Polluting Industry Agglomeration, Environmental Regulation, and Urban Air Quality

Abstract

In China, with the increasing emphasis on the concept of green sustainable development, polluting industries characterized by pollution and high energy consumption are facing unprecedented challenges. The development of the intermediate demand-type characteristics of polluting industries should be more reasonably laid out and regulated. In this paper, environmental regulation and environmental quality are introduced into the new economic geography model. On the basis of theoretical analysis, the IV regression method was used to study the interaction between polluting industry agglomeration, environmental regulation, and their effects on urban air quality with key cities as research objects. The results show that an increase in the agglomeration of polluting industries leads to significant deterioration in urban air quality and that this effect is linear, whereas an increase in the intensity of environmental regulation significantly dampens this effect. Each 1% increase in the intensity of environmental regulation results in a 1.17% reduction in air pollution. Therefore, to effectively protect the environment, the development of polluting industries should be relatively decentralized. Additionally, city governments should fully consider their urban eco-geographical characteristics, directly reduce and indirectly inhibit the degree of agglomeration of polluting industries and simultaneously strengthen the intensity of environmental regulation.

1. Introduction

As China’s economy develops rapidly and its international status rises, so do its environmental problems, especially air pollution [1]. As a carrier of economic development, urban air quality is closely related to people’s quality of life and sustainable economic development [2]. Improving urban air quality has become an urgent desire of the people and is central to determining the success of sustainable development strategy in China. According to the China Ecological Environment Status Bulletin, in 2023, 136 prefecture-level cities, accounting for 40.1% of all cities, failed to meet environmental standards. The average concentration of PM 10 in the ambient air at background points across the country increased by 14.6% compared to 2022, with industrial sources accounting for the largest share of all pollution sources, especially sulfur dioxide and particulate matter, both of which accounted for more than 60% of the total. Therefore, in order to improve air quality and enhance the capacity for sustainable development, it is necessary to specialize in the pollution of key industries, starting from industrial pollution sources. As an important source of industrial pollution, polluting industries should inevitably become one of the main targets for improving environmental quality and realizing green and sustainable development. How to do a good job of pollution remediation by polluting industries is the research theme and research purpose of this paper.

From an economic growth perspective, industrial agglomeration fosters positive externalities such as labor sharing, input–output linkages, and knowledge spillovers [3,4,5], while industrial decentralization often leads to resource inefficiency and redundant construction [6], proposing agglomeration as a key feature of economic growth and a theoretical basis for urban formation [7]. However, the impact of industrial agglomeration is debated regarding environmental protection. Many scholars argue that in its initial stages, agglomeration exacerbates pollution due to expanded production and external spillovers. Over time, however, this negative effect diminishes as agglomeration promotes technological innovation, resource sharing, and improved infrastructure [8,9], ultimately enhancing environmental quality and playing a positive role in promoting environmental protection and sustainable development [10]. This analysis pertains to general industries. However, polluting industries, as essential intermediate demand-based sectors in economic development and significant sources of industrial pollution, have environmental functions that differ markedly from other industries. This performance involves conundrums. How does the agglomerative development of polluting industries influence urban air quality? Does it played a positive role in promoting environmental protection and sustainable development? Furthermore, as a key tool for environmental protection, environmental regulation plays a critical role in determining the location of polluting enterprises and the agglomeration of polluting industries. What role does environmental regulation play in mediating the relationship between polluting industry agglomeration and urban environmental quality? To address these questions, this study analyzes the relationship between polluting industry agglomeration, environmental regulation, and urban environmental quality.

The marginal contributions of this research are primarily reflected in the following aspects:

Firstly, this study refines the focus of analysis to the level of polluting industries, specifically analyzing whether the agglomeration of polluting industries generates Marshallian externalities, whether such positive externalities can cover the negative impacts of polluting industries’ agglomeration on urban air quality, and whether the environmental regulation of polluting industries can play the role of the “cost effect” and the “Porter’s Hypothesis”. This is the innovation of this paper’s research perspective.

