Environmental Decentralization, Heterogeneous Environmental Regulation, and Green Total Factor Productivity—Evidence from China

Abstract

The effective enhancement of green total factor productivity (GTFP) through macro-regulatory tools—environmental decentralization and environmental regulation and thus the promotion of high-quality and sustainable economic development—is a hot topic of current research. However, many studies have focused on how environmental decentralization or environmental regulation affects green total factor productivity, lacking attention to the relationships and impact paths among the three. To clarify the mechanisms of action of the three effects, this paper measures the GTFP of 30 Chinese provinces and cities from 2010 to 2020 through the Super-SBM model. The mediating effect of environmental regulation between environmental decentralization and GTFP is examined. Firstly, the study findings suggested that environmental decentralization is significantly negatively related to GTFP, while different environmental regulations are all significantly positively related to GTFP. Secondly, environmental decentralization suppresses GTFP in eastern China, which has a non-significant effect in central China. It has a catalytic effect on GTFP in western China. Finally, environmental decentralization can enhance GTFP by promoting public participation in environmental regulation. The findings of this paper have implications for adjusting environmental decentralization, environmental regulation policies, and formulating green economic transition and development strategies.

1. Introduction

Over a very long period in the past, China’s economic development was traded for environmental pollution and massive energy consumption. For the past few years, the government has tried its best to reverse the situation in which economic development, ecological, and environmental protection do not develop synergistically. In March 2020, the Central Committee of the Communist Party of China and the State Council issued the “Guidance on Building a Modern Environmental Governance System,” which proposes to improve the system of leadership and responsibility for environmental governance, enhance the decentralized governance mechanism of environmental governance between the central government and local governments, and increase the enthusiasm of market players and public participation. In November 2021, The Opinions on Deepening the Battle of Pollution Prevention and Control issued in November 2021 calls for continuing to strengthen the comprehensive management of the ecological environment. It deepens the battle of pollution prevention and control and strengthens ecological and environmental zoning control, with particular emphasis on fully understanding ecological and environmental publicity’s important status and role. The 2022 Government Work Report emphasizes the above, highlighting the government’s adoption of effective environmental decentralization and environmental regulation. The government’s determination to take practical environmental decentralization and regulation measures to promote eco-environmental protection and green economic development is highlighted in the 2022 government work report. Therefore, it is essential to explore the influence of environmental decentralization and environmental regulation on GTFP, and it has implications for the government to develop an environmental governance system with strong implementation and multiple participation.

There is extensive literature examining the relationship between environmental decentralization, environmental regulation, and GTFP. For example, Wu et al. (2020) show differences in the impacts of heterogeneous environmental regulation on provincial GTFP, where environmental decentralization and environmental administrative decentralization promote regional green development, while environmental supervision decentralization and environmental monitoring decentralization have a negative effect on it [1]. Moreover, Song et al. (2018) analyzed panel data for 273 cities in China and found that the effect of local environmental regulations on GTFP is influenced by the political attributes of cities [2]. Some studies support the contribution of each to GTFP [3,4], while others conclude that each has an inhibitory effect on GTFP [5,6]. In contrast, some studies show that the respective effects of both on GTFP are non-linearly determined [7,8]. However, existing studies do not agree on the relationship between these three factors. Environmental decentralization allows local governments to perform some environmental regulation functions as trustees, and both environmental decentralization and environmental regulation can affect GTFP, so it is necessary to clarify the mechanisms of their interactions.

The environmental decentralization system has given local governments discretionary space in environmental policy formulation and implementation, enhancing the flexibility of environmental regulation [9]. With the central government’s advocacy to shift from the pursuit of high economic growth to all-round development, local governments have subsequently shifted and upgraded to environmentally friendly industries. Environmental regulation as an effective restraint and control means to reduce environmental damage by economic behavior, through mandatory or non-mandatory ways to strengthen environmental awareness and clear social responsibility; local enterprises tend to choose to improve production processes and technologies, reduce the intensity of pollution emissions, and improve production efficiency [10]. In an environmental decentralization system, the increase in GTFP due to the intensity of environmental regulation depends on whether the “innovation compensation” effect is larger or smaller than the “compliance cost” effect. This is different from the mechanism by which environmental decentralization directly affects GTFP. Under China’s environmental decentralization system, within a reasonable interval of the degree of environmental decentralization, the advantages of local governments in terms of capital, talent, and technological innovation can be maximized to enhance the promotion of GTFP. However, the supply of public goods and services for environmental protection is allocated differently among local governments in different regions, and the development goals of local governments include not only environmental protection, but also economic development and tax revenue growth. There is a possibility of “bottom-up competition” between neighboring regions in order not to treat the environment so that neighboring regions can enjoy the benefits [11].

In summary, this paper tries to address the questions listed below, which are also the contributions of this paper: (1) What is the relationship between environmental decentralization, heterogeneous environmental regulations (command-and-control environmental regulations, market-incentive environmental regulations, and public participation in environmental regulations), and GTFP? (2) Is the influence of environmental decentralization on GTFP consistent across the east and west–central regions of China? (3) Can heterogeneous environmental regulation transmit the effects of environmental decentralization on GTFP?

The rest of the paper is organized as follows: Section 2 reviews the literature on relevant research directions. Section 3 elaborates on the measurement of GTFP. Section 4 takes on the model setting and variable descriptions. Section 5 takes on the empirical analysis, the findings, and relevant discussions. Section 6 summarizes the findings and provides policy implications.

2. Literature Review

2.1. Environmental Decentralization and Green Total Factor Productivity

Scholars’ research on decentralization first started with the influence of fiscal decentralization on the environment. Then, as the research progressed, researchers found inconsistencies between fiscal decentralization and environmental decentralization, thus evolving the branch of environmental decentralization based on fiscal federalism [12,13]. Under such a system, local governments not only play an important role in local economic development, but also play an irreplaceable role in ecological and environmental protection [14]. Scholars differ in their views on analyzing the role of environmental decentralization on GTFP.

There are three main reasons to support environmental decentralization to increase green total factor productivity: (1) Environmental decentralization will lead local governments to choose socially beneficial local environmental quality standards and public services from the perspective of considering the well-being of local residents, improving the matching of supply and demand, and thus increasing green total factor productivity [15,16]; (2) environmental decentralization will drive local governments to enact more rigorous environmental laws and regulations to give local governments the flexibility to respond to local conditions [17,18,19]; (3) environmental decentralization increases Research and Development (R&D) investment, which in turn increases green technological innovation capacity, optimizes industrial structure, generates the Porter effect, and thus promotes green total factor productivity [3]. However, it is worth noting that environmental decentralization should not be taken as a policy prescription but should consider the policy context and add constraints [20].

