollution should be viewed as an integrated whole, and dealt with synergistically rather than separately (Du and Li, 2020).

To explore the effects of SPPC in depth, scholars have conducted research from multiple dimensions. In the spatial dimension, certain studies have revealed the heterogeneity of SPPC. For example, Li et al. (2020) employed the four-quadrant classification method and the Moran index to find significant regional variations in SPPC. Dong et al. (2019) employed the LMDI decomposition method to quantify SPPC, revealing that coordinated emission reduction serves as the predominant driving force behind PM_2.5 abatement.

Certain studies have evaluated the effectiveness of specific environmental policies on SPPC. For example, the United States’ policies to reduce fossil fuel combustion effectively reduced CO₂, particulate matter, and toxic substances (Boyce and Pastor, 2013). China’s Three-Year Action Plan for Air Pollution Prevention and Control enhanced SPPC. Meanwhile, researchers have investigated SPPC effects of air pollution control regulations (Gao et al., 2022) and low-carbon pilot policies (Wang et al., 2024). However, although studies have pointed out the effect of policies in SPPC, there remains a research gap in the role of GFSP in SPPC from the perspective of incentive-based policies.

The environmental effect of ESER

As a typical GFSP, ESER has been examined by many studies from different perspectives, especially its environmental effects. Some studies have found that ESER can reduce air pollutions and carbon emissions. For example, Xu et al. (2022) discovered that ESER can reduce carbon emissions by 15.26%. ESER can also reduce wastewater discharges, PM_2.5, industrial dust and SO₂ in the short term (Lin and Zhu, 2019a). However, Wang and Qiu (2021) pointed out that its long-term effectiveness gradually weakens. Meanwhile, ESER also plays a positive role in SPPC. By issuing targets for combined multi-pollutant control, ESER directly promotes local governments to achieve synergistic governance. Lin and Zhu (2019a) found that the policy significantly influences the eco-efficiency; however, the effect was only significant after three years. Fan and Liang (2023) drew similar conclusions following a study of the combined impact of pollutants and carbon emissions. Another strand of literature explores the spillover effects of the ESER. Lin and Zhu (2019b) found that the policy has no spillover effect on the non-policy target pollutant PM_2.5. Sun and Feng (2023) argue that although the policy is capable of driving energy conservation and emission reduction within the city, it fails to generate a notable spillover effect on the energy conservation and emission reduction efforts of neighboring cities.

Moreover, in exploring how ESER policies influence SPPC, existing studies mainly focused on the roles of technological innovation, industrial structure and energy structure elevation. Technological innovation and reducing energy intensity are regarded as important drivers for achieving SPPC (Du and Wang, 2022). Some scholars argue that policy effectiveness benefits from industrial structure upgrading and energy efficiency improvements (Sun and Feng, 2023). Weakening the assessment of economic growth targets is also a key for ESER to achieve SPPC (Fan and Liang, 2023).

However, existing studies have overlooked that enterprises’ perception of the government’s attitude towards environmental governance, namely climate policy uncertainty, is the key factor in determining whether enterprises adopt environmental actions. Enterprises require long-term investment for technological innovation and industrial upgrading, which hinges on a stable policy environment. ESER imposes stricter environmental standards on pilot cities, demonstrating the government’s resolve in environmental governance. This provides enterprises with a stable external policy environment, encouraging technological innovation and industrial structure upgrading to achieve SPPC. Therefore, the role played by climate policy uncertainty in the impact of ESER on SPPC requires further exploration.

Research gaps

There are certain research gaps in exploring the impact of GFSPs on SPPC. First, most studies have focused on a single environmental objective, such as carbon emission or air quality, whereas the SPPC of GFSPs has been disregarded. For example, Zhang et al. (2020) and Cai et al. (2023) focused on the impact of decreasing carbon emissions and empirically analysed the carbon mitigation potential of GFSPs and their influencing factors. Similarly, Guo and Zhang (2024) focused on air quality improvement and explored the contribution of GFSPs to achieving this. However, none of these studies have addressed the integrated effects of GFSPs in lowering pollutant and carbon emissions.

Second, when exploring the mechanisms, the existing literature primarily concentrates on the role of technological innovation and industrial transformation. For example, Du and Wang (2022) studied the influence of GFSP on technological innovation and how this can promote the improvement of environmental quality, while Sun and Feng (2023) examined the role of GFSP on the optimization of industrial structure and how this optimization can enhance environmental benefits from the perspective of industrial transformation. However, none of these studies have explored the potential impact of climate policy uncertainty on the environmental benefits of GFSPs. Climate policy uncertainty may lead to investment decision-making attitudes of enterprises, which may affect the implementation effect of GFSPs.

Third, in assessing the environmental benefits of GFSPs, previous studies have primarily considered the direct impacts of the policies on targeted pollutants or emissions. However, GFSPs may have spillover effects on non-target emissions. Such spillover effects are often overlooked in existing studies, which may lead to an underestimation of the environmental effects of GFSPs.

Research hypotheses

ESER effectively promotes SPPC through the dual mechanism of mandatory constraints and financial incentives. The pilot program establishes an environmental assessment for pilot cities, and those that fail to meet the standards will have their incentive funds withdrawn or be disqualified from the pilot program. Pilot city governments therefore tend to set more stringent mandatory emission standards to ensure enterprises fulfill the assessment’s requirements (Abbass et al., 2022). On the one hand, strict emission standards raise the cost of emissions for enterprises, forcing them to undergo green transformation. On the other hand, the fiscal incentives (Zhou et al., 2023) reduce the fiscal pressure on the technological upgrading of the enterprises, prompting them to achieve SPPC. According to the analysis above, we propose the first hypothesis:

H1: ESER policies can improve the efficiency of SPPC in pilot cities.

Climate policy uncertainty may have a negative impact on the effectiveness of green fiscal incentives. This uncertainty is mainly reflected in frequent policy changes, unstable implementation and unclear environmental standards, which will increase the investment risk of enterprises and inhibit their investment in R&D (Ren et al., 2022). By setting stricter and clearer carbon and pollutant control targets in the pilot cities, ESER sends a stable and clear policy signal to the market. The incentive fiscal funds and policy guidance can reduce the risk of uncertainty faced by enterprises, prompting them to be more proactive in making long-term green investments and accelerating the application of clean technologies (Fuss et al., 2009), thus enhancing the efficiency of environmental governance. We therefore propose hypothesis 2a:

H2a: ESER will improve SPPC in cities by reducing climate policy uncertainty.