Secondly, regional geo-ecological characteristics should be considered on the basis of pollution emissions when analyzing urban air quality. This study draws on the new economic geography model to integrate polluting industry agglomeration, environmental regulation, and environmental quality, and innovatively constructs a mathematical and theoretical model about the relationship between the three, which is the innovation point of the theoretical part of this paper.

Lastly, by referencing the China Urban Statistical Yearbook and China Environmental Yearbook, we obtained the air quality and other indicators of key monitoring cities of the Ministry of Environmental Protection (MEP). We also obtained the indicators of the output value of the polluting industries and the number of employed persons by summing up the database of China’s industrial enterprises. We also calculated the degree of polluting industry agglomeration in the cities by using the location entropy index, and we studied the interaction of polluting industrial agglomeration, environmental regulation, and their effects on urban air quality by using the IV regression method. This is the innovation of the empirical part of the paper. Overall, this study provides new insights and strategies for improving urban air quality and realizing sustainable economic development from the perspectives of polluting industrial agglomeration and environmental regulation.

The other sections of this research are arranged as follows: the Section 2 offers a brief literature review; the Section 3 develops a theoretical model and proposes hypotheses; the Section 4 explains the econometric model setting, variable selection, and data sources; the Section 5 details the empirical results and heterogeneity analysis, and the Section 6 presents the main findings of this research and relevant policy recommendations.

2. Literature Review

2.1. Polluting Industrial Agglomeration and Air Quality

There is relatively little literature on industrial agglomeration and air quality, and scholars have focused more on the relationship between industrial agglomeration and environmental pollution, analyzing the positive [11,12], negative [13,14], or “U”-shaped [15,16] relationship between the two from different perspectives. Some scholars even think that the relationship between industrial agglomeration and pollution reduction is complicated by the alternating effects of “congestion effect”, path dependence, and structural rigidity [17]. However, the impact of industrial agglomeration on various pollutants is different, and compared with water pollution, air pollution is more diffuse, and the sources are more complex [18], so it is inappropriate to directly apply the conclusion of the analysis between industrial agglomeration and environmental pollution to air pollution. Based on this, some scholars began to pay attention to the relationship between industrial agglomeration and air pollution. There are three main points of view related to the conclusions. First, there is a high positive correlation between industrial agglomeration and air pollution [19,20]. Second, industrial agglomeration is conducive to reducing air pollution, and there is a negative correlation between the two [21,22,23]. Third, when a certain threshold is reached, there is a “U” or “N” shaped relationship between the two [24,25,26]. Different industrial structures correspond to different pollution emission structures, and there are significant differences between different types of industries and the direction of change in pollution emission intensity [27,28], and the impact of different industrial agglomerations on air quality is also different. So how exactly does the spatial concentration of polluting enterprises and the clustering of polluting industries affect air quality? These questions urgently require further study.

2.2. Environmental Regulation, Agglomeration of Polluting Industries, and Air Quality

The relationship between industrial agglomeration and environmental pollution is complex and may involve a variety of influencing mechanisms, such as the existence of a number of “moderating variables” affecting the relationship between the two, with environmental regulation being one of the important “moderating variables” [29,30]. Environmental regulations, such as environmental taxes and regulations, usually raise the cost of production for firms, which makes polluting firms more inclined to cluster in areas with lax environmental regulations [31,32], leading to higher pollution and the formation of pollution havens [33,34]. Although the strengthening of environmental regulations increases the pollution abatement costs and marginal costs faced by enterprises, it also gives them the incentive to reform and innovate environmentally friendly and energy-saving technologies [35], which in turn improves the quality of the ecological environment and brings about an innovation compensation effect [36,37]. The direction of the impact of environmental regulation on industrial agglomeration depends on the relative size of these two roles [38]. Chen et al. (2024) [30] found that stricter environmental regulations lead to a decrease in manufacturing agglomeration and an increase in productive services agglomeration, which reduces air pollution. Polluting industries are both essential intermediate demand industries in the process of economic development and important sources of industrial pollution. Are the environmental effects of their agglomeration different from those of other general industries? Does their agglomeration bring different environmental effects compared to other general industries? Does it exacerbate air pollution or reduce it? What is the moderating role of environmental regulation as a moderating variable on the impact of the agglomeration of polluting industries?