The main reasons against environmental decentralization to increase GTFP are as follows: (1) Environmental decentralization can lead to GDP-driven local governments, and local profit-seeking behavior is more likely to lead to tax competition mechanisms [21] and the phenomenon of “bottom-up competition,” thus reducing green total factor productivity [22]; (2) environmental decentralization is often linked to local government corruption, which can lead to collusion between local governments and enterprises. It has also connived local governments to tolerate highly polluting enterprises, thus resulting in inefficient environmental pollution control [1,5]; (3) the current lack of incentives for local government spending and the lack of autonomy of local environmental protection departments lead to environmental decentralization instead of enhancing environmental pollution [23]. Under China’s current cadre appraisal system and environmental decentralization system, the marginal effect of economic performance is much greater than environmental performance in the promotion process of local government officials [24].

However, some scholars argue that the influence of environmental decentralization on local governments is two-sided and cannot be generalized, e.g., Ran et al. (2020) suggest that it can lead to both “bottom-up competition” and “top-down competition,” depending on different forms of environmental decentralization [7]. Zou et al. (2019) and Shan et al. (2020) have similar conclusions [25,26].

2.2. Environmental Regulation and Green Total Factor Productivity

In order to minimize the negative impact on the natural environment [27], governments regulate the behavior of polluting the environment. The discussion on the impact of environmental regulation on GTFP is still not unified, and environmental regulation can effectively address the externalities of environmental pollution [10]. Current scholarly research on the impact of environmental regulation on GTFP focuses on the following three areas: the Porter hypothesis, the compliance cost hypothesis, and the non-linear relationship between environmental regulation and GTFP [28].

The first type of research supports Porter’s hypothesis that moderate environmental regulation will inspire technological innovation and offset the costs of technological innovation, thereby increasing GTFP, the “innovation compensation effect” [29,30]. Wang and Shen (2016) use the Global Malmquist Luenberger (GML) index to calculate industrial productivity in China by considering environmental factors, and environmental regulation and GTFP are positively correlated, which to some extent validates the Porter hypothesis [31]. Liu et al. (2020) suggest that a moderate level of environmental regulation can stimulate “innovation compensation effects”, which can stimulate the choice of differentiated strategies to improve resource efficiency and achieve cleaner production [32]. A recent study by Song et al. (2022) also validates Porter’s hypothesis by finding that green technological progress significantly increases GTFP by increasing unit labor productivity and that environmental regulations “force” green technological progress through external pressure [4].

The second type of research supports the cost-following hypothesis, also known by some scholars as the environmental regulation disincentive theory, where economies follow the “benefit maximization” principle [33]. However, environmental regulations add additional costs to economies by inhibiting technological innovation and resource allocation efficiency, hindering the development of green total factor productivity [6]. Hancevic’s (2016) study finds that total factor productivity declines due to environmental regulations ranging from 1% to 2.5% on average, with yield losses ranging from 1% to over 6% [34]. The results of Yuan and Xiang (2018) oppose the Porter hypothesis by finding that in the long-term view, environmental regulations crowd out R&D investments, inhibit patent output, and hinder the development of GTFP [35].

The third category of studies argues that the connection between environmental regulation and GTFP is non-linear and that the effect of local environmental regulation on GTFP is influenced by the political attributes of cities and the choice of heterogeneous environmental regulation policy instruments [36,37]. Sanchez et al. (2013) find that the link between environmental regulation and productivity in Mexico is non-linear, which weakens in the manufacturing sector [38]. The empirical results of Qiu et al. (2021) showed a U-shaped association between environmental regulations and GTFP, while China is still in the left half of the U-shaped curve [8].

2.3. Environmental Decentralization, Environmental Regulation, and Green Total Factor Productivity

There is no shortage of the current literature examining the relationship between environmental decentralization and environmental regulation [9,39,40]. Under dual pressure to protect the environment and maintain economic growth rates, the central government needs to decide how much power to delegate to local governments [41]. The 21st century, known as the “green economy development” period, has witnessed new developments and extraordinary advances in technology [42]; environmental laws and regulations and ecological protection policies and measures introduced by local governments will influence local production and life activities; total factor productivity, as the only sustainable force in the new normal [43], will link environmental decentralization, environmental regulation, and GTFP; many social problems will be solved [44]. In an in-depth study, researchers have broken down the effects of heterogeneous environmental decentralization and heterogeneous environmental regulation on GTFP, respectively [1,20,28], and different types of environmental regulations may differ significantly in their environmental objectives. Ignoring the heterogeneity of environmental regulations may lead to bias between environmental regulations and policy impact assessment [45]. Meanwhile, many studies have found regional heterogeneity in the effects of both on GTFP [3]. Wu et al. (2020) examined the impact of environmental regulation on green total factor energy efficiency under different environmental decentralization [46]. However, fewer studies have included environmental decentralization, environmental regulation, and GTFP in the same framework. Ghosal et al. (2019) researched the influence of environmental decentralization, environmental regulation, and enforcement policies on mill-level GTFP growth in the pulp and paper industry in Sweden. Their findings indicated that Sweden’s decentralization and special environmental regulation had a positive effect [47]. Chen and Zhao (2020) verify the effectiveness of local government environmental decentralization in regulating environmental regulations to promote GTFP based on provincial panel data in China from 2001–2015 [48].

In summary, (1) firstly, the paper finds that the past literature has not clarified the relationship between environmental decentralization, environmental regulation, and GTFP. (2) Secondly, this paper finds that the past literature has not considered the relationship between environmental decentralization and GTFP under the effect of heterogeneous environmental regulations. (3) Finally, this paper finds that the measurement of GTFP in China is concentrated before 2016, and the mechanism of action transmission between environmental decentralization, environmental regulation, and GTFP may have changed over the past few years as the Chinese government’s environmental governance system has improved.