ESER promotes corporate green technology innovation through incentives and environmental regulation, thereby enhancing SPPC. According to the Porter’s hypothesis (Liu and Li, 2022), environmental policies can promote the technological innovation and increase productivity of enterprises. In the process of policy implementation, incentive funds from the central government and matching funds from local governments can reduce the costs of R&D (Wang and Qiu, 2021); while strict emission regulations (e.g., penalties for non-compliance) increase the environmental costs of high-polluting enterprises (Ambec et al., 2013), resulting in the “innovation compensation” effect, which will jointly push enterprises from passive pollution control to active emission reduction. Consequently, hypothesis 2b is proposed:

H2b: ESER policies will improve the efficiency of SPPC by promoting technological innovation.

To meet environmental assessment targets, pilot city governments set stringent emission limits and strengthen regulations. Enterprises must comply with specific limits or face high non-compliance costs. For instance, Shenzhen city has implemented total control of key pollutants, requiring new construction projects to adopt a two-fold reduction in VOC emissions as a substitution, and limiting NOₓ from new natural gas boilers to 30 mg/m³. These strict measures raise non-compliance costs, creating a strong deterrent (Chen et al., 2022). Under policy pressure, enterprises invest in green and low-carbon transformation by optimizing processes and adopting cleaner technologies. This reduces pollutant emissions directly and cuts carbon emissions via technological upgrades, achieving SPPC. Thus, we propose the second sub-hypothesis H2c:

H2c: ESER policy will improve the efficiency of SPPC by increasing the level of corporate emissions regulation.

Method and data

DEA model for estimating SPPC

The data envelopment analysis (DEA) model uses non-parametric methods to evaluate decision-making units (DMUs) by defining an optimal production frontier as a benchmark. It converts multiple inputs and outputs into efficiency ratios, avoiding measurement inconsistency. DEA is widely applied in energy and environmental efficiency assessments. The directional distance function (DDF), a generalization of the radial DEA model (Cheng et al., 2022a), accounts for non-desired outputs but assumes proportional changes in inputs/non-desired outputs relative to desired outputs, often leading to efficiency overestimation (Cheng et al., 2022b). Zhou et al. (2012) proposed the non-radial directional distance function (NDDF) to address this limitation, enabling non-proportional adjustments of outputs/inputs and reducing relaxation bias. Thus, we adopted the NDDF-DEA model proposed by Zhou et al. (2012) for SPPC evaluation.

Assuming that each city is a DMU, the input factors comprising capital $K$, labour $L$, and energy $E$ for each DMU are expected to produce outputs as GDP; however, non-desired outputs, such as air pollution $PM$ and carbon emissions $C$ are simultaneously produced. Therefore, a production technology $T$, that contains non-desired outputs can be defined, which satisfies the principle of strong disposability for inputs and desired outputs, as well as weak disposability for non-desired outputs (Färe et al., 2007):

$$T=\{(K,L,E,Y,C,PM):(K,L,E)\,{\rm{can}}\,{\rm{produce}}(Y,C,PM)\}$$

(1)

This leads to the definition of a NDDF under the production technology T:

$$\overrightarrow{D}(K,L,E,Y,C,PM;{\bf{g}})=\sup \{{{\bf{w}}}^{T}\beta :[(K,L,E,Y,C,PM)+{\bf{g}}\cdot diag(\beta )\in T]\}$$

(2)

where ${{\bf{w}}}^{T}={({w}_{K},{w}_{L},{w}_{E},{w}_{Y},{w}_{C},{w}_{PM})}^{T}$ is a vector of weights for inputs, desired outputs, and undesired outputs; $\beta =({\beta }_{K},{\beta }_{L},{\beta }_{E},{\beta }_{Y},{\beta }_{C},{\beta }_{PM})\ge 0$ is a scaling factor indicating the proportion of inputs and outputs that may vary or be redundant; and ${\bf{g}}=({g}_{K},{g}_{L},{g}_{E},{g}_{Y},{g}_{C},{g}_{PM})$ is a vector of directions of input and output adjustments.

When the NDDF $\overrightarrow{D}=0$ of the DMU is in the direction g, then this DMU is regarded as being on the efficiency frontier. The linear programming model can be expressed as follows,

$$\begin{array}{lll}{D}_{\tau }({\bf{x}},{\bf{y}},{\bf{b}};{\bf{g}})=\mathop{\max }\limits_{\beta ,\lambda }{{\bf{w}}}^{T}\beta ,{\bf{x}}=K,L,E,{\bf{y}}=Y,{\bf{b}}=C,PM\\ s.t.\mathop{\sum}\limits _{\tau \in {\varGamma }_{t}}\mathop{\sum }\limits_{j=1}^{J}{\lambda }_{j\tau }{{\bf{x}}}_{j\tau}\le {\bf{x}}+diag({\beta }_{{\bf{x}}})\times {{\bf{g}}}_{{\bf{x}}}\\\qquad \!\mathop{\sum}\limits _{\tau \in {\varGamma }_{t}}\mathop{\sum }\limits_{j=1}^{J}{\lambda }_{j\tau }{{\bf{y}}}_{j\tau }\ge {\bf{y}}+diag({\beta }_{{\bf{y}}})\times {{\bf{g}}}_{{\bf{y}}}\\\qquad\! \mathop{\sum}\limits _{\tau \in {\varGamma }_{t}}\mathop{\sum }\limits_{j=1}^{J}{\lambda }_{j\tau }{{\bf{b}}}_{j\tau }={\bf{b}}+diag({\beta }_{{\bf{b}}})\times {{\bf{g}}}_{{\bf{b}}}\\ \,\,\,\quad\qquad\beta \ge 0;{\lambda }_{jr}\ge 0,j=1,{\mathrm{..}}.,J\end{array}$$

