Research on the Impact of Atmospheric Self-Purification Capacity on Environmental Pollution: Based on the Threshold Effect of Environmental Regulation

Abstract

In China, Hebei Province and Guangdong Province have many pollution-intensive industries; yet Hebei suffers more serious atmospheric or environmental pollution than Guangdong. To explain the phenomenon, this paper chooses the statistical data of 286 prefecture-level cities in the Chinese mainland from 2005 to 2018 and empirically tests the spatial effect and threshold characteristics of atmospheric self-purification capacity on environmental pollution by using the spatial autoregression model (SAR), spatial Durbin model (SDM), and panel threshold model. As evinced, the local atmospheric self-purification capacity has a significant inhibitory effect on environmental pollution, and the absolute value of the direct effect coefficient reaches 1.337. Simultaneously, atmospheric self-purification capacity has a threshold effect on environmental pollution. The threshold value of environmental regulation as a threshold variable is 11.6349. This means that when the level of environmental regulation proves higher than 11.6349, it attenuates the inhibitory effect of the atmospheric self-purification capacity on environmental pollution. As the heterogeneity research unveils, atmospheric self-purification capacities in various regions form a significant correlation with environmental pollution. This paper suggests that local governments should strengthen environmental regulation and construct an inter-regional joint prevention and control system for atmospheric-pollution governance to enhance the atmospheric self-purification capacity.

1. Introduction

In China, Hebei Province and Guangdong Province have many pollution-intensive industries; yet Hebei suffers more serious atmospheric or environmental pollution than Guangdong. Why? Environmental pollution is not only related to the emission of pollutants, but is also affected by the self-purification capacity of the environment. In terms of the same pollution source, the degree of environmental pollution degree varies in regions with as their different environmental self-purification capacities are different. This paper mainly examines the impact of atmospheric self-purification capacity on environmental pollutants.

When the value of the atmospheric self-purification capacity remains high, it means that the atmosphere has a strong capacity to remove pollutants; otherwise, it means that the atmosphere has a weak capacity to remove pollutants. However, atmospheric self-purification capacity is limited. When the harmful substances that enter the atmospheric environment exceed its self-purification capacity, environmental pollution occurs. This paper attempts to answer these questions. Can the current atmospheric self-purification capacity effectively reduce environmental pollution? Given the spatial spillover effect of environmental pollution, is there a spatial correlation between atmospheric self-purification capacity and the governance of environmental pollution in local and neighboring regions? Considering the differences in economic development, urbanization process, population density, and industrial structure among various regions, under what conditions can atmospheric self-purification capacity play a better role in the governance of environmental pollution? Today, China vigorously fosters high-quality development. It is of practical and theoretical significance to research the impacts of atmospheric self-purification capacity on environmental pollution.

Atmospheric self-purification capacity signifies the capacity to purify environmental pollutants with physical and chemical actions. The concentration and toxicity of pollutants naturally dilute, drop, and disappear, so that environmental elements return to their originally clean state. The movement of the atmosphere per se can remove pollutants in the atmosphere. For instance, similar to the purification capacity of water, Examples are the diffusion and dilution of strong wind caused by the transit of cold air and wet removal of atmospheric pollutants by precipitation (similar to the purification capacity of water). Atmospheric self-purification capacity is defined as the capacities of diffusion, dilution, and wet removal of pollutants in the atmosphere owing to the movement of the atmosphere per se (Zhu Rong et al., 2018) [1]. The pollutants that enter the atmosphere can be diffused and diluted to a wide space via physical and chemical actions, substantially reducing their concentration. Owing to the action of gravity and the washing of rain, heavier particulate matters fall on the ground. The pollutants can also decompose under the irradiation of light and the participation of other physical and chemical actions, which purifies the air and reduces atmospheric-pollution concentration. Moreira et al. (2005) [2] and Sharan et al. (2003) [3] reveal that low wind speed and weak turbulence make pollutants easily accumulate around the emission source. DeGaetano et al. (2004) [4] confirm that air pollution in New York mostly occurs in the southwest-wind environment with high humidity, high temperature, and low wind speed. Wang Zhen et al. (2020) analyze and conclude that atmospheric relative humidity forms a positive correlation with PM2.5 in Changzhou, China, that precipitation has a removal effect on PM2.5, and that wind speed forms a negative correlation with PM2.5 concentration [5]. Theoretically speaking, when the quantity of pollutant emission is determined, atmospheric self-purification capacity represents the capacity of the atmosphere to contain pollutants. In other words, lower atmospheric self-purification capacity means a higher mass concentration of pollutants and vice versa (Tang Yingxiao et al., 2019) [6].

So far, most scholars agree on the impact of atmospheric self-purification capacity on environmental pollution. To put it in another way, the improvement of atmospheric self-purification capacity plays a vital role in mitigating air pollution. Chinese scholars mostly take a city or region in China as a research object, analyze the change characteristic of atmospheric self-purification capacity, discuss the relationship between atmospheric self-purification capacity index and air quality in the sample regions, and draw similar conclusions. For example, scholars observe that in Tianjin when the atmospheric self-purification capacity remains low, the heavy-pollution events are most likely to occur (Tang Yingxiao et al., 2019) [6]. In Hohhot, the atmospheric self-purification capacity index forms a negative correlation with air pollution index (Liu Xin et al., 2020) [7]. In the Yellow River Basin and main urban districts of Zhengzhou, the atmospheric self-purification capacity index forms a significant negative correlation with PM2.5 concentration, or when the atmospheric self-purification capacity index reaches a high level, the corresponding PM2.5 concentration remains low, with good air quality (Liu Mei, 2021 [8]; Liu Mei, 2022 [9]). Luo Yu et al. (2021) argue that a low atmospheric self-purification capacity index means weak air diffusion capacity that hinders the removal and diffusion of pollutants, with high air quality index. A high atmospheric self-purification capacity index promotes the ventilation and diffusion of pollutants and the removal capacity of precipitation [10].

Under what conditions can atmospheric self-purification capacity play a better role in controlling environmental pollution? As scholars conclude, the impact of atmospheric self-purification capacity on environmental pollution is restricted by multiple factors. The first ones are meteorological factors. Meteorological conditions, e.g., temperature inversion, low pressure, high humidity, weak wind, stable atmospheric stratification, low mixing-layer thickness, and weak precipitation, easily cause the accumulation of pollutants and constitute major meteorological factors for heavy pollution weather (Liu Houfeng et al., 2015) [11]. When the emission of external pollutants remains relatively stable, the degree and change of air pollution mostly hinge on meteorological factors, or the dilution effect of atmospheric self-purification capacity on pollutants varies with different meteorological conditions (Dong Xuguang et al., 2018) [12]. Season serves as an important factor that affects meteorology. For example, Linfen City, Shanxi Province, has the highest wind speed, the lowest relative humidity, the highest mixing-layer height, and the best atmospheric self-purification capacity in spring (Han Yan et al., 2019) [13]. Heilongjiang Province has the highest atmospheric self-purification capacity in spring, higher in autumn, and the lowest in winter (Zhu Hongrui et al., 2020) [14]. Xinjiang has a high atmospheric self-purification capacity in spring and summer, and a low atmospheric self-purification capacity in autumn and winter (Wang Yu et al., 2021) [15]. Spatial changes also affect atmospheric self-purification capacity. In China, the regions with the lowest atmospheric self-purification capacity are distributed in basins, whilst the regions with the highest atmospheric self-purification capacity are distributed in plateaus, plains, and islands (Zhu Rong, 2018) [1]. Some scholars focus on provinces and cities. For example, the spatial distribution of the atmospheric self-purification capacity index in Heilongjiang Province generally looks low in the north and high in the south (Zhu Hongrui et al., 2020) [14]. The spatial change of atmospheric self-purification capacity in the Chengdu Plain Economic Zone of Sichuan Province shows an overall trend of low in the southwest and high in the northeast (Bai Ge and Ni Changjian, 2021) [16]. In addition to natural factors, human factors affect the impact of atmospheric self-purification capacity on environmental pollution. Yu Zhenyan et al. (2017) reveal that urbanization is one of main reasons for the worsening air quality in Zhejiang Province. The deepening of urbanization increases the underlying surface roughness. This, in turn, reduces ground wind speed and weakens atmospheric self-purification capacity [17]. Zhu Rong et al. (2019) discover that the urban scale of the Beijing–Tianjin–Hebei region and its neighboring regions forms a reverse relationship with ground wind speed. Therefore, the rapid increase in heavily polluted meteorological conditions is also affected by urbanization to some extent [1].