3. Measurement and Analysis of Green Total Factor Productivity

3.1. Measurement of Green Total Factor Productivity

The Data Envelopment Analysis (DEA) method does not need to set the production function, but the traditional DEA method has the problem of input and output slackness. Tone (2001) proposed the SBM model to remove the inefficiency caused by the slackness [49]. This paper selects the panel data of 30 Chinese provinces (excluding Tibet, Hong Kong, Macao, and Taiwan) from 2010 to 2020. It uses MaxDEA 8 Ultra software to measure the Malmquist index applying the Super-SBM model considering non-desired output, variable scale payoffs, and the measure results in the growth rate of GTFP [50]. If the Malmquist index is greater than 1, the GTFP increases compared to the previous year, and vice versa, it decreases. It is assumed that the GTFP in 2010 is 1, and the GTFP in 2011 is the GTFP in 2010, which is multiplied by the corresponding Malmquist index, etc.

The input indicators involved in GTFP mainly include labor input, capital input, and energy input, and the output indicators mainly include desired output and non-desired output:

Labor input. The optimal labor input should be standard labor hours, but due to data availability, the literature on GTFP in China has almost always considered only the number of workers [51]. This paper uses the number of employed persons in the whole society at the end of the year in each region as an indicator, calculated by adding up the number of urban units, private enterprises, and individuals employed at the end of the year in each region.

Capital inputs. Using the actual annual physical capital stock of each region as an indicator, Shan (2008) estimated the capital stock in China using the Perpetual Inventory Method (PIM) [52]. Chinese scholars have developed a consistent system of benchmark capital stock and depreciation rates, and the estimation formula is as follows:

$${\mathrm{K}}_{\mathrm{t}}=\frac{{\mathrm{I}}_{\mathrm{t}}}{{\mathrm{P}}_{\mathrm{t}}}{+1 - \mathsf{\delta } \mathrm{K}}_{\mathrm{t}-1}$$

(1)

where K_t is the actual physical capital stock in year t; I_t is the current investment in year t. In this paper, we use total fixed capital formation as the data for investment, and due to missing data, we use the growth of fixed-asset investment over the previous year to calculate the investment data in 2018 and 2019. P_t is the price index of fixed asset investment in year t. δ is the depreciation rate of fixed assets. Compared to the Zhang et al. (2004) study that chooses 9.6% [53], the Shan (2008) study takes 10.96%, which is more in line with the trend of economic development in the new era. The earlier the base period is chosen, the less the error is arising from the estimation of capital stock in the base period, which affects the subsequent years [52]. The year 2000 is selected, and the fixed capital stock in 2000 is adopted as the sum of the average depreciation rate of 10.96% over the actual capital formation in 2001. In each province and the average investment, growth rate between 2001 and 2005 is found with the following formula:

$${\mathrm{K}}_1=\frac{{\mathrm{I}}_2}{\mathrm{g}+\mathsf{\delta }}$$

(2)

Energy input. Use the total annual energy consumption of each region as an indicator.

Expected output. Using each region’s annual real GDP as an indicator and using 2000 as the base period, the real GDP for 2001 is multiplied by the GDP of 2000 by the GDP index. Then, the real GDP for 2002 is multiplied by the real GDP of 2001 by the GDP index, and so on.

Non-desired output. The measurement of non-desired outputs in the different literature is broadly classified as follows. The first one uses total carbon emissions to indicate non-desired outputs [36]. The second one uses some or all industrial wastewater emissions, industrial waste gas emissions, industrial solid pollutant generation, SO₂ emissions, chemical oxygen demand emissions, soot emissions, and CO₂ emissions as indicators [1,35]. The third one uses the principal component or entropy method to calculate the integrated pollution index as an indicator [54,55]. Since the first two methods cannot reflect the different weights of pollutant emissions and have the problem of inconsistency in the scale, this paper adopts the entropy value method to measure the comprehensive environmental pollution index. The use of industrial chemical oxygen demand emissions as an indicator of industrial wastewater emissions, industrial sulfur dioxide emissions, industrial soot emissions as an indicator of industrial waste gas emissions, and industrial solid waste generation as an indicator of industrial waste emissions. The weights of industrial chemical oxygen demand emissions, industrial sulfur dioxide emissions, industrial smoke, dust emissions, and industrial solid waste generation in environmental pollution are calculated separately, and then the comprehensive environmental pollution index is derived with the following specific steps: (1) dimensionless processing of data to avoid the influence of different levels of indicators on the calculation results,

$${\mathrm{P}}_{\mathrm{ij}}=\frac{{\mathrm{X}}_{\mathrm{ij}}}{{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{X}}_{\mathrm{ij}} }$$

(3)

where P_ij is the specific gravity of the jth indicator of the ith region in the sum of the jth indicator of all regions, X_ij are the initial data, m represents the total number of regions, i.e., 30; (2) calculating the entropy value of each indicator,

$${\mathrm{e}}_{\mathrm{j}}=- \frac{1}{\mathrm{lnm}}{{\displaystyle \sum }}_{\mathrm{i}=1}^{\mathrm{m}}{\mathrm{P}}_{\mathrm{ij}}{\mathrm{lnP}}_{\mathrm{ij}}$$

(4)

where e_j is the entropy value of the jth indicator; (3) calculating the weight of each indicator,

$${\mathrm{W}}_{\mathrm{j} }=\frac{{1 - \mathrm{e}}_{\mathrm{j}}}{{{\displaystyle \sum }}_{\mathrm{j}=1}^{\mathrm{n}}{1 - \mathrm{e}}_{\mathrm{j}}}$$

(5)

where n represents the total number of indicators, i.e., 4; (4) calculating the comprehensive index of environmental pollution,

$${\mathrm{F}}_{\mathrm{i}}={\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{j}}{\mathrm{X}}_{\mathrm{ij}}$$

(6)

3.2. Measurement Results and Analysis

The GTFP of 30 provinces from 2011 to 2020 was measured by the method described in the previous section, and the above provinces were divided according to the eastern and western regions, and the green total factor productivity distribution maps (Figure 1) for 2011 and 2020 were drawn using ArcGIS 10.4.1 software. From

Table 1. GTFP in 30 Chinese provinces, 2011–2020.