(3)

where ${\lambda }_{jr}$ is the weight of the j_th DMU in the observation period, $\tau$. The NDDF is the maximum weighted sum of the adjustment components, $\beta$, such that $[diag({\beta }_{{\bf{x}}})\times {{\bf{g}}}_{{\bf{x}}},diag({\beta }_{{\bf{y}}})\times {{\bf{g}}}_{{\bf{y}}},diag({\beta }_{{\bf{b}}})\times {{\bf{g}}}_{{\bf{b}}}]$ can be produced under a given production technology. Depending on the needs, each input-output vector can be assigned different weights. In the absence of a priori information, inputs, along with desired and non-desired outputs, are assumed to be of equal importance; therefore, all three are assigned the same weight in this study. Furthermore, inputs consist of capital, labor, and energy consumption, and non-desired outputs consist of CO₂ and PM_2.5; therefore, the final weight vector is determined as ${{\bf{w}}}^{T}={(\frac{1}{9},\frac{1}{9},\frac{1}{9},\frac{1}{3},\frac{1}{6},\frac{1}{6})}^{T}$. Considering the need to reduce inputs and non-desired outputs while increasing desired outputs, we refer to Zhang and Choi (2013) to set the direction vector ${\bf{g}}=(-{g}_{K},-{g}_{L},-{g}_{E},{g}_{Y},-{g}_{C},-{g}_{PM})$. According to the solution, the non-efficiency value ($Dv$) is obtained. To ease the handling and interpretation of the results, the non-efficiency value is standardized as $\frac{1}{1+Dv}$, and defined as the pollution and carbon reduction collaborative management efficiency SPPC. SPPC is a variable between 0 and 1, and the larger the SPPC, the closer it is to the optimal efficiency frontier, and pollution and carbon reduction collaborative management is therefore more highly efficient. SPPC estimation offers advantages by integrating economic development and environmental protection goals, unlike single emission targets. It enables relative efficiency comparisons across cities, identifying best practices and improvement areas to assist policymakers in formulating effective environmental policies. Finally, SPPC allows dynamic temporal evaluation to monitor policy implementation and environmental governance progress.

Staggered DID estimation specification

DID method is an effective method for assessing the effects of policies that has been widely used in recent years in important areas such as energy economics and environmental governance. The basic premise is to subtract the changes before and after the policy shock from the changes before and after the treatment group was subjected to the policy shock, with the aim of obtaining the net effect of the policy shock. However, the change in the treatment group that remains unaffected by the policy shock is a counterfactual situation that cannot be observed. Consequently, the alteration in the time trend of the control group, which has characteristics like those of the treatment group, is commonly chosen as a substitute for the counterfactual situation. The policy impact is determined by the variance between the changes in the treatment group and the control group during the pre- and post- policy periods.

Considering that ESER was declared to be among the pilot cities in three different years, we constructed an asymptotic double-difference model that can adapt to the gradual implementation characteristics of the policy to identify the impact of the policy on the efficiency of collaborative governance. The benchmark model was set in the following way,

$$SPP{C}_{it}=\alpha +\beta di{d}_{it}+\eta {Z}_{it}+{\mu }_{i}+{\lambda }_{t}+{\varepsilon }_{it}$$

(4)

where $i$ represent cities and t denotes years; $SPP{C}_{it}$ represents the synergistic performance of PM_2.5 control and carbon mitigation; ${Z}_{it}$ is a vector of control variables, including characteristics like economic growth levels, demographic composition, industrial composition, level of urbanization, and degree of openness; ${u}_{i}$ and ${\lambda }_{t}$ denote individual and year fixed effects, respectively; while ${\varepsilon }_{it}$ represents a random error. $di{d}_{it}=trea{t}_{i}\times tim{e}_{it}$ is a dummy variable indicating whether or not city $i$ implemented the pilot in year t; $trea{t}_{t}$ denotes a dummy variable, $trea{t}_{i}=\left\{\begin{array}{l}1,{\rm{if}}\,{{\rm{city}}}_{i}\,{\rm{is}}\,{\rm{the}}\,{\rm{treatment}}\,{\rm{group}}\\ 0,{\rm{if}}\,{{\rm{city}}}_{i}\,{\rm{is}}\,{\rm{the}}\,{\rm{control}}\,{\rm{group}}\end{array}\right.$.$tim{e}_{it}=0$ before a certain policy is implemented, and $tim{e}_{it}=1$ after a policy is implemented. $\beta$ is the coefficient of interest in this study, representing the net effect of the policy. A significantly positive $\beta$ indicates that the implementation of the ESER can remarkably enhance the efficiency of the city’s SPPC.

When applying the staggered DID approach, a crucial prerequisite is the parallel trends hypothesis. There are no structural disparities between pilot and non-pilot cities, nor are there such disparities among batches of pilot cities. Even if there is a difference, it is fixed, and the trend of the development of synergistic governance efficiency over time should be consistent. To examine whether the sample within this study meets the parallel trend assumption, the following model was constructed with reference to the methodology outlined by Lalonde et al. (1993):

$$SPP{C}_{it}=\alpha +\mathop{\sum }\limits_{t=-5\& t\ne -1}^{6}{\beta }_{t}di{d}_{it}+\eta {Z}_{it}+{\mu }_{i}+{\lambda }_{t}+{\varepsilon }_{it}$$

(5)

where $di{d}_{it}$ represents dummy variables that $di{d}_{it}=\left\{\begin{array}{l}1,{\rm{if}}\,{\rm{city}}\,i\,{\rm{implemented}}\,{\rm{the}}\,{\rm{policy}}\,{\rm{in}}\,{\rm{year}}\,t\\ 0,\text{otherwise}\end{array}\right.$. This study focuses on the ${\beta }_{t}$ in Eq. (5), which captures the disparity in SPPC during the year of policy implementation.

To analyse the specific mechanism of the policy’s impact on SPPC, we refer to Heckman et al. (2013) to assess the mechanism via the following equation:

$${M}_{it}={\beta }_{0}+{\beta }_{1}di{d}_{it}+\eta {Z}_{it}+{u}_{i}+{\lambda }_{t}+{\varepsilon }_{it}$$

(6)

where ${M}_{it}$ denotes mechanism variables: climate policy uncertainty, technological innovation, and emission regulation. Other variables maintain their meanings from the baseline regression equation.