As the core means for national agencies to control environmental pollution, environmental regulation helps to consolidate the path of regional green development and offset the lack of market governance (Hu W. and Wang D., 2020) [18]. Chinese and foreign scholars attain remarkable achievements in the research on the relationship between environmental regulation and environmental pollution. In terms of how to promote the effect of environmental regulation on emission reduction, relevant research mainly centers on the dominant role of the reversed emission reduction effect or the compliance cost effect in environmental regulation (Shen Zhao and Qu Xiao’e, 2022) [19]. Scholars who support the former view believe that strengthening environmental regulation can curb environmental pollution in various regions. There are two reasons. On the one hand, governments compel enterprises to conduct necessary pollution pre-treatment and reduce waste emissions by taking regulatory policies such as raising the threshold for the entry of polluting industries and levying high environmental pollution taxes (Guo Lingjun et al., 2022) [20]. On the other hand, long-term environmental regulation encourages enterprises to engage in more innovative activities, improve their production efficiency and competitiveness, and forge the innovation compensation effect to offset the compliance cost of environmental regulation. This spurs corporate technological innovation (Yang Wei, 2019) [21] and has a positive impact on the prevention and control of environmental pollution. Simultaneously, environmental regulation increases the production cost of some pollution-intensive industries. The drop in industrial profits squeezes out some enterprises with the inappropriate allocation of production factors and low efficiency, helping to achieve independent adjustment, optimization, and upgrading of industrial structure (Li Jiajia et al., 2022) [22]. Scholars who are in favor of the latter view affirm that environmental regulation internalizes the external environmental costs of enterprises in the short term. As the corporate cost or expenditure increases, it constrains the capital for technological innovation and impedes the enterprise’s technological innovation. Yin Xiguo and Chen Yaojun (2022) conclude that the implementation of environmental regulation counts against the transformation of new and old kinetic energies [23].

In brief, the existing literature makes great contributions to the research in this regard yet fails to discuss three topics satisfactorily. Hopefully, this paper can further the research.

First, the research findings on atmospheric self-purification capacity prove inadequate. Scholars commonly acknowledge that raising atmospheric self-purification capacity has a positive effect on improving air quality, which lays a solid foundation for this paper. Nevertheless, under what conditions can atmospheric self-purification capacity play a better role in controlling environmental pollution, and what factors affect the role? In this regard, the existing scholarship concentrates on human factors like meteorological condition and urban construction yet fails to fully discuss the role of comprehensive factors such as the natural purifying factors of the atmospheric environment and the rationality of human design. Scholars seldom research whether environmental regulation functions or how it functions in the impact mechanism of atmospheric self-purification capacity on environmental pollution. Based on the existing literature, this paper further analyzes the influence effect and spatial correlation effect of comprehensive factors such as atmospheric self-purification capacity and the rationality of human design on environmental pollution and scrutinizes the role of environmental regulation in the impact mechanism of atmospheric self-purification capacity on environmental pollution. This enormously enriches the research findings on the impact mechanism of environmental regulation and atmospheric self-purification capacity on environmental pollution.

Second, Chinese and foreign scholars score great success in the research on the relationship between environmental regulation and environmental pollution. Noticeably, whether environmental regulation promotes the reduction in environmental pollution remains controversial. Scholars hold different views on the dominant role of the reversed emission reduction effect or the compliance cost effect. That is to say, scholars have not agreed on the issue that environmental regulation augments or attenuates the impact mechanism of atmospheric self-purification capacity on environmental pollution. This paper investigates the threshold effect of environmental regulation and unmasks the relationship between atmospheric self-purification capacity and environmental pollution more accurately. Therefore, the research conclusion possesses more practical and instructive significance.

Third, the existing research focuses on empirical research, whose research methods and models set an example for relevant research in this field. However, grounded in regional samples, the existing research highlights practical problems or empirical tests in provinces and regions (Lu Xuehuan and Bai Tingting, 2020) [24]. The lack of national data limits the guiding significance of the research conclusion. This paper chooses panel data from 286 prefecture-level cities in the Chinese mainland from 2005 to 2018, hoping to improve the commonality and guiding value of the research conclusion.

The rest of this paper is structured as follows. Section 2 introduces the theoretical mechanism and proposes the research hypothesis. Section 3 concerns research design. Section 4 conducts measurement estimation and analysis and presents the results of empirical research. Section 5 concludes and gives suggestions on policies.

Structurally, this paper is designed as follows. Part two introduces the theoretical mechanism and proposes the research hypothesis. Part three concerns research design. Part four conducts measurement estimation and analysis and presents the results of empirical research. Part five concludes and gives suggestions on policies.

2. Theoretical Mechanism and Research Hypotheses

Owing to the influence of natural conditions (e.g., ventilation, precipitation, and humidity), as well as human factors (e.g., urban planning and layout and the layout of polluting industries), environmental pollution forms a spatial correlation. Atmospheric self-purification capacity mainly rests on the dilution, diffusion, and oxidation of the atmosphere, with a spatial spillover effect among regions. Meanwhile, in addition to atmospheric self-purification capacity, differences in socio-economic factors such as economic development, technological innovation, and environmental regulation affect the governance of environmental pollution. Therefore, in terms of theoretical mechanism, this paper proceeds to analyze the governance effect, spatial correlation effect, and threshold effect of atmospheric self-purification capacity on environmental pollution.

2.1. The Effect of Atmospheric Self-Purification Capacity on the Governance of Environmental Pollution

With the rapid development of the economy and society and the acceleration of urbanization, social activities aggravate the emission of pollutants aggravates, the environmental pollution intensifies, and the scope of pollution enlarges. When the emission of pollutants reaches the maximum allowable emission amount for the atmosphere of an ambient-air-protection target, it signals the environmental self-purification capacity of atmospheric pollutants [6]. In recent years, human activity, urban layout, industrial structure, and other factors have continuously increased the emission of environmental pollution, exacerbating air quality. Particularly, human factors change meteorological elements, further sustainably decrease atmospheric self-purification capacity, and aggravate environmental pollution. These cannot be ignored [25].

In terms of mechanism, the governance effect of regional atmospheric self-purification capacity on environmental pollution operates in three ways. First, factors such as temperature, near-surface wind speed, and precipitation affect the removal and accumulation of pollutants in local regions by changing the transport and diffusion capacities of the near-surface atmosphere [26,27]. When the environment is polluted, it can gradually eliminate pollutants and achieve natural purification via physical, chemical, and biological actions. The dilution, diffusion, oxidation, and other physical or chemical actions of the atmosphere make the emitted pollutants appear. For example, particulate matter emitted into the atmosphere falls to the ground after being washed by rain and snow for atmospheric self-purification. Fully understanding and utilizing atmospheric self-purification capacity can reduce the concentration of pollutants and the harm of pollution. Second, the environmental bearing capacity sets a hard constraint for the formulation of the polices on the governance of environmental pollution. Therefore, the implementation of these policies should base itself on scientifically understanding the environmental self-purification capacity. Relevant parties determine the environmental bearing capacity, lay down corresponding operable policies on the treatment of environmental pollution, and control the concentration of environmental pollutants within the practical scope. The environmental policy serves as an important means to adjust industrial structure. For one region, it can achieve the goals of strengthening the environmental bearing capacity and of better adapting to the environmental bearing capacity by adjusting industrial structure and layout [28]. Third, atmospheric self-purification capacity reversely promotes urban green planning, and makes rational use of environmental resources to optimize urban planning and layout in line with regional features [5]. To enhance the atmospheric self-purification capacity, a region can enlarge the greening area, plant trees, and establish a nature reserve. Based on environmental improvement and protection, the ecological city that is people-oriented and suitable for human settlement will be built.

In summary, this paper proposes research Hypothesis 1.

Hypothesis 1 (H1).

Augmenting the environmental self-purification capacity helps to alleviate environmental pollution in one region.

2.2. The Spatial Correlation Effect of Atmospheric Self-Purification Capacity on the Governance of Environmental Pollution

The spatial effect of atmospheric self-purification capacity on environmental pollution depends on the view that environmental pollution in various regions has a spatial spillover effect. For example, whenever the PM2.5 concentration in neighboring regions increases, the PM2.5 concentration in one region will increase accordingly [29]. For atmospheric self-purification capacity under different meteorological conditions, the dilution and diffusion capacities of pollutants change. Therefore, atmospheric self-purification capacity probably displays a spatial correlation effect among various regions. Coupled with spatial and geographical factors, environmental pollution may have an inter-regional effect.

In terms of mechanism, the spatial correlation between atmospheric self-purification capacity in neighboring regions and local environmental pollution operates in three ways. First, the positive externality of the improvement of atmospheric self-purification capacity: ventilation and precipitation-based atmospheric self-purification capacity has a close tie with air-environment quality. When the atmospheric self-purification capacity index remains low, atmospheric diffusion capacity becomes weak, which hinders the removal and diffusion of pollutants and causes atmospheric environmental pollution [17]. The improvement of atmospheric self-purification capacity in neighboring regions will directly reduce environmental pollution from these regions by raising urban greening coverage and transferring polluting industries. In its neighboring regions, the drop in environmental pollution also reduces environmental pollution in one region. Second, the positive externality of air-pollution governance: cities with serious air pollution can hardly attract talent, high-tech, and other elements, which affects the accumulation level of urban capital and further undermines the quality and sustainability of economic growth. To better curb environmental pollution promotes the high-quality development of one region and enhances its attractiveness. Third, as “hitchhiking” often occurs in environmental governance, the atmospheric self-purification capacity of one region can drive the governance of environmental pollution in neighboring regions (or a demonstration effect) in a limited space. To meet people’s growing yearning for a better life, neighboring regions learn and implement advanced ideas of environmental-pollution governance, thus forming the spatial correlation effect of atmospheric self-purification capacity on environmental-pollution governance [30].