Group	Province	GTFP
Eastern Region	Beijing	1.442
	Tianjin	0.871
	Hebei	1.118
	Liaoning	1.103
	Shanghai	1.168
	Jiangsu	1.151
	Zhejiang	1.056
	Fujian	0.939
	Shandong	0.788
	Guangdong	0.975
	Hainan	0.593
	Mean	1.019
Central Region	Shanxi	0.942
	Jilin	1.070
	Heilongjiang	0.986
	Anhui	0.567
	Jiangxi	0.968
	Henan	0.995
	Hubei	0.934
	Hunan	1.021
	Mean	0.935
Western Region	Inner Mongolia	0.903
	Guangxi	0.933
	Chongqing	1.078
	Sichuan	1.077
	Guizhou	0.959
	Yunnan	0.951
	Shaanxi	0.904
	Gansu	0.948
	Qinghai	0.435
	Ningxia	0.611
	Xinjiang	0.854
	Mean	0.878
National		0.945

, it appears that the overall level of GTFP development in the country has risen and fallen, with an upward trend in the central and eastern regions and a downward trend in the western region. It is shown that the introduction and implementation of environmental laws and regulations have been practical, and China’s economy is being transformed and upgraded to a resource-saving and environment-friendly one. From Figure 2, we can see that no matter the GTFP in the east, central, west, or the whole country, it shows a decline and then an increase in the period of 2010–2020. To analyze the reasons for this, the government’s 12th and 13th Five-Year Plans for environmental protection have strengthened the assessment of energy conservation and environmental protection objectives, improved the market-based mechanism for energy conservation and environmental protection, and implemented an effective mechanism for local governments to impose constraints and incentives in parallel [50]. However, in the early stages of policy implementation, there is a buffer period to balance the synergy between environmental protection and economic growth as local governments need some time to change their economic development. China’s GTFP shows a stepwise distribution from east–central–west, with only the eastern region having a GTFP greater than 1. Considerations may be due to differences in centrally delegated environmental rights and different development goals across regions, with the eastern region pursuing economic benefits more than environmental management, while the western region’s most pressing goal is ecological management.

4. Materials and Methods

4.1. Model Setting

The basic econometric model constructed to test the influence of GTFP by environmental decentralization and environmental regulation is as follows:

GTFP_it = α + β₁ED_it + β₂ER_jit +β₃Z_it + ε_it

(7)

where i is the region and t is the year, GTFP is green total factor productivity, ED is environmental decentralization, ER_j represents command-and-control (CER), market-incentivized (MER), and public participation environmental regulation (PER), Z_it represents control variables, and ε_it is an error term.

To test the indirect role of environmental regulation in the channel between environmental decentralization and GTFP, this paper constructs a model with environmental regulation as a mediating variable. It draws on the mediation test procedure combining the sequential test and Sobel test, proposed by Wen et al. (2004) [56]; this method has the advantage of being easy to calculate, simple, and feasible, so it is widely used:

GTFP_it = a₀ + α₁ED_it + δ₁Z_it + ε_it

(8)

ER_it = b₀ + β₁ED_it + δ₂Z_it + ε_it

(9)

GTFP_it = c0 + φ₁ED_it + φ₂ER_it + δ₃Z_it + ε_it

(10)

4.2. Variable Selection and Data Description

4.2.1. Core Explanatory Variables

Environmental decentralization (ED). This paper has used the ratio of the number of people in local environmental protection systems to the total number of people in national environmental protection systems, which is an indicator of the intensity of environmental decentralization. By calculating the ratio of the number of people in local environmental protection systems to the total number of people in national environmental protection systems, an economic scale proportionality factor is added to correct for errors that may result from economic differences between regions [46]. The specific advantages of this measurement method are the following three points: the first point is that institutions and personnel are the carriers of government operations [57], and the number of personnel in the local government environmental protection system directly reflects the degree of decentralization of environmental authority and the ability to fulfill its responsibilities. The second point is that the overall size of the environmental protection sector is stable and that changes in the personnel of the environmental protection system can reflect the changes and developments in the decentralized environmental management system [1]. The third point is that environmental decentralization is essentially management decentralization [46], and, therefore, the allocation of personnel is more in line with the logic of environmental decentralization and management than the allocation of financial expenditures. The formula is as follows:

$${\mathrm{ED}}_{\mathrm{it}}=\left[ \frac{{\mathrm{LEPP}}_{\mathrm{it}}/{\mathrm{POP}}_{\mathrm{it}}}{{\mathrm{NEPP}}_{\mathrm{t}}/{\mathrm{POP}}_{\mathrm{t}}} \right] \times \left[ 1 -\left({ \mathrm{GDP}}_{\mathrm{it}} /{ \mathrm{GDP}}_{\mathrm{t}} \right) \right]$$

(11)

where ED_it is the environmental decentralized intensity in region i in year t, LEPP_it is the number of urban units in water, environment, and public facility management in region I in year t, POP_it is the population size in region i in year t, NEPP_t is the number of urban units in water, environment, and public facility management nationwide in year t, POP_t is the population size nationwide in year t, and GDP_it and GDP_t are GDP in region i in year t and GDP nationwide in year t, respectively.

Environmental regulation (ER). Command-and-control environmental regulation (CER) is the government’s use of coercive means, such as environmental laws and regulations and environmental standards, to impose environmental restrictions on production and life in the region [58]. This paper applies the number of environmental administrative penalty cases received by each region in the current year as an indicator, and the missing data years are filled in with the environmental administrative penalty cases in the current year notified by the Ministry of Ecology and Environment of China. Market-incentive environmental regulation (MER) is a non-coercive environmental regulation tool, and the revenue from emission fees has a catalytic effect on the environmental regulation of each local government [59], using the discharge fees unpaid into the treasury of each region in the current year as an indicator. It should be noted, in particular, that the Environmental Protection Tax Law of the People’s Republic of China has been implemented since 1 January 2018, and the emission fees have been changed to environmental protection tax; public participation-based environmental regulation (PER) uses the current year of each region The numbers of the National People’s Congress (NPC) proposal are undertaken as indicators.