Variables

Dependent variable

Synergistic performance of air pollution control and carbon mitigation (SPPC). The measurement of SPPC included inputs, desired outputs, and undesired outputs. As reported by Zhang and Choi (2013), capital, labor, and energy consumption are treated as input variables, while GDP is considered the desired output. Specifically, the fixed capital stock calculated via the perpetual inventory method was regarded as the capital input in this study, the city’s year-end employment as the labor input, the city’s energy consumption as the energy input, and the GDP per capita as the desired output, and CO₂ emissions and PM_2.5 as non-desired outputs. The main reason for using these two as indicators of SPPC is that excessive carbon emissions are widely recognized as the main cause of global warming, while PM_2.5 is one of the most harmful components of air pollution, and these two indicators are representative of the study of pollution and carbon reduction. A higher SPPC value indicates that the city has a stronger synergistic governance capacity.

Independent variable

GFSP. The specific meaning of this variable is that a city takes the value of 1 when ESER is implemented and after its implementation, otherwise it takes 0. Three batches of 30 cities were selected for the policy, and the policy was implemented in the second half of the year in each of the three batches; considering the policy’s lagging nature, the policy start times in this study were set to 2012, 2014, and 2015. In addition, considering that there were missing data from individual cities at the level of explanatory and control variables, Haidong, Tangshan, Qiqihar, and Liaocheng cities were excluded from the sample in this study; the remaining 26 selected cities were included in the treatment group and the others in the control group.

Control variables

To control the influence of other factors, while avoiding the error of omitted variables and in combination with the existing studies, we incorporated the characteristics of the city’s urbanization level, economic development level, population, industrial structure, and external openness level into the control variables. Table 1 shows the descriptive statistics of each variable, which are briefly described as follows:

Table 1 Descriptive statistics.

The urbanization rate (urban_rate) is expressed as non-farm population/resident population (Lin and Zhu, 2019a). Urbanization may lead to improvements in urban infrastructure and public services, which are conducive to environmental protection; however, it may also exacerbate resource depletion, air pollution, and ecological damage.

The economic development level of the city (lngdp) is expressed by GDP per capita (Quan et al., 2020). Economic development has a significant impact on the effectiveness of environmental management. Cities boasting a prosperous economic development are more likely to have higher carbon emissions than other cities owing to industrial development and population density, but they also have greater financial and technological conditions to promote SPPC.

Population density (lnpop) is expressed as the logarithm of population per square kilometer (Zhu et al., 2021). Excessive population density may increase energy consumption and pollutant emissions (Li et al., 2019), but the concentration of talent can also boost green technology innovation and accelerate the low-carbon transition of cities.

The greening rate of built-up area (green_rate) is expressed as greening area/built-up area (Zhu et al., 2021). Some studies have proved that the urban greening level positively impacts the environmental quality (Nowak et al., 2006). A high level of greening can increase the area of urban green space, improve urban air quality, and enhance the stability of urban ecosystems, thus creating a more liveable urban environment.

Industrial structure (second) is expressed as output value of secondary industry/gross regional product (Cheng et al., 2018). The secondary industry is the main sector of air pollution and carbon emissions (Wu et al., 2017); if the proportion of the secondary industry increases, air pollutants and carbon emissions also increase, thus affecting environmental quality.

The external openness level (fdi) (Wen et al., 2022) is expressed as FDI/regional GDP. With China’s growing openness to the world, foreign investment as well as trade has increased substantially, and while promoting economic growth, the impact on the environment cannot be disregarded.

Data sources

Drawing on the panel data of 281 cities in China from 2007 to 2022, we investigated the impact of GFSP on SPPC. Data on the green coverage rate of built-up area, population density, secondary industry ratio, GDP per capita, and foreign direct investment ratio were sourced from the China City Statistical Yearbook and the various provincial statistical yearbooks and government work reports of each province. As reported by Lin et al. (2014), the PM_2.5 concentration data were obtained from the satellite-based raster data published by Columbia University, USA. Carbon emission data were obtained from the study by Chen et al. (2020). Climate policy uncertainty index data from the study by Ma et al. (2023) were obtained by investigating the tweets of six Chinese mainstream news media containing the topic of climate uncertainty on social media, and was computed using manual cross-checking as well as the MacBERT24 deep learning model.

Results and discussion

Synergic performance of air pollution reduction and carbon mitigation

Figure 2 illustrates the changes in SPPC efficiency of the 281 cities across China during the period 2007–2022. The range of the blue markers on the graph gradually expands over time, indicating that an increasing number of cities were included in the scope of the pilot synergistic governance. The blue color representing the SPPC efficiency of the pilot cities and the green color representing the SPPC efficiency of the non-pilot cities deepen simultaneously, indicating that SPPC gradually improved nationwide.

Fig. 2: Spatio-temporal trends of SPPC in China.

a-e SPPC efficiency of the 281 cities across China in year 2007, 2012, 2014, 2015, and 2022, respectively. The blue legend and the green legend represent the governance efficiency of pilot cities and non-pilot cities respectively, and the darker the color, the larger the SPPC value, i.e., the higher the efficiency of synergistic governance for pollution reduction and carbon reduction.

Over the period 2007 to 2022, the national average level of SPPC efficiency experienced a notable upward trend, rising from 0.398 to 0.490, with an average annual growth rate of around 1.31%. In contrast to the national average, the governance of the pilot cities exhibited a superior performance. The average SPPC level of the pilot cities was 0.397 in 2007, and by 2022, this value rose to 0.498, recording an average annual growth rate of around 1.43%. In contrast, while the average efficiency of non-pilot cities increased, this growth was significantly slower than that of pilot cities. In 2007 and 2022, the efficiency was 0.398 and 0.489, respectively, with an average annual growth rate of 1.30%.

Impacts of green fiscal policy on synergic performance

Baseline regression analysis

Table 2 reports the baseline regression results, where column (1) shows the regression results without the control variables and fixed effects, column (2) displays the regression results with the control variables and fixed effects, and column (3) uses the clustering standard errors based on column (2). The results show that ESER has a significant positive effect on SPPC. This finding preliminarily confirms hypothesis H1 proposed in this study. Specifically, the average SPPC efficiency in pilot cities was 1.60% higher than that in non-pilot cities.

Table 2 Role of GFSP in the SPPC.