2.3. Review of Studies

Analyzing the above literature, it can be found that there are several deficiencies in the existing studies. Firstly, the analysis of regional industrial agglomeration still remains at the level of general industry, without refining the focus of the analysis to differentiate. However, the environmental effects of different industrial agglomerations are different. Is the environmental effect of polluting industrial agglomerations different from that of general industries? What is the role of environmental regulation for polluting industries? Therefore, this paper refines the analysis to the level of polluting industries and empirically examines their impact on air pollution and the regulating role of environmental regulation. Secondly, previous studies have neglected regional geo-ecological characteristics, while environmental quality is the result of the combination of pollution emissions and regional ecological conditions, so this paper incorporates regional geo-ecological characteristics into a unified framework in the analysis and fully discusses the heterogeneous effects of regional ecological characteristics. Thirdly, analyses in the context of China mostly remain at the inter-provincial level, disregarding the city level. However, cities, as distinct block economies, are fundamentally driven by population and industry agglomerations [39]. As a result, discussions on inter-provincial polluting industry agglomeration in existing studies may lack realistic and persuasive power. To address these gaps, this paper considers the nature of urban eco-geography, and endogenous environmental regulation, environmental regulation, and environmental quality are introduced into the new economic geography model. The IV regression method is used to study the interaction between polluting industry agglomeration and environmental regulation and its impact on urban air quality. Additionally, the impact of polluting industrial agglomeration on urban air quality under the influence of environmental regulations is further analyzed in terms of heterogeneity dimensions such as the level of economic development, the current industrial structure, and geographic and climatic characteristics.

3. Theoretical Modeling and Proposition Formulation

Building on the new economic geography model, this paper conceptualizes environmental regulation as the cost of pollution control for firms. It examines the impact of polluting industrial agglomeration on urban environmental quality and investigates the moderating role of environmental regulation, proposing research hypotheses based on this framework.

3.1. Basic Modeling

This research develops a 2 * 2 * 2 new economic geography model including 2 regions and 2 sectors. The regions encompass local and foreign, and the sectors contain agricultural and industrial sectors. Each region includes both the agricultural sector and the industrial sector. The agricultural sector has constant returns to scale and perfect competition, while the industrial sector has increasing returns to scale and imperfect competition. Only the firms in the industrial sector cause pollution. Production factors are limited to human capital H and labor L. The industrial sector uses both capital and labor factors, while the agricultural sector utilizes only labor factors for production. The local variables in the model are not superscripted, the foreign variables are superscripted with *, and the variables of the whole economic system are superscripted with $w$. The two regions are identical in terms of consumer preferences, production technology, factor endowments, and transaction costs.

The utility function of consumers is $U=C_M^{\mu }{C}_A^{1-\mu }{Q}^{\gamma }$,$C_M={\left(\sum _{i=1}^{n^w}{c_i}^{(\sigma -1)/\sigma }\right)}^{\sigma /(\sigma -1)}$,

Where $C_M$ is the consumption of industrial products,$C_A$ denotes the consumption of agricultural products, Q indicates the quality of the environment, and $0<\mu <1<\sigma , \gamma >0.$

The cost function of the agricultural sector is $w_L{a}_A$. Agricultural products implement the marginal cost pricing method, while agricultural products are in the two places between the non-existence of transaction costs. The cost function of each industrial enterprise is $w+w_L{a}_{M}x$. The consumer’s demand for industrial goods is $c_j=\mu E{p_j}^{-\sigma }/{p_M}^{1-\sigma }$, where E shows the total expenditure.