4.2.2. Control Variables

Considering the potential influencing factors of GTFP, this study, therefore, adopts the level of economic development, the non-linear influence of the level of economic development, foreign direct investment, industrial structure, infrastructure level, local financial autonomy, human capital quality, and scientific and technological capital investment as control variables. Economic development level (PGDP): the environmental Kuznets curve predicts that the level of environmental pollution is associated with the level of economic development [60,61], so the GDP per capita of each region is selected as an indicator. Non-linear impact of the level of economic development (PGDP2): adding the square of GDP per capita to measure the non-linear impact of the level of economic development on GTFP. Foreign direct investment (FDI): on the one hand, the technology spillover effect of FDI facilitates the introduction of advanced technology and management methods, thus promoting GTFP, while on the other hand, the “pollution haven hypothesis” suggests that developed countries or regions transfer environmental pollution to low-income regions through FDI, resulting in GTFP losses [62]. Therefore, this paper adopts the actual annual use of foreign direct investment by region to characterize. Industrial structure (IS): there are significant differences in energy consumption, pollution emission levels, and input–output ratios among industries, and the transformation and upgrading of primary and secondary industries to tertiary industries may have a significant impact on GTFP [63], measured using the annual share of secondary sector output in GDP for each region. Infrastructure (INFRA): infrastructure, as a non-competitive and non-exclusive public project, determines the leading role of the government in the construction process [64] and the level of infrastructure which affects the development of GTFP. Therefore, the area of road occupation per capita is used as an indicator of the level of infrastructure. Fiscal decentralization (FD): the degree of fiscal decentralization determines the local government’s disposable finances, which may have a positive contribution to local green TFP development or may hinder local green TFP development by reducing environmental investments due to local government competition [2]; therefore, using the ratio of fiscal budget revenue to fiscal budget expenditure measures. Human capital quality (HR): human capital can influence GTFP by promoting technological progress and facilitating knowledge “spillovers,”. It promotes industrial upgrading and raising awareness of green development [60]. This paper uses the average years of schooling by region to measure the level of human capital [65,66,67]. There are three reasons for adopting this method: first, it has easy operability. Second, the average years of education of the population visually reflects the quality of regional labor resources, which is an internationally accepted indicator of human resource effectiveness and innovation capacity. Third, although the average years of education of the population is subject to different education systems [68], China’s education systems are similar and the statistical caliber provided in the China Statistical Yearbook is consistent. Referring to the formula for the average years of education for the population, as specified in the China Education Monitoring and Evaluation Statistical Indicator System (2020 Edition) issued by the Chinese Ministry of Education, the years of education at each level of primary, middle, high school, and college and above in China are 6, 9, 12, and 16 years, respectively. The calculation is as follows:

$${\mathrm{HR}}_{\mathrm{I}}=\frac{\left(16 \times \mathrm{College} \mathrm{and} \mathrm{above}+12 \times \mathrm{High} \mathrm{School}+9 \times \mathrm{Junior} \mathrm{High} \mathrm{School}+6 \times \mathrm{Primary} \mathrm{School}\right)}{\mathrm{Population} 6 \mathrm{years} \mathrm{old} \mathrm{and} \mathrm{above}}$$

(12)

Capital investment in science and technology (TI): government support for the science and technology sector enhances technological progress, energy efficiency, and regional green economy development [1], so the share of investment in education and science and technology in GDP of each region is used as an indicator.

4.2.3. Data Description

Considering data availability, this paper excludes Tibet, Hong Kong SAR, Macau SAR, and Taiwan Province, and counts the relevant data of 30 provinces and cities. Labor input, capital input, desired output, environmental decentralization intensity, economic development level, industrial structure, infrastructure level, local financial autonomy, human capital quality, and technological innovation are from China Statistical Yearbook 2010–2020, energy input is from China Energy Statistical Yearbook 2010–2020, the non-desired output is from China Environmental Statistical Yearbook, command-and-control environmental regulation (CER), market-incentive environmental regulation (MER), and public participation environmental regulation (PER) are from China Environmental Yearbook 2010–2020, and foreign direct investment is from China Business Yearbook 2010–2020 and the annual national economic and social development statistical bulletin of each province. The results of descriptive statistics are reported in

Table 2. Descriptive statistics.

	Variable Symbol	N	Mean	SD	Min	Max
Explained Variable	GTFP	330	0.950	0.220	0.328	2.118
Core Explanatory Variable	ED	330	1.134	0.474	0.503	3.201
	CER	330	4460	6323	32	45,140
	MER	330	68,769	60,240	2849	358,900
	PER	330	219.406	173.928	11	1196
Control Variable	PGDP	330	50,510	26,445	10,309	164,220
	PGDP2	330	3.248 × 109	3.784 × 109	1.063 × 108	2.697 × 1010
	FDI	330	5,148,702	4,944,993	2949	2.257 × 107
	IS	330	0.447	0.086	0.162	0.590
	INFRA	330	15.054	4.666	4.040	26.200
	FD	330	0.502	0.192	0.148	0.931
	HR	330	9.104	0.915	6.766	12.681

.

5. Results and Discussion

5.1. Basic Regression Results

To achieve the scientific validity of the model selection, this paper applies a fixed-effects model for the baseline regression (Prob>chi2 = 0.0001). As shown in model (1) of

Table 3. Basic regression results.

Variables	(1) GTFP	(2) GTFP	(3) GTFP	(4) GTFP
ED	−0.088 *** (−3.09)	−0.088 *** (−3.11)	−0.079 *** (−2.83)	−0.051 * (−1.68)
CER		0.00000388 ** (2.45)
MER			0.000000635 *** (3.60)
PER				0.0002177 *** (3.08)
PGDP	−0.000 *** (−4.57)	−0.000 *** (−4.70)	−0.000 *** (−3.66)	−0.000 *** (−5.04)
PGDP2	0.000 *** (6.17)	0.000 *** (6.35)	0.000 *** (5.57)	0.000 *** (6.70)
FDI	0.000 *** (5.11)	0.000 *** (4.13)	0.000 *** (3.68)	0.000 *** (3.93)
IS	−0.671 *** (−4.95)	−0.635 *** (−4.69)	−0.654 *** (−4.91)	−0.583 *** (−4.26)
INFRA	−0.013 *** (−5.66)	−0.013 *** (−5.88)	−0.016 *** (−6.66)	−0.013 *** (−5.80)
FD	−0.377 *** (−4.46)	−0.379 *** (−4.51)	−0.341 *** (−4.07)	−0.391 *** (−4.67)
HR	0.067 *** (3.03)	0.067 *** (3.03)	0.052 ** (2.37)	0.072 *** (3.28)
TI	−5.929 *** (−6.77)	−5.784 *** (−6.64)	−5.212 *** (−5.90)	−5.468 *** (−6.23)
Constant	1.476 *** (6.68)	1.463 *** (6.66)	1.511 *** (6.96)	1.329 *** (5.95)
Observations	330	330	330	330
R-squared	0.513	0.522	0.532	0.527
F test	37.381	34.767	36.200	35.490

Note: t-statistics in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1.