In brief, this paper proposes research Hypothesis 2.

Hypothesis 2 (H2).

The improvement of atmospheric self-purification capacity in its neighboring regions helps to reduce environmental pollution in one region.

2.3. The Threshold Effect of Atmospheric Self-Purification Capacity on the Governance of Environmental Pollution

In consideration of the differences in regional economic development and social systems, the impact of atmospheric self-purification capacity on environmental pollution is also affected by multiple factors, environmental regulation in particular.

With the improvement of policies and standards on environmental regulation, governments control the environmental-protection level of foreign investment and other factors more strictly and further strengthen the review of polluting industries. In this way, various regions enhance their atmospheric self-purification capacity and expand the pollution-reduction effect. Besides, various regions regulate or reduce the emission of environmental pollutants and control the emission within the scope of atmospheric environmental bearing capacity, to alleviate environmental pollution (Xu Pengjie and Lu Juan, 2018) [31]. However, strict environmental regulation inhibits the reduction-emission effect of pollutants. The intensity of environmental regulation directly determines whether enterprises plan to include the cost of the governance of environmental pollution in production costs. Additionally, when governments tighten the grip on environmental regulation, corporate compliance cost increases, which restricts innovation in production technology to some degree [32]. Simultaneously, the rise of the expenses on environmental-pollution governance and the price of production factors cut down the upgrading of the enterprise’s products. This not only retards pollution reduction with technological innovation but also entices enterprises to expand production scale or increase pollutant emission driven by the goal of maximizing production profits, weakening the inhibitory effect of atmospheric self-purification capacity on environmental pollution.

To sum up, this paper proposes research Hypothesis 3.

Hypothesis 3 (H3).

When environmental regulation reaches a high level, the inhibitory effect of atmospheric self-purification capacity on environmental pollution will be weakened.

3. Research Design

3.1. Framework and Variable Description

The theoretical-analysis framework of atmospheric environmental pollution follows the STIRPAT model and the research framework of Huo Luping and Zhang Yan (2020) [33]. This paper summarizes the research on environmental pollution in previous studies, chooses economic development level, industrial structure, and population density as control variables in the model, and takes environmental regulation as the threshold variable in the model. The basic linear model of this paper is shown as follows:

lny_it = β₀ + β₁lnx_it + β₂lnk_it + β₃lnm_it + ε_it

(1)

In particular, i and t stand for city and year, respectively, β₀ stands for the constant term, β₁–β₄ stand for the estimation coefficient of each factor, x stands for the explanatory variable, y stands for the explained variable, k stands for the threshold variable, m stands for the control variable, and ε stands for the random error term.

AEP stands for the degree of atmospheric environmental pollution. This paper follows the practice of Lei Yutao and Liu Minglu (2017) [34] and uses the atmospheric environmental pollution comprehensive index to express the degree of atmospheric environmental pollution, i.e., atmospheric environmental pollution comprehensive index = (industrial sulfur dioxide/GDP + industrial smoke (dust)/GDP + PM2.5/GDP)/3, as shown in Equation (2). First, this paper determines the relative emission level of the first type of pollution in city i. Specifically, pli stands for the emission of the first pollution unit GDP of city i (absolute quantity of pollution emission/GDP). The higher value of pxli (at least more than 1) means a higher relative emission level of the first pollutant in the city i nationwide. As pxli per se denotes a dimensionless variable, it is important to sum and average as follows. In a robustness test, this paper follows the research conclusion of Lu Xuehuan et al. (2020) [32], uses ArcGIS software to extract the grid data of the global annual mean value of PM2.5 concentration based on satellite monitoring released by Columbia University of the United States, and collates the annual average PM2.5 concentration (PM) of 286 prefecture-level cities from 2005 to 2018 to measure the degree of atmospheric environmental pollution.

$${\mathit{px}}_i=\frac{\mathit{1}}{\mathit{3}}({\mathit{px}}_{\mathit{1}i}+{\mathit{px}}_{\mathit{2}i}+{\mathit{px}}_{\mathit{3}i})$$

(2)

.

API stands for atmospheric self-purification capacity. In the calculation methods of atmospheric self-purification capacity index, scholars lay stress on the ventilation, diffusion, and purification capacities of the atmosphere. For example, Yu Zhenyan et al. (2017) [17], Dong Xuguang et al. (2018) [12], and Zhu Hongrui et al. (2020) [14] use such indexes as ventilation capacity, precipitation, and unit area in the calculation formula of atmospheric self-purification capacity. Zhu Rong et al. (2018) [1], Luo Yu et al. (2021) [10], and Liu Mei et al. (2021) use such indexes as ventilation capacity, precipitation rate, air-quality control-concentration of typical pollutants, and unit area. Based on the existing research methods and the availability of data, this paper chooses the statistical data of relevant indexes, such as the maximum wind speed, the average two-minute wind speed, the cumulative precipitation, the maximum daily precipitation, the number of days with daily precipitation ≥0.1 mm, the average air pressure, the average temperature, the average relative humidity, the sunshine hours, urban green space area, population density, and urban built-up areas. The standardized score of atmospheric self-purification capacity is calculated according to the standardized index value and the index weight obtained from Delphi method (expressed by API). The specific calculation process is expressed as follows:

API_it = β₁(AT_it) + β₂(SH_it) + β₃(ARH_it) + β₄(AA_it) + β₅(MWS_it) + β₆(AWS_it) + β₇(MDP_it) + β₈(AP_it) + β₉(DP_it) + β₁₀(BUA_it) + β₁₁(PD_it) + β₁₂(GRB_it) + β₁₃(WA_it)

(3)

where β stands for the weight of different indicators, i stands for city, and t stands for year. The values in brackets are the standardized values of different indicator data, which can be obtained by standardizing the collected data. The standardization processing formula of the indicator is (y − y_min)/(y_max − y_min), in which y_min means the minimum value of the indicator data and y_max means the maximum value of the indicator data.

In terms of the choice of the control variable, this paper follows the practice of Huo Luping et al. (2016) [33] and chooses population density (den) to represent the concentration degree of the urban population, follows the research of Li Jianbao et al. (2019) [35] and chooses built-up area (bua) to represent the area that has actually been developed and constructed in a block with public facilities in the prefecture-level administrative region, and follows the research of Zhang Yu et al. (2020) [36] and chooses per capita GDP of prefecture-level cities to represent the economic development level (ey) of the research region. The above three factors mirror the urbanization process of various regions from different angles. The deepening of urbanization increases the underlying-surface roughness, which reduces the ground wind speed and weakens atmospheric self-purification capacity (Yu Zhenyan et al., 2017) [17]. This paper chooses the proportion of the secondary industry to GDP to demonstrate the impact of industrial structure (in) on environmental pollution (Lei Yutao et al., 2017) [34]. The reason lies in that the secondary industry dominates the heavy industry, and sulfur dioxide formed by fossil-fuel combustion in heavy industry and smoke and dust from the construction industry clearly constitutes important sources of environmental pollution, thus affecting the emission of pollutants (Han Nan, 2015) [37].

In terms of the choice of the threshold variable, this paper chooses environmental regulation (er). This paper follows the research conclusion of Smarzynska and Wei (2001) [38] and uses the removal amount of industrial sulfur dioxide to measure environmental regulation. On the relationship between environmental regulation and environmental pollution, there are two views. The first one admits that environmental regulation urges enterprises to engage in technological and organizational innovation, improve their production efficiency and market competitiveness, and ultimately achieve a win–win result of environmental protection and economic growth (Song Shuang, 2017) [39]. The second view oppugns that environmental regulation can effectively reduce environmental pollution, and proposes a “green paradox” (Ren Xiaojing, 2018) [40].