The derivation shows that the profit function for industrial firms is based on Equation (1).

$$w=b{E}^{w}B/H^{w }$$

(1)

where $b\equiv \mu /\sigma \in \left(\mathrm{0,1}\right)$, $B={\displaystyle \frac{s_E}{∆}}+\mathrm{\varnothing }{\displaystyle \frac{s_E^{*}}{{∆}^{*}}}{s}_E=E/E^w∆=s_H+\varnothing {s_H}^{*}$, can be solved for

$$x=\left(\sigma -1\right)w/w_L{a}_M. \mathrm{Then}, a_M=\left(\sigma -1\right), \mathrm{and} x=w.$$

3.2. Introduction of Environmental Regulation

Environmental regulation is now introduced into the model as the cost of pollution control for firms. Since production by industrial firms causes pollution, the amount of pollution per unit of output is assumed to be $q_0$. If industrial firms are subjected to environmental regulation with a regulatory standard q, which requires each industrial firm to extract ${qq}_{0}w$ from its operating profit to reduce its corresponding amount of pollution, set $q_0<1, q\ge 0$. Assuming that there is no diffusion of pollution between regions, and considering that the amount of pollution from firms can be simply superimposed, we have

$$Q=Q_0+H{q}_0\left(qw-x\right)$$

(2)

where $Q_0$ is the initial environmental endowment. Carrying $x=w$ and (1) into (2) yields $Q=Q_0+H{q}_{0}b{E}^{w}B\left(q-1\right)/H^w$, which in turn has $H=s_H{H}^w$ and carries over to

$$Q=Q_0+q_{0}b{E}^{w}B{s}_H\left(q-1\right)$$

(3)

It can be found that when $q\ge 1$, environmental regulation is sufficient, $Q\ge Q_0$, environmental regulation can not only completely curb pollution but also improve the environmental quality, even at the initial endowment level. We call it sufficient environmental regulation. When $q<1$, environmental regulation is insufficient, $Q<Q_0$. Then, environmental regulation cannot completely curb environmental pollution, and the environmental quality decreases from its initial endowment level. We call it an insufficient level of environmental regulation. Considering the actual situation in China, in the following models of this paper, we set an insufficient level of environmental regulation (i.e., $q<1$), i.e., although enhancing the degree of environmental regulation can improve the environmental quality, it is insufficient to compensate for pollution’s damage to the initial environmental endowment.

Aggregate local consumer spending is the sum of labor wages and capital gains. Thus, we further solve for B:

$$\mathrm{B}={\displaystyle \frac{\left(1+\varnothing \right)+b(1-\varnothing )(2s_H-1)}{2(2s_H-1)(1-\varnothing )(1+\varnothing +b{s}_H)}}$$

3.3. Formulation of Propositions

Bringing equation B into Equation (3) obtains a function of the intensity of environmental regulation q, the share of industrial agglomeration $s_H$, and the environmental quality Q. This function is

$$Q=Q_0+{\displaystyle \frac{q_{0}b{E}^w\left(q-1\right)}{2(\varnothing -1)}}\left\{{\displaystyle \frac{\left[\left(1+\varnothing \right)+b(1-\varnothing )(2s_H-1)\right]s_H}{(2s_H-1)(1+\varnothing +b{s}_H)}}\right\}$$

(4)

Let {} in Equation (4) be X. We can derive $\partial X/\partial s_H<0$. Assuming that $\varnothing$ is not equal to 0 or 1 ($\varnothing$ = 0 means that the market is completely closed, and $\varnothing$ = 1 means that the market is completely open, so it is reasonable to assume that $\varnothing$ is not equal to 0 or 1); therefore, $q_0, b,$ and $\varnothing$ are constant parameters between (0, 1), and due to the previous assumption $q<1$, it can then be shown that

$${\displaystyle \frac{\partial Q}{\partial s_H}}={\displaystyle \frac{q_{0}b{E}^w\left(q-1\right)}{2(\varnothing -1)}}{\displaystyle \frac{\partial X}{\partial s_H}}<0$$

(5)

This leads to

Hypothesis 1.