In this study, ESER contributes to the efficiency of SPPC, which is consistent with the results obtained from the established literature on studies conducted at the provincial scale (Zhu et al., 2024). However, provincial studies yielded efficiency gains as high as 14.390%, which is significantly different from the estimates in this study. This suggests that provincial-level studies may mask the spatial heterogeneity of policy effects due to data aggregation, leading to an overestimation of policy effects. Therefore, studies at the city scale are recommended, as they would allow for more accurate estimates.

Parallel trend check

Figure 3 reports the parallel trends test results. The parallel trends test is a method in DID modeling used to verify whether the trends of the treatment and control groups are parallel before the intervention. When facing limited pre- and post-policy data, studies often aggregate data (Zhu et al., 2021). Given the limited data for 5 years before and 6 years after policy implementation, this study aggregated pre-policy data into period −5 and post-policy data into period 6. We chose period −1 as the base period for three reasons: First, excluding one period avoids multicollinearity between year dummies and treatment group interaction terms. Second, selecting period −1 (closest to the policy implementation) under the no-anticipation assumption minimizes structural differences in early data, ensuring ESER effect estimation accuracy. Finally, using period −1 as the base period is a common practice in academia, enhancing comparability with existing studies (Li et al., 2024; Cheng et al., 2024).

Fig. 3: Parallel trend checks.

The parallel trend test needs to satisfy the principle of ex ante and ex post significance. The confidence intervals of the coefficients of the pre- period in the figure are all through 0, indicating that there is no significant difference between the cities before the implementation of the policy; the coefficients of the post period have no significant difference at the beginning, but in the 5th and 6th period, the confidence intervals of the coefficients do not go through 0, which indicates that the implementation of the policy resulted in a significant difference, and indicates that the test for parallel trends was passed.

The parallel trends test requires the assumption that coefficients are insignificant pre-policy and significant post-policy. Results show that none of the coefficient estimates were statistically significant before ESER implementation, indicating similar trends between treatment and control groups without policy intervention. From the 5th period after ESER implementation, the confidence interval for the coefficient difference no longer included zero, confirming significance. This suggests ESER significantly improves SPPC efficiency in pilot cities with a lagged effect, validating the baseline regression that ESER effectively enhances SPPC efficiency.

Placebo test

Existing placebo tests mainly include time placebo and city placebo tests. The time placebo test involves constructing false policy timing (Lin and Zhu, 2019b), while the city placebo test constructs false treatment groups (Cai et al., 2016). For the time placebo test, as shown in Table 3, we first assumed ESER was initially implemented in 2006, with the three batches of pilot cities included in 2006, 2008, and 2009. The second assumption set ESER’s initial implementation in 2007, with batches included in 2007, 2009, and 2010, thereby constructing false policy timings. If SPPC efficiency coefficients then became significantly positive, it would indicate improvements stemmed from other factors rather than ESER. Results show pre-policy coefficients were all insignificant, confirming that SPPC improvements originated from ESER rather than other factors, thus validating the causal link between ESER and SPPC efficiency.

Table 3 Time-placebo test.

For the city placebo test, we randomly selected 26 sample cities as the spurious treatment group and designated the rest as the control group. This approach enables the estimation of ESER’s impact on SPPC. Repeating this 500 times generates 500 coefficient estimates and p-values. As shown in Fig. 4, the horizontal axis plots placebo ESER coefficients, the left vertical axis shows p values for significance testing, and the right vertical axis is the kernel density, which reflects the probability density distribution of the values of the spurious regression coefficients. The higher the curve, the higher the frequency (density) of occurrence of the corresponding regression coefficient. The regression coefficients are mainly distributed on both sides of the 0-value, presenting normal distribution characteristics, which are significantly different from the benchmark regression results (coefficient = 0.016). Most of the p-values are greater than 0.1, and there is no significant difference in SPPC between the cities in the spurious treatment and control groups. This result indicates that the study sample passes the city placebo test, ruling out that the baseline regression results in this paper are due to unobservable factors.

Fig. 4

City-placebo test.

Robustness check

To further validate the findings, we conducted a series of robustness tests, with results shown in Table 4. First, to reduce the impact of outliers, we winsorised explanatory variables at the 1% and 5% levels in columns (1) and (2). Second, recognizing that GFSP’s effect on SPPC is not immediate, we lagged the model city construction dummy variable by one period in column (3). Finally, accounting for city-specific temporal trends in air pollution and carbon emissions, we added province-year interaction terms to the baseline model to identify cleaner measures in column (4). Meanwhile, to address potential unobservable factors in pilot city selection, we employed the propensity score matching (PSM) DID method to mitigate endogeneity from sample selection bias. By matching each treatment group sample with the most similar control group counterpart (one-to-one matching), we formed a validated treatment group for evaluation, with results in column (5) presented below. Additionally, other energy-saving policies besides ESER may influence carbon emissions. China implemented the low-carbon city pilot in 2010 and the carbon trading market policy in 2011. To isolate their effects, we added dummy variables for these policies to the baseline model. Columns (6) to (8) report results after controlling for other energy-saving and emission reduction policies. Finally, the premise of accurately identifying the effects of ESER is that the pilot and non-pilot cities should be randomly selected. In practice, however, the selection of model cities is often based on various city characters. Differences in the inherent characteristics of these cities may have different impacts on their collaborative governance efficiency over time, leading to estimation bias, which is the problem with non-random selection. To address this, we followed Cheng et al. (2018) by adding interaction terms between pilot selection criteria, control variables, and time trends to the baseline model. Selection criteria included whether cities were provincial capitals, sub-provincial cities, or one of two control zones cities. Time trends comprised linear and quadratic terms. These interactions control for time-varying urban effects on emissions, mitigating bias from non-random selection. Column (9) includes control-time trend interactions, while column (10) adds quadratic terms. The coefficients are significant, validating baseline results.

Table 4 Robustness test results.

The results show that all estimated coefficients passed the test at various significance levels and these coefficients were comparable to those of the baseline regression results; therefore, the baseline regression results can be considered robust: ESER helps to improve SPPC.

Mechanism analysis

In this study, climate policy uncertainty, emissions regulation, and technological innovation were used as variables in the mechanism test, and Table 5 presents the outcomes.

Table 5 Mechanism analysis results.