3.2. Measurement Model and Method

Atmospheric self-purification capacity not only reduces environmental pollution in one region, but also affects its neighboring regions. Therefore, to verify Hypothesis 1 and Hypothesis 2, this paper considers regional correlation while conducting empirical analysis. The spatial autoregression model (SAR) includes spatial correlation between different things in a general regression model, which offsets the lack of spatial correlation of variables in a general regression model. For this reason, this paper introduces the spatial autoregression model, which focuses on analyzing the spatial correlation of the explained variables. Owing to its natural hysteresis, environmental pollution has an inter-regional effect. Accordingly, this paper designs the space lag and time lag of environmental pollution in SAR. By connecting local atmospheric self-purification capacity with neighboring regions, this paper exposes the impact of atmospheric self-purification capacity on environmental pollution more truthfully. In this paper, the estimation model of spatial correlation is expressed as follows:

$$lnAE{P}_{it}=\rho {{\displaystyle \sum }}_{j=1}^N{\mathrm{W}}_{it}lnAE{P}_{it}+{\beta }_{1}wlnAP{I}_{it}+{\beta }_{2}lne{r}_{it}+{\beta }_{3}lnde{n}_{it}+{\beta }_{4}lnbu{a}_{it}+{\beta }_{5}lne{y}_{it}+{\beta }_{6}lni{n}_{it}+{\mu }_i+{\lambda }_i+{\epsilon }_{it}$$

(4)

where 𝜌 stands for spatial lag (autoregression) coefficient, W_it stands for the element in row i and column j of standardized non-negative spatial weight matrix W in N * N dimension, and 𝜇_i and 𝜆_i stand for spatial (individual) effect and temporal effect, respectively.

In the spatial measurement model, the spatial Durbin model (SDM) considers spatial correlations of the explained variables and the explanatory variables respectively. The spatial Durbin model can obtain unbiased coefficient estimation, it better estimates the spillover effect generated by different observation individuals and measures the spatial spillover effect based on panel data. This paper adopts spatial Durbin model (SDM) estimation based on spatial panel estimation to better investigate the impact of relevant factors.

$$\begin{gathered} lnAE{P}_{it}=\rho {{\displaystyle \sum }}_{j=1}^N{W}_{it}lnAE{P}_{it}+{\beta }_{1}wlnAP{I}_{it}+{\beta }_{2}lne{r}_{it}+{\beta }_{3}lnde{n}_{it}+{\beta }_{4}lnbu{a}_{it}+{\beta }_{5}lne{y}_{it}+{\beta }_{6}lni{n}_{it}+ \\ {\beta }_7{{\displaystyle \sum }}_{j=1}^N{W}_{it}wAPI+{\beta }_8{{\displaystyle \sum }}_{j=1}^N{W}_{it}lne{r}_{it}+{\beta }_9{{\displaystyle \sum }}_{j=1}^N{W}_{it}lnde{n}_{it}+{\beta }_{10}{{\displaystyle \sum }}_{j=1}^N{W}_{it}lnbu{a}_{it}+{\beta }_{11}{{\displaystyle \sum }}_{j=1}^N{W}_{it}lne{y}_{it}+ \\ {\beta }_{12}{{\displaystyle \sum }}_{j=1}^N{W}_{it}lni{n}_{it}+{\mu }_i+{\lambda }_i+{\epsilon }_{it} \end{gathered}$$

(5)

According to Hypothesis 3, environmental regulation and atmospheric self-purification capacity probably have a threshold effect. To examine the possible threshold effect, this paper introduces the threshold value of environmental regulation as an unknown variable into the measurement model, constructs the piecewise function of atmospheric self-purification capacity on environmental pollution, and tests and estimates its threshold value and threshold effect. In the case of the single threshold effect, the measurement model is expressed as follows:

$$\begin{gathered} lnAE{P}_{it}={\beta }_0+{\beta }_{1}lne{r}_{it}\left(lne{r}_{it}<\gamma \right)+{\beta }_{2}lne{r}_{it}\left(lne{r}_{it}\ge \mathsf{\gamma }\right)+{\beta }_{3}wlnAP{I}_{it}+{\beta }_{4}lnde{n}_{it}+{\beta }_{5}lnbu{a}_{it}+{\beta }_{6}lne{y}_{it}+ \\ {\beta }_{7}lni{n}_{it}+{\epsilon }_{it} \end{gathered}$$

(6)

The multiple threshold estimation equation is extended from the above equation.

3.3. Data Sources and Descriptive Statistics

There are 338 prefecture-level regions in Chinese mainland. This paper chooses the data of 286 prefecture-level cities in the Chinese mainland from 2005 to 2018 as samples. The samples are highly representative. Data sources of relevant indicators cover the China Urban Statistical Yearbook, China Urban Construction Statistical Yearbook, China Environmental Statistical Yearbook, China Water Conservancy Statistical Yearbook, China Regional Economy Statistical Yearbook, statistical yearbooks of some provinces, statistical yearbooks of water conservancy in some provinces, Zhejiang Province Natural Resources Yearbook, Zhejiang Province Natural Resources and Environment Statistical Yearbook, China Meteorological Data Service Center, and National Intellectual Property Administration. For some missing data, to prevent the mutation of data in the time sequence, this paper mainly uses the autocorrelation fitting method to estimate. Noticeably, in terms of days of air quality with Grade II or above, the data are available in only 80 prefecture-level cities, which are missing in some years. The missing data in some years are filled with the provincial average value. To eliminate the influence of data dimension and improve the goodness of fit of the model, this paper standardizes all indicators. The descriptive statistics of variables are shown in

Table 1. The Descriptive Statistics of Variables.

Variable	Sample Size	The Mean Value	Standard Deviation	The Minimum Value	The Maximum Value
lnAEP	3990	−21.087	2.286	−27.250	0.000
lnPM	1120	5.706	0.209	3.892	5.903
lnAPI	3990	−0.211	0.266	−1.297	0.295
wlnAPI	3990	−0.210	0.262	−0.879	0.215
lner	3990	10.324	2.195	−13.816	16.728
lnden	3990	7.942	0.806	3.296	9.920
lnbua	3990	4.366	0.876	1.879	7.979
lner	3990	10.313	0.776	4.595	15.675
lnin	3990	3.847	0.250	2.197	4.511

Note: In line with the comparison between standard deviation and mean value, this paper winsorizes lnAPI at the 99% level only when the standard deviation of the lnAPI proves higher than the mean value.

.

4. Measurement Estimation and Analysis

4.1. Estimation Results and Analysis Statistics of Spatial Correlation Effect

To verify hypotheses 1 and 2,

Table 2. The Analysis of Spatial Correlation Effect.

	The Total Effect		The Direct Effect		The Indirect Effect
Model	(1) SAR	(2) SDM	(1) SAR	(2) SDM	(1) SAR	(2) SDM
Variable	lnAEP	lnAEP	lnAEP	lnAEP	lnAEP	lnAEP
LR_Direct
wlnAPI	−0.173	−1.553 *	−0.094	−1.337 ***	−0.079	−0.216
	(−0.164)	(−1.873)	(−0.168)	(−4.461)	(−0.160)	(−0.272)
lner	−0.049 *	−0.240 ***	−0.026 *	−0.023	−0.023 *	−0.217 ***
	(−1.781)	(−3.481)	(−1.777)	(−1.521)	(−1.777)	(−3.523)
lnden	−0.002	0.132	−0.001	−0.091 ***	−0.001	0.223 *
	(−0.036)	(0.958)	(−0.040)	(−2.644)	(−0.031)	(1.772)
lnbua	−1.013 ***	−0.630 *	−0.539 ***	−0.380 ***	−0.474 ***	−0.251
	(−6.635)	(−1.903)	(−6.667)	(−5.447)	(−6.343)	(−0.850)
lney	−2.408 ***	−1.992 ***	−1.281 ***	−1.583 ***	−1.126 ***	−0.409 **
	(−25.758)	(−11.431)	(−24.288)	(−25.727)	(−17.907)	(−2.493)
lnin	0.672 **	2.902 ***	0.358 **	0.523 ***	0.314 **	2.379 ***
	(2.453)	(7.308)	(2.447)	(3.447)	(2.443)	(7.070)
N	3990	3990	3990	3990	3990	3990
R2	0.051	0.441	0.051	0.441	0.051	0.441
adj. R2

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

shows the empirical results of spatial correlation effect of atmospheric self-purification capacity on environmental pollution by using SAR and SDM. As Table 2 suggests, the statistical test of the estimation value of spatial lag parameter of environmental pollution proves significant, indicating that environmental pollution has significant spatial correlation and spatial spillover phenomenon. Spatial correlation effect of atmospheric self-purification capacity on environmental pollution can be summarized as the total effect, the direct effect, and the indirect effect. The direct effect means the average impact of changes in atmospheric self-purification capacity on environmental pollution in one region, or the spillover effect of atmospheric self-purification capacity in one region. The indirect effect means the average impact of atmospheric self-purification capacity in neighboring regions on environmental pollution in one region, or the spatial spillover effect of atmospheric self-purification capacity in neighboring regions. The total effect means the sum of the direct effect and the indirect effect. Among them, SDM has a higher fitting degree. This paper mainly explains and analyzes the SDM results.