, the coefficient of environmental decentralization is −0.088 and significant at the 1% level, implying that environmental decentralization has a depressing effect on GTFP, which is different from the conclusions of previous studies, where the GTFP decreases in regions where the intensity of environmental decentralization increases. The possible reason for this result is that due to information asymmetry, the central government tends to use GDP, for example, as an explicit criterion in its assessment of local governments [69]. Although the state is also strengthening the assessment of local governments’ environmental performance, when economic performance takes priority, local governments will use their higher governance autonomy to pursue the principle of GDP supremacy [70], thus creating a serious “bottom-up competition” effect on the environment [71].

The types of environmental regulations in Equations (2)–(4) in Table 3 are command-and-control (CER), market-incentive (MER), and public participation environmental regulations (PER), and all three environmental regulations are significantly and positively correlated with GTFP, indicating that as the degree of environmental regulations increases, GTFP increases, which verified the conclusion of Zhang et al. (2011) [72]. The promotion of GTFP by command-and-control environmental regulation (CER) may be due to the relatively complete environmental laws, regulations, and related standards in China. Meanwhile, the promotion of GTFP by market-inspired environmental regulation is mainly because market-based instruments stimulate technological innovation and contribute to the optimal allocation of innovation resources [73]. The public’s participation and environmental protection supervision motivate them to raise and report problems that harm their rights and interests, reducing environmental pollution and promoting green and sustainable economic and social development. The public can be the most direct, effective, and widely distributed regulator of the government’s ecological performance [74]. This paper also finds that the three effects on green TFP differ in magnitude, with public participation in environmental regulation having the most considerable effect, followed by command-and-control environmental regulation (CER). Market-incentive environmental regulation (MER) has the smallest effect. Compared with the other two environmental regulations, the environmental regulation based on the autonomous market regulation mechanism has a long response time to the behavior of economic agents. It lacks coercion on economic agents, so its impact is relatively small.

5.2. Heterogeneity Test

To investigate whether there is regional heterogeneity of environmental decentralization on GTFP, this paper divides three samples for regression in the east, west, and central.

Table 4. Heterogeneity test results.

Variables	Eastern Region GTFP			Central Region GTFP			Western Region GTFP
ED	−0.133 ** (−2.38)	−0.090 * (−1.71)	−0.123 ** (−2.27)	−0.036 (−0.63)	−0.047 (−0.85)	−0.096 (−1.55)	0.065 (1.56)	0.072 * (1.67)	0.079 * (1.76)
CER	0.00000585 ** (2.41)			−0.00000265 (−1.55)			0.0000279 *** (3.48)
MER		0.0000019 *** (5.14)			−0.00000078 *** (−3.41)			0.00000118 ** (2.53)
PER			0.0003741 *** (3.75)			−0.0002963 ** (−2.25)			0.000244 * (1.75)
PGDP	−0.000 ** (−2.10)	−0.000 (−0.42)	−0.000 *** (−2.92)	−0.000 *** (−2.84)	−0.000 *** (−3.37)	−0.000 *** (−2.72)	−0.000 *** (−5.46)	−0.000 *** (−4.55)	−0.000 *** (−5.23)
PGDP2	0.000 *** (3.21)	0.000 * (1.84)	0.000 *** (3.93)	0.000 *** (3.40)	0.000 *** (3.84)	0.000 *** (3.46)	0.000 *** (3.58)	0.000 *** (2.66)	0.000 *** (3.70)
FDI	0.000 ** (2.34)	0.000 (0.29)	0.000 *** (2.67)	0.000 *** (3.17)	0.000 *** (2.89)	0.000 *** (3.78)	0.000 ** (2.08)	0.000 *** (3.65)	0.000 ** (2.30)
IS	−1.104 *** (−2.90)	−1.162 *** (−3.39)	−0.978 *** (−2.65)	0.055 (0.22)	−0.334 (−1.26)	−0.204 (−0.75)	−1.420 *** (−6.56)	−1.723 *** (−8.56)	−1.622 *** (−7.57)
INFRA	−0.013 *** (−3.13)	−0.024 *** (−5.55)	−0.013 *** (−3.21)	−0.026 *** (−3.53)	−0.034 *** (−4.58)	−0.026 *** (−3.60)	−0.013 *** (−2.80)	−0.016 *** (−3.67)	−0.017 *** (−3.69)
FD	−0.594 *** (−3.16)	−0.412 ** (−2.32)	−0.721 *** (−4.03)	−0.896 *** (−3.53)	−0.836 *** (−3.48)	−0.523 * (−1.77)	0.560 *** (3.55)	0.280 * (1.69)	0.375 ** (2.33)
HR	0.116 *** (2.75)	0.048 (1.16)	0.165 *** (3.91)	0.216 *** (3.73)	0.226 *** (4.11)	0.214 *** (3.76)	0.079 *** (2.91)	0.100 *** (3.76)	0.114 *** (4.27)
TI	−16.385 *** (−4.68)	−13.049 *** (−3.95)	−15.432 *** (−4.56)	3.821 * (1.90)	6.179 *** (3.04)	1.469 (0.66)	−3.438 *** (−3.03)	−4.218 *** (−3.72)	−3.945 *** (−3.37)
Constant	1.717 *** (3.58)	2.020 *** (4.61)	1.281 *** (2.64)	−0.112 (−0.22)	0.123 (0.25)	0.028 (0.05)	1.369 *** (6.13)	1.451 *** (6.33)	1.295 *** (5.39)
Observations	121	121	121	88	88	88	121	121	121
R-squared	0.611	0.670	0.637	0.693	0.725	0.703	0.792	0.781	0.775
F test	17.29	22.32	19.32	17.38	20.30	18.22	41.76	39.28	37.84

Note: t-statistics in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.

reports the regression results, and the results show that the eastern region is consistent with the previous results. However, the impact of the central region is not obvious, and environmental decentralization in the western region can contribute to the development of GTFP. For developed regions, the improvement of environmental technology and the effective reduction in environmental pollution can lead to the reduction in human capital investment. It leads to a lower degree of environmental decentralization [75].

In contrast, for less developed regions, total factor productivity is more severe for environmental pollution and the degree of environmental governance efficiency is backward. It still needs a considerable number of environmental protection system personnel to complete the environmental protection work, and thus the high cost of environmental protection governance. Moreover, the pressure on environmental protection in the western region is greater than the assessment of economic performance. It is more strictly implemented and uses the central government’s environmental governance authority, which has a stronger incentive for local environmental protection and green economy development [76].