An increase in the level of agglomeration of polluting industries leads to a deterioration in environmental quality.

Further derivation of Equation (5) with respect to the intensity of environmental regulation q leads to

$${\displaystyle \frac{\partial \frac{\partial Q}{\partial s_H}}{\partial q}}={\displaystyle \frac{q_{0}b{E}^w}{2(\varnothing -1)}}{\displaystyle \frac{\partial X}{\partial s_H}}>0$$

(6)

Since $\partial Q/\partial s_H<0$ and ${\displaystyle \frac{\partial \left(\frac{\partial Q}{\partial s_H}\right)}{\partial q}}>0$, it is assumed that ${\displaystyle \frac{\partial Q}{\partial s_H}}=-3$, and an increase in q will cause ${\displaystyle \frac{\partial Q}{\partial s_H}}$ to increase to $-2$.

This leads to

Hypothesis 2.

An increase in environmental regulation reduces the negative impact of polluting industrial agglomeration on environmental quality.

This paper proposes that an increase in the level of polluting industry agglomeration leads to a linear deterioration in environmental quality. However, heightened environmental regulation mitigates the negative impact of polluting industry agglomeration on environmental quality and, to some extent, alleviates the environmental pollution caused by such agglomeration.

4. Measurement Models, Variables, and Data

4.1. Econometric Modeling

Based on the above theoretical analysis and with reference to the econometric equation of Chen et al. (2018) [40], this research develops Equation (7):

$${api}_{it}={\alpha }_0+{\alpha }_1{agg}_{it}+{\alpha }_2{er}_{it} \ast {agg}_{it}+\beta C_{it}+{\epsilon }_{it}$$

(7)

where i and j denote city and year, respectively. ${api}_{it}$ is the annual average air pollution index of the city used to characterize the air quality of the city. A high ${api}_{it}$ indicates a low level of air quality in the city. ${agg}_{it}$ indicates the concentration of industrial exhaust pollution in the city. A large ${agg}_{it}$ signifies a high concentration of polluting industries. ${\alpha }_1$ measures the direct impact of polluting industries’ concentration on air quality. If ${\alpha }_1>0$, the exacerbation of pollution caused by the degree of concentration of polluting industries will increase the pollution index of the city and degrade the air quality, which is in line with the theoretical hypothesis. ${er}_{it}$ represents environmental regulation, ${\displaystyle \frac{\partial \left(api\right)}{\partial \left(agg\right)}}={\alpha }_1+{\alpha }_2\left(er\right)$. If ${\alpha }_2<0$, then the increase in environmental regulation will reduce the deterioration of air quality by the agglomeration of polluting industries, and the theoretical hypothesis is further verified.

To check the robustness of the estimation results, the research selects the following control variables: $C_{it}$: ① urban economic growth, measured by GDP per capita (${\mathrm{l}\mathrm{n}pgdp}_{it}$), and its square term to control the non-linear impact of economic growth on environmental quality, which examines the environmental Kuznets curve; ② relevant indicators of urban development status, including population density (${\mathrm{l}\mathrm{n}densp}_{it}$), industrial structure (${\mathrm{l}\mathrm{n}str}_{it}$) measured by the proportion of output value of the tertiary industry, the level of scientific and technological development (${\mathrm{l}\mathrm{n}tec}_{it}$) measured by the number of scientific and technological employees in the city, and the degree of opening up to the outside world (${\mathrm{l}\mathrm{n}fdi}_{it}$) measured by the amount of foreign capital actually utilized by the city. The above data mainly come from the China Urban Statistical Yearbook; ③ indicators of the city’s characteristics, including whether it is a provincial capital city (${capital}_i$), whether it is a southern city (${south}_i$), and whether it is a coastal city (${coastal}_i$), which are variables that do not change over time.