Table 2 basically investigates the direct effect and indirect effect of atmospheric self-purification capacity (wlnAPI) on environmental pollution (lnAEP), or the impacts of atmospheric self-purification capacity in one region and its neighboring regions on environmental pollution. The total effect and direct effect both show that the coefficient of atmospheric self-purification capacity (wlnAPI) proves negative and passes the significance test. This demonstrates that atmospheric self-purification capacity helps to mitigate environmental pollution in one region and tallies with theoretical prediction. The reason is that atmospheric self-purification capacity signifies an inherent function of the environment to automatically eliminate pollutants and purify them. The pollutants that enter the atmosphere can be diffused and diluted to a wide space via physical, chemical, and biological actions under natural conditions, substantially reducing their concentration. Owing to the action of gravity and the washing of rain, heavier particulate matters fall on the ground. The pollutants can also decompose under the irradiation of light and the participation of other physics, which purifies the air and reduces atmospheric-pollution concentration. In terms of the impact of atmospheric self-purification capacity in neighboring regions, the coefficient of its indirect effect (wlnAPI) proves negative yet insignificant. To put it in other way, atmospheric self-purification capacity in neighboring regions has a spillover effect on local environmental pollution, with insignificant spillover effect. The reason lies in that atmospheric self-purification capacity is affected by many factors such as precipitation, ventilation, urban planning, pollution-source intensity, and vegetation coverage, which embody large spatial differences. With the transformation of social contradictions in China, central and local governments attach more importance to environmental issues. By increasing urban greening coverage, transferring polluting industries, optimizing and upgrading industrial structure, and adjusting urban layout, one region can improve atmospheric self-purification capacity, thus strengthening the removal of environmental pollution by atmospheric self-purification capacity. For example, coastal regions like Fujian, Guangdong, and Zhejiang give full play to the purification of precipitation on environmental pollution. Simultaneously, in high-level cities, economic development plays a significant role in promoting atmospheric self-purification capacity, which remarkably reduces environmental pollution.

As for the total effect or indirect effect of environmental regulation (lner), the coefficient proves significantly negative and has a significant inhibitory effect on environmental pollution. Environmental regulation can form the reversed emission reduction effect and the compliance cost effect to alleviate environmental pollution and achieve emission reduction. As China formulates and implements a series of measures and policies on the governance of environmental pollution, environmental regulation scores a success in reducing pollution emission. Simultaneously, in neighboring regions, environmental regulation and environmental-pollution emission both have significant spatial spillover effects and show a symbiotic pattern of interdependence.

In terms of the control variables, as the coefficient of population density (lnden) suggests, there is a significant negative correlation between local population density and local environmental pollution. The rise of population density facilitates the centralized supply and use of direct energy consumption, reduces the emission of polluting gases represented by carbon emission, and further mitigates environmental pollution. The indirect effect of population density has a significant positive correlation, indicating that the increase in population density in one region aggravates environmental pollution in neighboring regions. The reason lies in that with the increase in population density in one region, the pressure on various resources and infrastructure in local and neighboring regions soars, and pollutant emission jumps, which pollutes and destroys the environment in neighboring regions. Yet, in the total effect, there is no significant correlation between population density and atmospheric environmental pollution. Population density affects environmental pollution in two ways, i.e., the scale effect and the agglomeration effect. In regions with high population density, citizens raise high requirements for clothing, food, housing, and transportation in daily life, which all cause lots of environmental pollution, and high residential density prevents the diffusion of pollutants. The increase in population density also poses pressure on the bearing capacity of land, expands the demand for land, causes the reduction in forest area, land degradation and other resource problems, and intensifies environmental pollution. Notably, albeit the increase in population has a negative impact on environmental pollution, population density, as an indirect indicator, cannot directly reflect the impact.

The coefficients of the total effect and direct effect of built-up area (lnbua) show that there is a significant negative correlation between built-up area and environmental pollution. As the built-up area enlarges, urban scale expands. On the one hand, with greater pressure and capacity in environmental governance, they can increase investment in environmental governance, continually develop pollution-prevention technologies, and ameliorate the quality of atmospheric environment. On the other hand, with the expansion of urban scale, per capita income climbs, and citizens yearn for a better green life. They augment environmental awareness and appeal and prefer green products in their consumption structure. Various environmental-protection organizations continue to popularize environmental protection, launch green products and services, and improve local environmental quality.

As the effect coefficient of per capita GDP (lney) suggests, local per capita GDP has a significant negative correlation with environmental pollution, indicating that per capita GDP weakens local environmental pollution. The spatial spillover effect of per capita GDP proves negative, indicating that per capita GDP has a negative effect on environmental pollution in neighboring regions. The increase in local per capita GDP reduces the level of environmental pollution in neighboring regions. One possible reason is that the rise of local per capita GDP forms a siphonage, resulting in the inflow of industries and reducing the emission of environmental pollution in neighboring regions. Supposing the population size remains the same, the level of environmental pollution in neighboring regions can be remarkably reduced.

The increase in the proportion of the secondary industry in GDP (lnin) has a significant positive effect on environmental pollution in local and neighboring regions. This is mainly because industry is the main source of the emission of environmental pollution, and various regions achieve the goal of rapid economic development by dint of industrial development. These exacerbate environmental pollution. By optimizing and adjusting of industrial structure, these industries and production factors feature high pollution and low efficiency and have a negative impact on the atmospheric environment of the transfer-in regions.

4.2. The Estimation Result and Analysis of Threshold Effect

To determine the specific form of the threshold model, this paper tests whether environmental regulation has a threshold effect, as well as a specific number of thresholds, as shown in

Table 3. The Threshold Effect Values and 95% Confidence Intervals of Variables.

	Thresholds	95% CI
Single Model	11.6349	11.6145	11.662
Double Model
Ito1	11.800	11.6817	11.8099
Ito2	10.274	10.0872	10.2887

. Besides, this paper analyzes the threshold values of the single threshold and double threshold of environmental regulation (lner), as well as the 95% confidence intervals. In terms of environmental regulation, its single threshold value is 11.6349, and double threshold values are 11.800 and 10.274. Table 3 shows 95% confidence intervals in a detailed way.

Then, this paper presents the threshold effects of variables. As the results of the self-sampling test suggest, the single threshold effect of environmental regulation passes the significance test of 5%, and its double threshold effect passes the significance test of 5% (

Table 4. The Results of the Self-Sampling Test of the Threshold Effect of Variables.

Model	Threshold	F	p-Value	BS-Reps	10%	5%	1%
Single Threshold	Single	27.340	0.040	300	22.772	25.6212	32.5621
Double Threshold	Single	27.590	0.030	300	21.877	26.3749	31.5228
	Double	23.990	0.043	300	17.258	23.6188	28.9096

). Therefore, the variable environmental regulation has a single threshold and double threshold effects.

Table 5. The Estimation of Threshold Effect.

	(1)	(2)
	lnAEP	lnAEP
wlnAPI	−1.310 **	−1.331 **
	(−2.114)	(−2.154)
lnden	−0.069 *	−0.078 *
	(−1.672)	(−1.903)
lnbua	−0.773 ***	−0.766 ***
	(−8.235)	(−8.183)
lner	−2.204 ***	−2.150 ***
	(−38.391)	(−36.858)
lnin	0.611 ***	0.606 ***
	(3.621)	(3.593)
0._cat#c.lner	−0.046 **	0.001
	(−2.530)	(0.070)
1._cat#c.lner	−0.080 ***	−0.032 *
	(−4.547)	(−1.747)
2._cat#c.lner		−0.065 ***
		(−3.632)
_cons	3.521 ***	2.762 ***
	(4.507)	(3.464)
N	3990	3990
R2	0.580	0.582
adj. R2	0.547	0.549

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

shows the regression results of the single threshold effect (1) and double threshold effect (2) of environmental regulation (lner). This paper will analyze the difference in the impact of atmospheric self-purification capacity on environmental pollution under the single threshold effect of environmental regulation. According to the statistical results in Table 5, when environmental regulation (lner) reaches less than 11.6349, the regression coefficient of environmental regulation to environmental pollution (lnAEP) proves significant and reaches −0.080; when environmental regulation reaches higher than 11.6349, the regression coefficient of environmental regulation to environmental pollution proves significant and reaches −0.065.

As evinced, with the increase in the natural logarithm of environmental regulation (the removal of industrial sulfur dioxide), its inhibitory effect on environmental pollution weakens, which basically accords with Hypothesis 3. The reason is that in general, environmental regulation in China largely depends on a command-control approach, with the aid of economic incentive. Economic-incentive-oriented environmental regulation bestirs people’s behavior motivation with market signals. For example, in the control of air pollution, emission taxes are levied on sulfur dioxide, nitrogen oxides, and smoke and dust emissions. Some countries start to execute carbon taxes and other policies. The cost of pollution expands with the increase in pollution, forcing enterprises to save production costs by supplementing investment, promoting technological innovation, adopting cleaner production technology, and introducing more advanced production equipment. If enterprises cannot upgrade pollution-control equipment or technology in a short term, they will depress production scale, decrease the production of polluting products, or turn to the production of green products in a bid to ease environmental pollution. This demonstrates that under the threshold effect of environmental regulation, atmospheric self-purification capacity can significantly reduce environmental pollution. The command-control approach regulates various activities of pollution emission. To meet the requirements of environmental regulation, enterprises necessarily increase their production costs, and to maximize their profits, enterprises expand their production scale. These multiply the emission of pollutants and cripple the removal effect of atmospheric self-purification capacity on pollutants. With respect to dynamic efficiency, performance efficiency lowers the cost of technological transformation to a certain extent, with higher efficiency than the technology standard. However, as command-control environmental regulation needs to consider the production costs of enterprises, it causes higher transaction costs of environmental regulation and affects the actual implementation efficiency.