Column (1) shows that ESER significantly reduces climate policy uncertainty. Under high uncertainty, enterprises adopt conservative strategies, cutting investments in new technology R&D and clean production upgrades to avoid risks from policy volatility and market fluctuations. Ren et al. (2022) found climate policy uncertainty significantly lowers enterprises’ total factor productivity, hinders R&D investment, and deters clean technology innovation. With reduced uncertainty, firms can better assess investment payback periods, investing resources in energy-saving R&D with greater confidence and lowering extra costs from policy changes. Moreover, lower uncertainty enhances policy implementation efficiency and credibility, ensuring long-term stable progress toward policy goals and providing institutional support for environmental improvement and green economic development. Fuss et al. (2009) further emphasized that reduced policy uncertainty drives policies toward more effective and sustainable trajectories, promoting environmental quality and green growth.

In this study, the three industrial wastes (industrial wastewater discharge, SO₂, fume and dust emissions) were used as the main components of enterprise emissions. Principal component analysis was used to reduce the dimensionality and determine the weights of the different indicators, thereby constructing a composite indicator (Bonelli and Manni, 2019), which was used as a proxy variable for enterprise emissions. Its reduction directly reflects the strictness of regional emissions regulation, and stronger emissions regulation is regarded as a key factor to enhance the efficiency of SPPC. Column (2) indicate that effect of ESER on enterprise emissions was negative and passed the significance test at 1% level. This demonstrates that the ESER pilot cities implemented more stringent emission regulations to promote enterprise emission reduction, which in turn helped to improve SPPC.

In this study, the ratio of technology expenditure to GDP was selected to gauge the level of technology innovation. This ratio intuitively reflects the government’s support and attention to technology innovation activities. The increase in technology expenditure not only means that the government has invested more directly in R&D funding, scientific research projects, and talent training, but it also reflects the government’s determination and action to promote technological innovation, industrial upgrading and green transformation. Under such a policy orientation, enterprises are more likely to effectively deal with environmental problems by developing advanced clean technologies, optimizing industrial structure, thus improving the overall efficiency of SPPC. Column (3) of Table 6 indicates that the ESER significantly increased technology expenditure. The increase in technology expenditure can effectively promote R&D and the application of green technology, which in turn reduces pollution emissions and improves environmental quality.

Table 6 Spillover effect.

The regulatory and innovation effects of ESER at the provincial level remain robust and effective at the city level, further confirming the robustness of the policy’s mechanism (Wang and Qiu, 2021; Zhu et al., 2024). Building on this, this study further validates the role of climate policy uncertainty in ESER’s synergistic governance. When pilot cities implement stricter emission standards, clear policy signals enable firms to perceive local governments’ commitment to environmental governance, prompting them to abandon short-term compliance strategies and shift toward long-term technological transformation. This finding not only helps accurately assess ESER’s environmental effects but also explains from the perspective of firm decision-making why the same policy yields different governance outcomes across cities, providing empirical evidence for optimizing policy design.

Further analysis

Spillover effect

The core purpose of green fiscal stimulus policies is to achieve SPPC; however, they may also have spillover effects on other non-targeted pollutants or emissions, as these pollutants and emissions originate from the same industrial production and urban economic activities (Fang et al., 2019; Gao et al., 2022).

We analysed methane (CH₄), nitrous oxide (N₂O), and sulphur dioxide (SO₂). Methane is the second most important greenhouse gas, trailing only CO₂. It has a global warming potential 25 times greater than CO₂ (Mendoza-Cano et al., 2023) and increases the risk of human cardiovascular disease. Nitrous oxide is a potent greenhouse gas with a non-negligible greenhouse effect. It not only damages the ozone layer but also threatens organism survival and ecosystem stability. Sulphur dioxide, primarily from coal combustion and industrial activities, reacts in the atmosphere to form acids, causing acid rain. This damages ecosystems, worsens air pollution, and directly threatens human health, contributing to respiratory and cardiovascular diseases. Thus, monitoring and controlling emissions of these three gases is crucial for sustainable environmental governance.

When measuring SPPC efficiency, we separately included these types of pollutants or emissions as non-desired outputs, recalculated the efficiency, and conducted regression analysis. Table 6 shows the results of recalculating the SPPC before regression. The impact of GFSP on SPPC remained significant after including methane emissions. This implies that GFSP plays a key role in reducing methane emissions. In contrast, nitrous oxide (N₂O) and sulphur dioxide (SO₂) emissions did not show significant reductions in the model, suggesting ESER may prioritize emission control over pollutant reduction. Methane emissions grew at an average annual rate of 2.63% before ESER implementation; although still increasing post-policy, the rate dropped to 0.89%, indirectly confirming ESER’s effect on methane control. Additionally, in contrast to the findings of Lin and Zhu (2019a), who argued local governments only focus on target pollutant reduction, this study finds ESER has no spillover effect on non-target pollutants but significantly affects non-target methane emissions. Incorporating methane control even led some cities to achieve the optimal SPPC efficiency score of 1, which partially explains why previous studies (Wang and Qiu, 2021) underestimated policy effectiveness by reporting maximum efficiency scores below 1.

Heterogeneity analysis

Variations in economic structure, industrial layout, and strength of environmental policy implementation among cities can affect SPPC in various types of cities (Cai et al., 2023). The synergistic governance effect of GFSPs may be more significant because cities with more economic development have more adequate financial and resource inputs (Cheng et al., 2018).

Table 7 columns (1) to (3) portray the differences in SPPC due to geographic locations from three perspectives: western, central, and eastern. Columns (4) to (5) and columns (6) to (7) differentiate the differences in SPPC among cities from whether they are old industrial-base cities and whether they are resource-based cities; columns (8) to (9) distinguish cities from other cities by using the number of municipal districts under the jurisdiction of cities as the criterion of delineation, with reference to the Zivin et al. (2020) and Fonseca and Matray (2024). To characterize the differences in collaborative governance efficiency caused by the high or low level of competition within the jurisdictions, the sample was grouped according to a certain proportion: sample cities with the top 30% of the number of municipal jurisdictions were categorized as high-competitiveness cities, while those within the bottom 30% of the number of municipal jurisdictions were categorized as low-competitiveness cities.

Table 7 Further analysis.