4.2. Data Sources and Sample Selection

The data sources of this paper mainly involve three databases: the China Industrial Enterprises Database, the China Urban Statistical Yearbook, and the China Environmental Yearbook. The relevant statistical yearbooks do not publish the relevant indicators of the subdivided industries of each prefecture-level city, while the China Industrial Enterprises Database provides informative micro-enterprise production data. Therefore, some of the indicator data in this paper can be organized by the China Industrial Enterprises Database.

In this paper, drawing on Brandt et al. (2012) [41] and Bai et al. (2009) [42], the missing values and outliers in the industrial enterprise database are processed as follows: First, samples with missing variables are removed, such as samples with missing, zero, or negative sample values for major variables such as gross industrial output value and value added of industry. Second, samples of firms with age less than 0 as well as greater than 210 are deleted in order to ensure the validity of the age variable of the firms. Third, samples with R&D and innovation expenditures of less than 0 are deleted. On the one hand, the statistics in the China Industrial Enterprises Database are as from 2013. On the other hand, China’s Ministry of Environmental Protection (MEP) started to monitor the air quality conditions of key cities in 2001, and since then, the number of monitored cities has been increasing year by year, and the MEP changed the monitoring standards and statistical calibers in 2013. Therefore, this paper examines an unbalanced panel of MEP-monitored cities from 2001 to 2012.

The database of China’s industrial enterprises covers a wide range of indicators and has an irreplaceable role in the study of industrial agglomeration. Data from 2013 and before are scarce and representative for the study of a specific historical period, and during the time period studied in this paper, China was in the stage of rapid industrialization and economic transformation, in which there were features of unbalanced development among regions, and the industrial structure needed to be upgraded urgently. By analyzing the data of enterprises in different industries, we can provide data support for the relevant theoretical research in a specific period and further improve the theoretical system of the relationship between industrial agglomeration, environmental regulation, and air pollution. The findings of this paper can provide reference for some developing countries in the early stage of industrialization.

4.3. Description of Indicators

4.3.1. Urban Air Quality

The explanatory variable in this study is the annual average Urban Air Pollution Index ${api}_{it}$. The index comes from the data center of China’s Ministry of Environmental Protection and is used to measure the quality of urban atmospheric environment. Daily API data are averaged to calculate the annual API. The API simplifies the concentrations of routinely monitored air pollutants—such as smog, suspended particulate matter, nitrogen dioxide, sulfur dioxide, and carbon monoxide—into a single index value. This index reflects both the degree of air pollution and the overall air quality, which result from the combined effects of urban industrial pollution emissions and ecological and climatic conditions.

4.3.2. Polluting Industry Agglomeration

The core explanatory variable of this paper is the degree of polluting industry agglomeration ${agg}_{it}$, which is used to explain the air pollution index. To match this variable, when selecting polluting industries, industrial emissions should be taken as a criterion. The “first national pollution source census bulletin”, jointly announced by 2010 MEP and the Bureau of Statistics, pointed out that sulfur dioxide production and emissions accounted for the largest proportion of the main pollutants in industrial exhaust. In addition, sulfur dioxide is emitted from the top six industries for the production and supply of electricity and heat: the non-metallic mineral products industry, the ferrous metal smelting and rolling industry, the chemical raw materials and chemical products manufacturing industry, the non-ferrous metal smelting and rolling processing industry, the petroleum processing and coking industry, and nuclear fuel processing industry. These six industries emit 88.5% of sulfur dioxide emissions. Based on their large contribution, this paper selects the six industries as air polluting industries.