4.3. Robustness Test

Following the practice of Lu Xuehuan et al. (2020) [24], this paper replaces the explained variable atmospheric environmental pollution comprehensive index and adopts PM2.5 to measure the degree of atmospheric environmental pollution and conduct a robustness test. Akin to the atmospheric environmental pollution comprehensive index, higher concentration of PM2.5 signifies worse air quality and more serious atmospheric environmental pollution, and vice versa. As

Table 6. The Robustness Test of Spatial Correlation.

	The Total Effect		The Direct Effect		The Indirect Effect
Model	(1) SAR	(2) SDM	(1) SAR	(2) SDM	(1) SAR	(2) SDM
Variable	lnPM	lnPM	lnPM	lnPM	lnPM	lnPM
LR_Total
wlnAPI	−0.846 ***	−1.044 ***	−0.435 ***	−0.162 **	−0.411 ***	−0.882 ***
	(−6.120)	(−6.674)	(−5.936)	(−2.292)	(−6.006)	(−6.410)
lner	0.013 ***	0.015 **	0.007 ***	0.010 ***	0.007 ***	0.005
	(3.552)	(2.370)	(3.617)	(4.797)	(3.429)	(0.877)
lnden	0.060 ***	0.026 **	0.031 ***	0.039 ***	0.029 ***	−0.013
	(6.775)	(2.017)	(6.928)	(8.282)	(6.257)	(−1.098)
lnbua	−0.146 ***	−0.281 ***	−0.075 ***	−0.054 ***	−0.071 ***	−0.227 ***
	(−7.051)	(−8.310)	(−7.278)	(−5.005)	(−6.422)	(−7.284)
lner	−0.057 ***	0.005	−0.030 ***	−0.012	−0.028 ***	0.016
	(−4.854)	(0.272)	(−4.786)	(−1.316)	(−4.770)	(0.965)
lnin	0.203 ***	0.271 ***	0.105 ***	0.138 ***	0.099 ***	0.133 ***
	(5.499)	(5.933)	(5.234)	(6.336)	(5.563)	(3.335)
N	3990	3990	3990	3990	3990	3990
R2	0.030	0.000	0.030	0.000	0.030	0.000

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

suggests, the total effect, direct effect and indirect effect of atmospheric self-purification capacity (wlnAPI) on PM2.5 annual average concentration (lnPM) prove significantly negative. In terms of spatial spillover, after replacing the explained variables, atmospheric self-purification capacity has a significant negative correlation with the environmental pollution in one region and its neighboring regions. There are differences between PM2.5 and other pollutants. PM2.5 mostly contains fine particulate matters and possesses strong regional transmissibility. Therefore, the deterioration of PM2.5 pollution in one region has a negative impact on the PM2.5 concentration in neighboring regions and even farther regions. PM2.5 concentration is jointly determined by natural (external) factors and human (internal) factors. The intensity of pollutant emission forms the fundamental internal cause of high incidence of environmental pollution weather. Natural conditions, such as precipitation, wind speed, temperature, vegetation, and mountain barrier, constitute external factors that affect the rise of environmental pollution weather. When PM2.5 concentration exceeds atmospheric self-purification capacity, the hazard of environmental pollution intensifies, which undermines the ecosystem.

In terms of the threshold effect of the robustness test (as shown in

Table 7. The Threshold Effect Values and 95% Confidence Intervals of Variables.

	Thresholds	95% CI
Single Model	9.387	9.3661	9.4071
Double Model
Ito1	9.387	9.352	9.4071
Ito2	3.2189	.	.

and

Table 8. The Results of Self-Sampling Test of the Threshold Effect of Variables.

Model	Threshold	F	p-Value	BS-Reps	10%	5%	1%
Single Threshold	Single	56.79	0.01	300	35.0398	39.4667	52.9297
Double Threshold	Single	56.79	0.0033	300	36.7907	40.0855	53.0131
	Double	20.22	0.4033	300	34.2589	38.7254	44.4125

), this paper first analyzes the threshold values of the single threshold and double threshold of environmental regulation, as well as the 95% confidence intervals. In terms of environmental regulation, its single threshold value is 9.387, and double threshold values are 9.387 and 3.2189. Table 7 shows 95% confidence intervals in a detailed way.

In Table 8, this paper presents the threshold effects of variables. As the results of the self-sampling test suggest, the single threshold effect of environmental regulation passes the significance test of 5%, and its double threshold effect fails the significance test of 5%. Therefore, the variable environmental regulation has a single threshold effect.

Table 9. The Robustness Test of Threshold Effect.

	(1)	(2)
	lnPM	lnPM
wlnAPI	−0.827 ***	−0.822 ***
	(−10.479)	(−10.438)
lnden	0.043 ***	0.043 ***
	(8.153)	(8.187)
lnbua	−0.086 ***	−0.087 ***
	(−7.233)	(−7.315)
lner	−0.078 ***	−0.082 ***
	(−10.608)	(−11.070)
lnin	0.300 ***	0.302 ***
	(14.075)	(14.180)
0._cat#c.lner	−0.006 **	−0.037 ***
	(−2.004)	(−4.788)
1._cat#c.lner	0.002	−0.000
	(0.978)	(−0.154)
2._cat#c.lner		0.006 **
		(2.565)
_cons	3.170 ***	3.162 ***
	(30.962)	(30.951)
N	3990	3990
R2	0.225	0.229

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

shows the regression results of the single threshold effect (1) and double threshold effect (2) of environmental regulation. This paper will briefly explain the results of the single threshold effect. As the statistical results in Table 9 suggest, when environmental regulation reaches less than 9.387, the regression coefficient of environmental regulation to PM2.5 reaches 0.002; when environmental regulation reaches more than 9.387, the regression coefficient of environmental regulation to PM2.5 reaches 0.006. This demonstrates that after replacing the explained variable, atmospheric self-purification capacity also has an obvious environmental regulation threshold effect on environmental pollution.

4.4. Heterogeneity Analysis

4.4.1. The Analysis of Spatial Regression Test of Five Regions

In

Table 10. The Regression Results of SDM Model Heterogeneity in Five Regions.

		(1) The Yangtze River Delta	(2) The Pearl River Delta	(3) Beijing–Tianjin–Hebei Region	(4) Northeastern Region	(5) Chengdu–Chongqing Economic Circle
		lnAEP	lnAEP	lnAEP	lnAEP	lnAEP
The Total Effect	LR_Total
	wlnAPI	10.035 *	5.893	−19.463 ***	1.158	−19.533
		(1.801)	(0.863)	(−3.143)	(0.475)	(−1.355)
	lner	−1.361 ***	−0.548	−1.791 ***	−0.222	0.311
		(−3.301)	(−1.619)	(−3.645)	(−1.391)	(0.605)
	lnden	3.541 ***	0.057	−0.466	0.130	−0.303
		(3.561)	(0.074)	(−0.546)	(0.547)	(−0.878)
	lnbua	1.562 *	−1.780	1.618	1.191	−2.789 *
		(1.690)	(−0.925)	(1.434)	(1.419)	(−1.702)
	lney	−2.036 ***	−1.136	−0.855	−2.842 ***	−0.142
		(−3.383)	(−1.442)	(−0.957)	(−6.987)	(−0.207)
	lnin	12.605 ***	14.336 ***	5.284 ***	2.173 ***	−2.454
		(4.961)	(4.656)	(3.432)	(3.552)	(−1.169)
	N	574	126	182	490	266
	R2	0.537	0.654	0.783	0.443	0.209
The Direct Effect	LR_Direct
	wlnAPI	1.214	1.988	−4.253 ***	0.377	0.949
		(1.269)	(0.656)	(−2.824)	(0.335)	(0.426)
	lner	−0.271 ***	−0.041	−0.171	−0.073 *	−0.079
		(−5.519)	(−0.518)	(−1.428)	(−1.650)	(−1.059)
	lnden	0.187	0.112	−0.133	0.089	−0.131
		(1.470)	(0.502)	(−0.608)	(0.894)	(−1.568)
	lnbua	0.055	−0.468	0.219	0.151	−1.783 ***
		(0.490)	(−1.162)	(1.000)	(0.674)	(−6.309)
	lney	−1.512 ***	−1.518 ***	−1.743 ***	−2.898 ***	−0.123
		(−12.497)	(−5.798)	(−6.805)	(−11.745)	(−1.105)
	lnin	4.259 ***	5.136 ***	2.208 ***	2.891 ***	−1.822 ***
		(11.866)	(6.009)	(4.113)	(6.964)	(−4.237)
	N	574	126	182	490	266
	R2	0.537	0.654	0.783	0.443	0.209
The Indirect Effect	LR_Indirect
	wlnAPI	8.821 *	3.905	−15.210 ***	0.780	−20.482
		(1.734)	(0.673)	(−2.807)	(0.325)	(−1.600)
	lner	−1.090 ***	−0.507 *	−1.620 ***	−0.149	0.390
		(−2.899)	(−1.825)	(−3.821)	(−1.036)	(0.871)
	lnden	3.354 ***	−0.055	−0.333	0.042	−0.172
		(3.706)	(−0.088)	(−0.469)	(0.180)	(−0.542)
	lnbua	1.507 *	−1.312	1.398	1.040	−1.006
		(1.790)	(−0.814)	(1.420)	(1.429)	(−0.710)
	lney	−0.525	0.383	0.888	0.056	−0.020
		(−0.933)	(0.570)	(1.132)	(0.134)	(−0.032)
	lnin	8.346 ***	9.199 ***	3.076 **	−0.718	−0.632
		(3.633)	(3.881)	(2.503)	(−1.180)	(−0.336)
	N	574	126	182	490	266
	R2	0.537	0.654	0.783	0.443	0.209