5.3. Intermediate Effect Test

This paper investigates the three environmental regulations as mediating variables in the action path between environmental decentralization and GTFP.

Table 5. Results of intermediate effect test.

Variables	(5) GTFP	(6) CER	(7) GTFP	(8) GTFP	(9) MER	(10) GTFP	(11) GTFP	(12) PER	(13) GTFP
ED	−0.088 *** (−3.09)	32.977 (0.03)	−0.088 *** (−3.11)	−0.088 *** (−3.09)	−13,500.4 (−1.52)	−0.079 *** (−2.83)	−0.088 *** (−3.09)	−168.0 *** (−7.54)	−0.051 * (−1.68)
ER			0.000 ** (2.45)			0.000 *** (3.60)			0.000 *** (3.08)
PGDP	−0.000 *** (−4.57)	0.038 (0.67)	−0.000 *** (−4.70)	−0.000 *** (−4.57)	−2.175 *** (−4.40)	−0.000 *** (−3.66)	−0.000 *** (−4.57)	0.003 *** (2.68)	−0.000 *** (−5.04)
PGDP2	0.000 *** (6.17)	−0.000 (−1.07)	0.000 *** (6.35)	0.000 *** (6.17)	0.000 *** (3.13)	0.000 *** (5.57)	0.000 *** (6.17)	−0.000 *** (−3.24)	0.000 *** (6.70)
FDI	0.000 *** (5.11)	0.001 *** (5.84)	0.000 *** (4.13)	0.000 *** (5.11)	0.006 *** (6.44)	0.000 *** (3.68)	0.000 *** (5.11)	0.000 *** (6.00)	0.000 *** (3.93)
IS	−0.671 *** (−4.95)	−9238.9 * (−1.95)	−0.635 *** (−4.69)	−0.671 *** (−4.95)	−25,915.7 (−0.61)	−0.654 *** (−4.91)	−0.671 *** (−4.95)	−403.5 *** (−3.81)	−0.583 *** (−4.26)
INFRA	−0.013 *** (−5.66)	114.838 (1.46)	−0.013 *** (−5.88)	−0.013 *** (−5.66)	4642.9 *** (6.64)	−0.016 *** (−6.66)	−0.013 *** (−5.66)	0.740 (0.42)	−0.013 *** (−5.80)
FD	−0.377 *** (−4.46)	409.330 (0.14)	−0.379 *** (−4.51)	−0.377 *** (−4.46)	−57,489 ** (−2.18)	−0.341 *** (−4.07)	−0.377 *** (−4.46)	61.415 (0.93)	−0.391 *** (−4.67)
HR	0.067 *** (3.03)	98.218 (0.13)	0.067 *** (3.03)	0.067 *** (3.03)	22,878 *** (3.32)	0.052 ** (2.37)	0.067 *** (3.03)	−21.642 (−1.25)	0.072 *** (3.28)
TI	−5.929 *** (−6.77)	−37,554.2 (−1.22)	−5.784 *** (−6.64)	−5.929 *** (−6.77)	−1,129,571 *** (−4.14)	−5.212 *** (−5.90)	−5.929 *** (−6.77)	−2118 *** (−3.09)	−5.468 *** (−6.23)
Constant	1.476 *** (6.68)	3577.306 (0.46)	1.463 *** (6.66)	1.476 *** (6.68)	−54,979.2 (−0.80)	1.511 *** (6.96)	1.476 *** (6.68)	676.73 *** (3.92)	1.329 *** (5.95)
R-squared	0.513	0.274	0.522	0.513	0.368	0.532	0.513	0.522	0.527
F test	37.38 ***	13.42 ***	34.77 ***	37.38 ***	20.68 ***	36.20 ***	37.38 ***	38.76 ***	35.49 ***
Sobel	0.000 (Z = 0.033)			−0.009 (Z = −1.402)			−0.037 *** (Z = −2.855)
Goodman−1	0.000 (Z = 0.031)			−0.009 (Z = −1.358)			−0.037 *** (Z = −2.834)
Goodman−2	0.000 (Z = 0.036)			−0.009 (Z = −1.450)			−0.037 *** (Z = −2.877)
Mediating effect	0.000 (Z = 0.033)			−0.009 (Z = −1.402)			−0.037 *** (Z = −2.855)
Direct effect	−0.088 ** (Z = −3.114)			−0.079 *** (Z = −2.826)			−0.051 * (Z = −1.683)
Total effect	−0.088 ** (Z = −3.085)			−0.088 *** (Z = −3.085)			−0.088 *** (Z = −3.085)
Proportion of total effect that is mediated	−0.001			0.098			0.416

Note: t-statistics in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.

reports the results of the tests of mediating effects. First, whether environmental decentralization affects GTFP and the coefficient of environmental decentralization in Equations (5), (8) and (11) is significantly negative at the 1% level, allowing for the next test.

Next, the effects of environmental decentralization on heterogeneous environmental regulation are studied. The coefficients of environmental decentralization in Equations (6) and (9) are not significant, while the coefficient of environmental decentralization in Equation (12) is significant, indicating that the higher the degree of environmental decentralization, the higher the degree of public participation in environmental regulations. Next, the effects of environmental decentralization and the three environmental regulations on GTFP are examined, and the three environmental regulations in Equations (7), (10) and (13) are all significantly and positively correlated with GTFP.

Lastly, the Goodman-1 value of public participation in environmental regulation is significant in the Sobel test results of the three environmental regulations. The proportion of mediating effect is 41.581352%, indicating that environmental decentralization can suppress GTFP through the intermediary channel of public participation-based environmental regulation, and its influence path is “environmental decentralization—public participation environmental regulation—GTFP.” The influence path is “environmental decentralization-public participation environmental regulation-GTFP.” The National Environmental Protection 12th Five-Year Plan and the National Environmental Protection 13th Five-Year Plan emphasize the need to mobilize the whole society to take part in environmental protection [75,77], and environmental decentralization improves the public’s perception of government performance and thus increases the motivation to participate in environmental regulation. It, in turn, enhances the development of GTFP [20].

5.4. Robustness Test

Firstly, the three control variables of ownership structure (OS), technological innovation (TEI), and energy consumption structure (ECT) are added to reduce the problem of inaccurate results due to omitted variables [78]. As shown in model (14) in

Table 6. Robustness test results.