The results show that the SPPC of GFSPs was more significant in the eastern cities, and this was associated with the relatively higher GDP level in these areas, as indicated by Du and Wang (2022). In addition, the non-old-industrial-based and non-resource-based cities showed significant increases in synergistic governance efficiency. Furthermore, the estimated coefficients for cities featuring higher degrees of intra-jurisdictional competition were notably positive. This result aligns with that of Fan and Liang (2023).

The differential distribution of fiscal incentives may explain the differences in SPPC between cities. Second-generation fiscal federalism theory suggests that fiscal incentives play a key role in local government initiatives and environmental economic performance (Weingast, 2009). ESER would provide a certain amount of incentive funding to selected pilot cities, with 600 million, 500 million and 400 million yuan in funding awarded depending on the city level. To further verify whether the difference in fiscal incentives causes the difference in policy effects, we categorized the cities that received 500 million yuan or more in incentives as high-fiscal-incentive cities, and those that received 400 million yuan in funds as low-fiscal-incentive cities. We then constructed two 0–1 variables, corresponding to whether a city was a high-fiscal-incentive city and whether it was a low-fiscal-incentive city, respectively. For each variable, the value was set to 1 if the city fell into the corresponding category, and 0 otherwise. All non-demonstration cities are non-financially incentivized cities, and the values of the two variables were 0. Based on this, we constructed an interaction term between the core explanatory variables and these two variables and incorporated them into the model in column (10).

The results indicated that the coefficients of both interaction terms were significant. This suggests that fiscal incentives can effectively drive the execution of measures aimed at achieving SPPC. Further analysis showed that the coefficients for the interaction between the explanatory variables and cities with high fiscal incentives were more significant than those for the interaction between the explanatory variables and cities with low fiscal incentives. This implies that cities with high fiscal incentives achieve better results in enhancing the SPPC compared to cities with low fiscal incentives. This may be since high financial incentive cities have more resources to invest in environmental protection and low carbon development programs due to more financial support. This not only helps these cities advance the relevant projects quickly, but it also enables them to make greater breakthroughs in technology and innovation, thus realizing the goal of reducing pollution and carbon more effectively.

From the perspective of fiscal incentives, eastern cities, non-resource-based cities, and non-old industrial base cities are mostly provincial capitals and sub-provincial cities, enjoying higher fiscal incentive levels. These cities have advanced economic development, cleaner industrial structures, sophisticated production technologies, and high resource efficiency. Sophisticated technologies can not only optimize industrial processes and reduce energy consumption but also effectively control pollutant and carbon emissions. This advantage enhances their competitiveness in ESER implementation, thus improving the policy’s SPPC effectiveness. Further, their relatively strong fiscal bases help balance economic development and environmental governance. With ESER incentive funding, they can allocate more resources to SPPC initiatives. This favorable fiscal environment creates optimal conditions for environmental policy implementation, enabling ESER to exert more pronounced synergistic effects.

For central-western, resource-based, and old industrial base cities, ESER’s SPPC effects face multiple constraints. First, shaped by resource endowments and industrial foundations, these cities rely heavily on traditional resource-based industries and heavy chemical industries. The long-formed path dependence complicates industrial transformation, hindering ESER’s effectiveness (Sun and Feng, 2023). Second, these cities generally have lower fiscal incentive levels, and the capital needed for industrial transformation far exceeds ESER funding. Due to limited fiscal incentives, these cities with scarce funds are forced to prioritize economic stability projects over SPPC initiatives. Finally, resource-dependent regions often exhibit preferences for non-clean technologies, leading ESER incentives to be channeled into maintaining traditional industries, a phenomenon known as the “incentive trap”. Funds primarily flow to end-of-pipe governance, such as environmental equipment upgrades, rather than technological innovation or industrial restructuring. This means ESER funding not only fails to break existing path dependence but also reinforces the short-term viability of traditional industries, creating a vicious cycle of policy-supported high-carbon development that prevents ESER from achieving SPPC.

For cities with a low level of competition in their jurisdictions, fiscal incentives may not be able to stimulate green innovation among firms. Firms, in the absence of competitive pressures, may have no incentive to engage in green innovation and high-performance transformation. Governments and businesses may tend to maintain the status quo rather than actively seek new development opportunities and breakthroughs. This may lead to limited effectiveness of fiscal incentives in promoting industrial upgrading and low-carbon and environmentally friendly transformation, thereby affecting long-term green development.

Existing literature predominantly evaluates ESER’s synergistic governance effects at the provincial level. Although it considers regional characteristics like geographical location and resource endowments that influence policy effectiveness, aggregating intra-provincial city data obscures information on the differential allocation of central fiscal incentive funds (Zhu et al., 2024). This makes it difficult for prior studies to identify how fiscal resource allocation disparities shape heterogeneous governance outcomes, thereby affecting the accurate assessment of ESER impacts. This study investigates the impact of ESER on SPPC at the city scale, revealing that differential fiscal fund allocation is the inherent driving factor influencing ESER’s governance efficacy. This finding provides a new perspective for explaining policy effectiveness variations and offers empirical support for optimizing ESER’s fund allocation system to enhance policy impacts.

Scenario simulation analysis

In this section, for the cities in the further analysis where GFSP has not yet exerted significant SPPC effect, we analysed the scenario simulation to provide these cities with the optimal strategies to achieve SPPC under fiscal constraints.

In the baseline regression, inputs, desired and non-desired outputs were assumed to be of equal importance in this study, with inputs consisting of capital, labor, and energy consumption, and non-desired outputs consisting of CO₂ and PM_2.5, with weights of (1/9, 1/9, 1/9, 1/3, 1/6, 1/6). When changing the importance of different elements, the effect of ESER policy on SPPC may differ. Therefore, we investigated the causes of the insignificant policy effects in the sub-sample cities by adjusting the relative weights of the elements and computing the SPPC in various simulation scenarios. These scenarios followed the guidelines of giving precedence to pollution reduction, prioritizing carbon reduction, and emphasizing synergy respectively. Specifically, the weight vector of the pollution reduction priority scenario was set as (1/9, 1/9, 1/9, 1/3, 0, 1/3), that of the carbon reduction priority scenario was set as (1/9, 1/9, 1/9, 1/3, 1/3, 0), and that of the synergism priority scenario was set as (1/12, 1/12, 1/12, 1/4, 1/4, 1/4). The respective SPPC for each city was calculated for the period 2007–2022. The strategy corresponding to the maximum efficiency value for each city in each year was counted and used as the optimal strategy for that city in each year. The optimal strategy that occurred most frequently in each year was taken as the optimal strategy for the city to realize SPPC. The distribution of the optimal strategies to realize SPPC for these cities that have not achieved synergy was reported in the form of a graph.