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

, the comparison of the total effect reveals that atmospheric self-purification capacity has an inhibitory effect on environmental pollution in the Beijing–Tianjin–Hebei region and Chengdu–Chongqing Economic Circle, yet the effect varies. In the Beijing–Tianjin–Hebei region, the coefficient of the total effect reaches −19.463 and passes the significance test of 1%. This demonstrates that atmospheric self-purification capacity has a significant inhibitory effect on environmental pollution in the Beijing–Tianjin–Hebei region and an insignificant inhibitory effect in the Chengdu–Chongqing Economic Circle, where the coefficient of the total effect reaches −19.533. In the Beijing–Tianjin–Hebei region, both the direct effect and the indirect effect prove significantly negative, indicating that atmospheric self-purification capacity weakens environmental pollution in local and neighboring regions. The Beijing–Tianjin–Hebei region suffers from serious environmental pollution because of the high emission intensity of pollutants. In the region, there is unreasonable industrial structure, with a large proportion of heavy industry. Particularly, in Tangshan, Xingtai, and other cities, industrial structure mainly covers iron and steel and coal. As environmental pollution intensifies, China launches special plans for air pollution and raises stricter requirements and longer goals for environmental protection in major cities with environmental pollution, so as to promote the establishment of a working mechanism for joint prevention and control of air pollution in highly polluted regions. The Beijing–Tianjin–Hebei region has initially established joint prevention and control mechanism for air pollution, which forces relevant parties to optimize industrial structure in line with environmental regulation, gives full play to green-technology innovation in fog-haze governance, and ensures the emission of environmental pollutants within the scope of atmospheric self-purification capacity. Simultaneously, in the Beijing–Tianjin–Hebei region, with a large emission base of environmental pollutants, atmospheric self-purification capacity has a more significant effect on environmental pollution. Noticeably, in the Chengdu–Chongqing Economic Circle, the insignificant inhibitory effect of atmospheric self-purification capacity on environmental pollution probably results from the fact that pollutant emission mostly comes from motor vehicles. As an important platform for the development of the western regions in China, the Chengdu–Chongqing Economic Circle sees rapid economic development, and the number of motor vehicles keeps rising.

In terms of the total effect, atmospheric self-purification capacity aggravates atmospheric environmental pollution in the Yangtze River Delta, the Pearl River Delta, and the northeastern region, yet the effect varies. In the Yangtze River, there is a significant positive correlation between atmospheric self-purification capacity and environmental pollution, with a coefficient of 10.035. In the Pearl River Delta and northeastern region, the coefficients reach 5.893 and 1.158, respectively, yet prove insignificant, indicating that atmospheric self-purification capacity does not serve as a key factor that affects the concentration values of environmental pollution in the Pearl River Delta and northeastern region.

The Yangtze River Delta lies in southern China. With a developed economy and trade and high population density, cities are overloaded with population pressure, as well as excessive demand for resources and the environment. Besides, in industrial structure, the Yangtze River Delta relies heavily on industrialization, which becomes a main controlling factor for environmental pollution. With excess emission of pollutants, environmental bearing capacity oversteps the scope of atmospheric self-purification capacity, which aggravates environmental pollution in the region and its neighboring regions. China has issued The Action Plan for the Prevention and Control of Air Pollution and The Three-Year Action Plan to Win Blue Sky Protection Campaign as well as other documents, which take stricter measures on environmental assessment and pollution control in the Yangtze River Delta and provide policy support for improving environmental pollution. As Table 10 suggests, in the Pearl River Delta and northeastern region, environmental pollution is tied to apron strings of industrial structure. A higher proportion of secondary industry increases environmental pollution in the Pearl River Delta. The reason is that the rise of industrial output value kindles the increase in the emission of various pollutants and worsens environmental pollution. In the northeastern region, there is a significant negative correlation between economic development level and environmental pollution, or technological progress from economic development improves environmental level.

4.4.2. The Analysis of the Regression Results of Eastern, Central, and Western Regions

In China, there are distinct regional differences in economic development. In other words, the eastern region is significantly more developed than the central and western regions. Therefore, this paper researches the differences in how environmental self-purification capacity affects environmental pollution in different regions.

As the results of the SDM model suggest, in eastern, central, and western regions, environmental regulation (lner) passes the significance test of the direct effect and proves negative. The direct effects of population density (lnden) all prove negative, yet the central region does not pass the significance test. The coefficients of built-up area (lnbua) all prove negative, yet the eastern region does not pass the significance test. The economic development levels (lney) all pass the significance test and prove negative. The contribution of the economic development level of eastern region to environmental pollution control proves significantly higher than these of the central and western regions. In the direct effect of industrial structure (lnin), the coefficients of eastern and central regions prove significantly positive, whilst that of the western region proves significantly negative. In general, eastern and western regions embody relatively high environmental governance capacity, which probably results from the geographical environment. The eastern region possesses the advantages of developed economy, high-quality employees, advanced technologies, and high-tech talents. The eastern region not only achieves rapid economic development but also accumulates rich experience in environmental governance, which furnishes technological support for environmental improvement and establishes a foundation for a high quality of economic growth. The western region basically boosts economic development via the support of national policies. Accordingly, the eastern region enhances environmental self-purification capacity by technological means, whereas the western region improves environmental self-purification capacity or environmental quality based on natural conditions.

As

Table 11. The Regression Results of SDM Model Heterogeneity in Eastern, Central and Western Regions.

		(1) Eastern Region	(2) Central Region	(3) Western Region
		lnAEP	lnAEP	lnAEP
The Total Effect	LR_Total
	wlnAPI	−1.519	−0.353	−2.281
		(−1.293)	(−0.317)	(−1.601)
	lner	−0.398 ***	−0.294 **	0.020
		(−3.095)	(−2.492)	(0.244)
	lnden	0.467	0.548 **	−0.118
		(1.606)	(2.570)	(−0.743)
	lnbua	0.875 **	−0.623	−2.527 ***
		(2.172)	(−1.193)	(−4.098)
	lney	−2.167 ***	−1.932 ***	−1.302 ***
		(−8.243)	(−6.824)	(−4.361)
	lnin	7.304 ***	1.925 ***	0.078
		(9.752)	(3.254)	(0.126)
	N	1386	1400	1204
	R2	0.665	0.459	0.284
The Direct Effect	LR_Direct
	wlnAPI	−1.347 ***	−1.261 ***	−3.144 ***
		(−4.082)	(−3.313)	(−4.392)
	lner	−0.048 **	−0.054 **	−0.048 *
		(−2.165)	(−2.074)	(−1.726)
	lnden	−0.100 *	−0.072	−0.146 **
		(−1.854)	(−1.344)	(−2.098)
	lnbua	−0.059	−0.303 ***	−0.700 ***
		(−0.878)	(−2.717)	(−3.473)
	lney	−1.590 ***	−2.056 ***	−1.478 ***
		(−20.968)	(−17.781)	(−12.321)
	lnin	2.162 ***	1.639 ***	−0.761 **
		(10.122)	(7.059)	(−2.421)
	N	1386	1400	1204
	R2	0.665	0.459	0.284
The Indirect Effect	LR_Indirect
	wlnAPI	−0.172	0.908	0.863
		(−0.162)	(0.873)	(0.616)
	lner	−0.350 ***	−0.240 **	0.069
		(−3.077)	(−2.258)	(0.922)
	lnden	0.567 **	0.620 ***	0.029
		(2.194)	(3.194)	(0.197)
	lnbua	0.935 ***	−0.320	−1.827 ***
		(2.601)	(−0.682)	(−3.430)
	lney	−0.577 **	0.124	0.176
		(−2.465)	(0.467)	(0.626)
	lnin	5.142 ***	0.286	0.839 *
		(8.185)	(0.540)	(1.676)
	N	1386	1400	1204
	R2	0.665	0.459	0.284