Variables	(14) GTFP	(15) GTFP	(16) GTFP
ED	−0.115 *** (−3.84)	−0.093 *** (−3.34)	−0.083 *** (−2.94)
PGDP	−0.000 ** (−2.16)	−0.000 *** (−4.16)	−0.000 *** (−4.57)
PGDP2	0.000 *** (4.31)	0.000 *** (5.34)	0.000 *** (6.16)
FDI	0.000 *** (4.33)	0.000 *** (5.20)	0.000 *** (5.26)
IS	−0.918 *** (−6.08)	−0.662 *** (−5.04)	−0.671 *** (−4.94)
INFRA	−0.015 *** (−6.38)	−0.013 *** (−5.88)	−0.013 *** (−5.63)
FD	−0.346 *** (−4.10)	−0.374 *** (−4.59)	−0.373 *** (−4.40)
HR	0.056 ** (2.53)	0.061 *** (2.83)	0.065 *** (2.94)
TI	−6.407 *** (−7.45)	−5.949 *** (−6.93)	−5.965 *** (−6.80)
OS	0.513 *** (3.21)
TEI	0.000 ** (2.03)
ECT	0.087 ** (2.39)
Constant	1.450 *** (6.62)	1.516 *** (6.96)	1.488 *** (6.73)
Observations	330	330	330
R-squared	0.546	0.500	0.511
F test	31.79	35.50	37.19

Note: t-statistics in parentheses, *** p < 0.01, ** p < 0.05.

, the positive and negative correlations between the core explanatory variable environmental decentralization and GTFP remain unchanged and significant at the 1% level after adding the above three control variables. Secondly, Wang et al. (2022) [79] has drawn to apply a 1% two-way tail reduction and two-way truncation to each variable to reduce the effect of extreme values or outliers on the regression results, and model (15) shows that it is generally consistent with the baseline regression results. Finally, in order to verify the robustness of environmental decentralization indicators, this paper uses environmental decentralization as the core explanatory variable after removing the economic scale proportionality factor [80], and the significance and sign of each variable do not change in the results of model (16), confirming that the indicator selection is robust.

6. Conclusions and Policy Implications

This paper examines the relationship between environmental decentralization, heterogeneous environmental regulations, and GTFP using various indicators for 30 Chinese provinces and cities (excluding Tibet, Taiwan, Macau, and Hong Kong) from 2010 to 2020. Stepwise and Sobel tests are used to verify the mediating effect of heterogeneous environmental regulations, with the following main findings: (1) for the country as a whole, environmental decentralization has a depressing effect on GTFP. Command-and-control (CER), market-incentive (MER), and public participation environmental regulations (PER) all contribute to GTFP, with public participation environmental regulations (PER) contributing the most, command-and-control environmental regulations (CER) the second most, and market-incentive environmental regulations (MER) the least. (2) The role of environmental decentralization varies across regions in China, with environmental decentralization having a depressing effect on GTFP in the east, a non-significant effect in the center, and a facilitating effect in the west. (3) Command-and-control (CER) and market-incentive environmental regulations (MER) do not have intermediary effects in the influence channel of environmental decentralization and GTFP. In contrast, public participation in environmental regulations has partial intermediary effects, the proportion of their intermediary effects is 41.6%, and the path of action is environmental decentralization → public participation environmental regulations → GTFP.

According to the above research conclusions, the policy recommendations of this paper are mainly summarized as follows: (1) heterogeneous environmental decentralization policies are implemented according to local conditions and regional synergy is strengthened. The findings of this paper show that the impact of environmental decentralization on GTFP is heterogeneous across the eastern and western regions of China. The central government should fully consider the differences in economic development levels and ecological environmental protection pressures among the eastern and western regions and formulate precise environmental decentralization policies. The eastern region should appropriately reduce the degree of environmental decentralization, clarify the boundaries of responsibility for environmental management matters between the central and local governments, and drive environmental protection and economic development in the central and western regions as the regions that get economically rich first and enjoy policy benefits first. The central and western regions, with fragile ecological environments, should firmly implement sustainable development strategies, screen industries for environmental damage when introducing them, learn from the development experience of the east, and coordinate economic, social, and ecological relations. Regional cooperation incentives should be implemented to promote the long-term environmental protection mechanism between regions and to achieve value co-creation. (2) Guide local governments to moderately increase the intensity of environmental regulation, form a new type of environmental regulation system, and bring into play the role of environmental regulation in promoting economic growth. The three environmental regulation tools should not be equally strongly enforced, but should focus on increasing the intensity of the current public participation-based environmental regulation. The central and local governments should collaborate to strengthen the publicity and popularization of environmental protection to the public, encourage public participation in the government’s environmental protection decisions, broaden the channels for public reflection, and improve the utilization rate and convenience of environmental protection petitions to create a favorable atmosphere of universal supervision. Command-and-control and market-incentive environmental regulations also have a positive effect on the development of green total factor productivity. The central government should, on the one hand, improve the environmental protection legal system and regulate strict law enforcement, and on the other hand, give policy support and government subsidies, improve the emissions permit trading system, etc., guide the transformation and upgrading of industries in each region, and play a leading role in the market. (3) Actively assume ecological responsibility and build a community of human destiny. As China’s position in the global value chain rises, greening the industrial structure is an inevitable requirement for development and ecological protection. Under the decentralized environmental governance system, the transfer of primary manufacturing industries from eastern regions to central and western regions and other large emerging economies is bound to increase local environmental pressure in order to achieve environmental protection performance and improve economic efficiency. China should insist on building “one belt and one road” to promote common development, advocate for sustainable and green production and life style, help other large emerging economies introduce advanced production and environmental protection technologies, promote green transformation and technological innovation in other large emerging economies, improve local livelihood environment and ecological environment, and improve global environmental governance.

It is worth noting that, given the specificity of China’s national situation, the uneven development among provincial administrative regions leads to possible differences in the situation and conditions at the more microscopic local and municipal levels and even at the district and county levels, and often the more microscopic regions directly affect GTFP. It is possible to explore in the future what the patterns are between environmental decentralization, environmental regulation, and GTFP at the local and municipal or district and county levels, and we can explore the spatial heterogeneity and spatial spillover effects between regions. This paper only explores the impact of one mediating pathway, environmental regulation, and whether other pathways jointly affect the role of environmental decentralization on GTFP needs further study. The measurement of environmental decentralization is still a hot spot and difficult issue for scholars in various countries. It is still necessary to explore and develop more scientific and reasonable environmental decentralization indicators and improve the environmental decentralization performance evaluation system in the future.