Figure 5 presents the scenario simulation results. Sub-figures a–d represent the scenario simulation analyses for central and western cities, old industrial bases, resource cities, and cities with a low degree of jurisdictional competitiveness during the period from 2007 to 2022, respectively. The different colors of the sectors in the figures represent different governance strategies, with red sectors representing pollution reduction as a priority strategy, blue sectors representing carbon reduction as a priority strategy, green sectors representing synergy as a priority strategy, and orange sectors representing the baseline scenario. The size of the different colored sectors indicates the number of times the strategy was used as the optimal strategy during the time. The largest portion of the sector shows the city’s optimal situation in achieving SPPC. The results show that among the central and western cities, old industrial bases, resource cities, and cities with a low degree of jurisdictional competitiveness, the baseline governance scenario applied only to individual cities, and the optimal governance strategy for most of the cities was synergistic prioritization of governance, while a few of the cities used pollution-reduction-prioritization and carbon-reduction-prioritization as the optimal governance strategy. It is worth noting that Sanming, Nanping, Shaoxing, and Longyan Cities had the same frequency of occurrence of pollution reduction priority and synergy priority strategies, indicating that these four cities had dual optimal governance strategies: pollution reduction priority governance and synergy priority governance.

Fig. 5: Scenario simulation analysis diagram.

a-d Scenario simulation analyses for central and western cities, old industrial bases, resource cities, and cities with a low degree of jurisdictional competitiveness during the period from 2007 to 2022, respectively.The different colors of the sectors represent different governance strategies, with red sectors representing pollution reduction as the priority strategy, blue sectors representing carbon reduction as the priority strategy, green sectors representing synergistic priority strategies, and orange sectors representing the baseline scenario. The size of the different colored sectors indicates the number of times the strategy was used as the optimal strategy during the time period. The largest portion of the sector shows the city’s optimal situation in realizing the synergistic effect of reducing emissions and carbon.

The scenario simulation analysis indicates that the current governance strategy is not the optimal governance strategy in the cities where the baseline regression results are not significant. Improvement in SPPC can be realized by adjusting the policy-focused management direction and matching the corresponding measures.

Conclusion and policy implication

GFSP are an important tool of promoting SPPC and achieving the goal of sustainable development. Based on the pilot of ESER as a quasi-natural experiment, a model was developed by constructing the NDDF and an analytical framework of the staggered DID method. This model based on the data of 281 cities in China from 2007 to 2022, and the impacts of the GFSPs on the governance efficiency and its mechanism were examined, to provide empirical evidence for the construction of a synergistic governance system as well as policy formulation for other economies.

The results showed the following: First, the implementation of GFSP improved the SPPC of cities by 1.60% and was significant at the 1% level; this held after several robustness tests. Second, the policy’s dynamic effects were analysed and were found to be insignificant in the short term; they only began to be significant after three years, indicating that the green fiscal stimulus policy had a certain lag. Third, the SPPC efficiency improvement in the pilot cities was mainly realized through reducing climate policy uncertainty, promoting technological progress, and restricting enterprises emissions. Fourth, GFSP has a more significant effect on SPPC improvement in eastern, non-old industrial-based, non-resource-based cities and cities with a higher degree of competition in their jurisdictions. Fifth, GFSP has a spillover effect on the reduction of non-targeted emissions methane.

Drawing on the findings, the following policy suggestions are proposed: First, the government should summarize existing successful cases and select representative old industrial base cities or central-western cities as new-round pilot cities. By implementing diversified green fiscal policies in these cities, it should explore implementation strategies suitable for cities with different geographic and economic backgrounds. Given the lagging effect of policies, a dynamic evaluation system integrating short-term and long-term goals is recommended. This system should periodically conduct comprehensive reviews of policy implementation effectiveness, financial resource use efficiency, and program progress. Evaluation indicators should include, but not be limited to, key metrics like air quality advancement and energy consumption structure changes.

Second, the fiscal incentive effect of green fiscal stimulus policies should be emphasized, as differential fiscal distribution may be a key factor influencing policy effectiveness. When formulating policies, the government should analyse each city’s fiscal status, economic development level, and environmental protection needs to implement differentiated fiscal incentive strategies. For example, cities with strong fiscal capacity and urgent environmental needs can receive increased funding to lead high-standard environmental projects; cities with weak fiscal capacity but high environmental potential can have reduced financial burdens via lower project start-up capital thresholds and extended funding cycles, encouraging active green transformation.

Third, the uncertainty of climate policy should be reduced, technological progress should be promoted, and emissions regulation should be reinforced. The government should ensure the continuity and predictability of climate policies, reduce the uncertainty caused by policy changes, and provide clear policy signals for long-term urban planning and SPPC. In addition, it should encourage the industrial structure optimization, boost the development of high-tech industries, and guide enterprises to increase green technological innovation.

Fourth, a multi-objective synergistic governance assessment system should be built to refine and precisely formulate policies. The government should consider the synergistic control effects of various pollutants, formulate comprehensive assessment indicators, and regularly evaluate the city’s SPPC to identify problems and take measures to improve them in a timely manner. In addition, more detailed policy measures should be formulated to promote the reduction of pollutant gases. For example, for N₂O, the use of abatement technologies and optimization of fertilizer management can be encouraged. This will help to strengthen the effect of the policy, which will in turn achieving SPPC.

This study provides valuable evidence on the impact of GFSP on SPPC, however, it still has some limitations that should be taken into account. Due to the lack of enterprise-level data, the impact of stricter environmental standards for enterprises in ESER pilot cities on SPPC is difficult to verify at the enterprise level, meanwhile, as environmental standards for enterprises vary across ESER pilot cities, the impact of such differences on SPPC effectiveness may be further discussed in the near future.