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

evidences, in the eastern region, local atmospheric self-purification capacity has a significant inhibitory effect on environmental pollution. To put it in another way, in terms of the direct effect, the impact of atmospheric self-purification capacity on environmental pollution proves significantly negative at the level of 1%, with a coefficient of −1.347. The continuous increase in the concentration of atmospheric environmental pollution implies the rapid development of industrial enterprises and highly polluting industries, thus driving the fast development of the urban economy in the eastern region. However, when the emission of environmental pollutants exceeds environmental bearing capacity, it will destroy atmospheric self-purification capacity, exacerbate environmental pollution, and curb the development of the eastern region. In the eastern region, with the implementation of environmental regulation, such as national policy intervention and increased emission tax, enterprises continue to enlarge the R & D investment to green-technology innovation, promote the green transformation of production, and offset the consumption and pollution of environmental resources by technological means. To create a new situation of environment-friendly and high-quality economic development, the eastern region shifts to tertiary industry such as the software industry, light industry, finance and service industry in the transformation of industrial structure, which alleviates environmental pressure. However, it is an arduous task for the eastern region to promote the governance of environmental pollution. Particularly, in Beijing–Tianjin–Hebei region, the governance effect of air pollution remains unstable. Eastern region needs to take measures such as infrastructure construction and urban-greening construction to consolidate atmospheric self-purification capacity and improve the quality of the atmospheric environment.

In the central region, the direct effect of atmospheric self-purification capacity on environmental pollution proves significantly negative, with a coefficient of −1.261 at the level of 1%. The central region serves as the core region of China’s heavy industry. Traditional extensive production model fosters rapid economic development and causes serious environmental pollution. China initiated the strategy of “the Rise of Central China” in 2004. Since then, on the one hand, the central region undertook the industrial transfer of manufacturing from the eastern region and increased the proportion of manufacturing. On the other hand, corporate investment to green technology innovation remained inadequate, and unreasonable industrial structure became an important factor that aggravated environmental pollution in the central region. Subject to environmental regulation, governments guide enterprises in the central region to advance technological innovation, increase environmental taxes on highly polluting industries, and realize the coordination between environmental protection and economic growth. Yet, owing to the low level of economic development in the central region, technological innovation and industrial-structure transformation lag behind atmospheric environmental pollution and lead to serious environmental pollution.

In the western region, the direct effect of atmospheric self-purification capacity on environmental pollution proves significantly negative, with a coefficient of −3.144. Generally speaking, inconveniently located and sparsely populated, the western region is less economically developed than the eastern and central regions. The lower level of economic development objectively diminishes environmental pollution. In the western region, environmental pollution arises from the consumption of resources and the emission of pollutants in economic growth. In 2001, China implemented the strategy of development of the western regions in China, which improves the scale and quality of economic growth in the western region. On this basis, the western region reinforces the governance and control of polluting industries, increases investment to pollution treatment, realizes green and healthy development of enterprises, and transforms the traditional model into a new model in industrial development. This ameliorates atmospheric-environment quality in the western region.

5. Conclusions and Suggestions on Policies

Based on the above analysis, this paper discloses that weak and strong atmospheric self-purification capacities determine the degree of atmospheric (environmental) pollution in Hebei and Guangdong, two provinces with many pollution-intensive industries. Specifically, this paper comes to the conclusion at three levels.

First, atmospheric self-purification capacity has a significant inhibitory effect on environmental pollution. There is a significant negative correlation between (local) atmospheric self-purification capacity and environmental pollution. The absolute values of the coefficients of the total effect and direct effect of atmospheric self-purification capacity reach 1.553 and 1.337, respectively.

Second, atmospheric self-purification capacity has a spatial spillover effect. Atmospheric self-purification capacity forms a negative correlation with the environmental pollution in neighboring regions, yet the effect proves insignificant. The absolute value of the coefficient of the indirect effect of atmospheric self-purification capacity reaches 0.216.

Third, there is a threshold effect in the impact of atmospheric self-purification capacity on environmental pollution. With environmental regulation as a threshold variable, the threshold value reaches 11.6349. When environmental regulation reaches lower than 11.6349, the regression coefficient of environmental regulation to environmental pollution proves significant and reaches −0.080. When environmental regulation reaches higher than 11.6349, the regression coefficient of environmental regulation to environmental pollution proves significant and reaches −0.065. To conclude, a high degree of regional environmental regulation attenuates the inhibitory effect of atmospheric self-purification capacity on environmental pollution.

In line with the research conclusion, this paper proposes suggestions for policies as follows.

First, relevant parties should scientifically facilitate atmospheric self-purification capacity, especially in heavily polluted regions. Atmospheric self-purification capacity effectively curbs regional environmental pollution. Local governments should make full use of environmental self-purification capacity to improve air quality. Atmospheric self-purification capacity is affected by both natural factors and human factors. Natural factors lack elasticity. Seasonal change, terrain condition, atmospheric stability, maximum mixed-layer thickness, and other factors seldom change significantly. Therefore, by changing human factors (e.g., the ratio of built-up area, urban green planning, and reasonable industrial layout), relevant parties can form targeted strategies to improve regional atmospheric self-purification capacity. While planning energy structure, industrial development and urban-construction layout in a coordinated way, governments can comprehensively take technological measures to prevent and control pollution, and control the total emission and concentration of pollutants within the scope of environmental bearing capacity.

Second, relevant parties should reasonably formulate environmental policies in heavily polluted regions and give full play to the inhibitory effect of atmospheric self-purification capacity on environmental pollution. Atmospheric self-purification capacity has a significant threshold effect on environmental pollution. In other words, when the degree of regional environmental regulation remains relatively high, the inhibitory effect of atmospheric self-purification capacity on environmental pollution will be restricted. Local governments should take high-quality economic development as a guideline and lay down reasonable policies on environmental protection on the basis of promoting atmospheric self-purification capacity. In line with the reality in various regions, governments should scientifically determine the intensity of environ-mental regulation, improve corresponding laws and regulations, as well as policies on environmental protection, and broaden corresponding supervision and feedback channels. Governments should gradually improve the emission trading market, optimize the emission fee-collection system, and give full play to the positive incentive role of environmental regulation. Simultaneously, governments can attach importance to the indirect effect of environmental regulation. Local governments can strengthen the implementation of environmental regulation, force polluting industries to complete green upgrading, continuously increase the investment to in technological research and development of pollution enterprises (especially to clean environmental-protection technologies), reduce the risk in corporate green-technology innovation, and strengthen the cooperation and exchanges on green-technology innovation among governments to ensure the emission of pollutants within the threshold value of environmental bearing capacity (i.e., atmospheric self-purification capacity). Additionally, governments can bolster the role of environmental regulation in upgrading and optimizing industrial structure, boost the optimization and upgrading of industrial structure, guide enterprises to transform from high-emission and high-pollution models in-to clean and high-tech-oriented development direction, coordinate the goals of industrial development and environmental protection, and achieve green, circular, and low-carbon development based on reasonably determining the intensity of environ-mental regulation.

Third, relevant parties should transfer polluting industries based on the gradient of environmental self-purification capacity. They can strengthen inter-regional joint prevention and control and cooperative governance, establish inter-regional ecological compensation mechanisms, and innovate the pollution-regulation model. They need to take a holistic outlook and re-evaluate regional environmental pollution. In China, the impact of atmospheric self-purification capacity on the degree of environmental pollution varies tremendously in different regions. In the formulation of environmental pollution measures, various regions should highlight the regional spillover effect to avoid the lack-of-coordination governance model. They need to continuously expand the regional boundary of joint prevention and control of environmental pollution, gradually connect multiple regional joint prevention and control systems of environ-mental pollution, and quickly realize a regional collaborative governance model of environmental pollution. They can establish special organizations for environmental pollution management, scientifically set goals of environmental-pollution governance among local governments, clarify the responsibilities and obligations of local governments in environmental-pollution governance, conduct unified and centralized management of environmental pollution, build vertical management systems and collabo-rative networks for the prevention of environmental pollution with the pattern of “organisations serving cities and regions”, eliminate environmental pollution on time, hedge against the dynamic cumulative effect of environmental pollution, and reduce long-term negative pollution effects. Besides, they can establish joint environmental-pollution governance system among different industries and enterprises in different regions to avoid the “Pollution Haven Hypothesis”, focus on rectifying high-pollution industries like the chemical industry, thermal power, and coal, con-struct a common action program for enterprises to jointly control environmental pollution at the macro and micro levels, innovate the accountability model for pollution emitted among regional enterprises, optimize and improve regional linkage mechanism for the prevention of environmental pollution among enterprises, and maintain and raise the comprehensive results of regional environmental-pollution governance.