Drawing on O’Donoghue et al. (2004) [43], location entropy is used to measure polluting industry agglomeration. There are two ways to use location entropy to measure the degree of agglomeration of polluting industries ${agg}_{it}$: (1) location entropy measured by the output value ${agg1}_{it}$ = (output value of polluting industries in this city/total industrial output value of this city)/(output value of polluting industries in the whole country/total industrial output value of the whole country); and (2) location entropy measured by the number of employees ${agg2}_{it}$ = (number of employees employed by polluting industries in this city/total number of employees employed by the whole city)/(total number of employees employed by polluting industries in the whole country/number of employees employed by the whole country). This paper uses the degree of agglomeration measured in terms of output value ${agg1}_{it}$ and ${agg2}_{it}$ as a robustness test.

4.3.3. Measurement of Environmental Regulation and Its Endogeneity Treatment

The core explanatory variable in this paper is environmental regulation ${er}_{it}$. The measurement of environmental regulation has been widely discussed in the academic literature. Scholars typically approach it from two perspectives. The first one is environmental governance, including inputs such as investment in environmental management, urban infrastructure construction, and operating costs of three-waste management facilities [31], as well as outputs like the industrial sulfur dioxide removal rate or composite indices for pollutant treatment [44]. The second perspective is environmental regulation, including indicators such as environmental legislation, policy documents, and the number of environmental administrative penalties [45].

Given the availability of data, this paper adopts the following two measures. ① One is the level of environmental regulation based on governance inputs measured by the cost of industrial exhaust gas governance ($\mathrm{l}\mathrm{n}{er1}_{it}$). The China Environmental Yearbook has published data on industrial pollution control in key cities since 2001, which includes the cost of operating exhaust gas control facilities. This cost encompasses energy consumption, equipment depreciation, maintenance, personnel wages, management fees, pharmaceuticals, and other related expenses, representing the total cost incurred by industrial enterprises in the city for treating industrial exhaust gases (environmental regulation for gas). To align with the concentration of polluting industries, this approach adjusts the measure using the following equation: ${er1}_{it}$=${erall}_{it}$ * (polluting industry output value/total industrial output value). ② Based on the effect of environmental governance ${\mathrm{l}\mathrm{n}er2}_{it}$, the industrial sulfur dioxide removal rate is used as another indicator to measure the level of environmental regulation. In the benchmark regression, environmental regulation is measured by governance inputs $\mathrm{l}\mathrm{n}{er1}_{it}$, while ${\mathrm{l}\mathrm{n}er2}_{it}$ serves as a robustness test variable.

An important issue in the study of environmental regulation is endogeneity. The intensity of environmental regulation is endogenous to both environmental quality and economic development levels. Cities with higher environmental quality lack the incentive to combat pollution, while economically underdeveloped cities lack the capacity for effective regulation. To address this endogeneity, many scholars introduce a one-period lag of environmental regulation as an instrumental variable, estimating it using two-stage least squares (2SLS) [46] or instrumental variables (IV) [36]. This research addresses the endogeneity of environmental regulation using two approaches: first, by lagging one period for OLS regression, and second, by considering the lagged variable as an instrumental variable in 2SLS regression. The reason is that lagging the OLS regression by one order alleviates some of the endogeneity problem, but it may not solve it completely. When there are other unobserved factors that affect both environmental regulation and the explanatory variables, resulting in omitted variable bias, 2SLS regression utilizing instrumental variables can be more effective in addressing the endogeneity problem.

4.4. Descriptive Statistics for Indicators

4.4.1. Organization of Research Objects

Since the number of cities monitored by the MEP gradually increased from 2001 to 2012, and the China Environmental Yearbook published data for 112 key cities for environmental protection, we match these two datasets to obtain the unbalanced panel city data.

Table 1. Number of cities visited in different years.

Particular Year	2001–2003	2004–2010	2011	2012	(Grand) Total
Number of cities	46	79	104	105	900

represents the number of cities examined each year.

4.4.2. Descriptive Statistics

Table 2. Descriptive statistics for each variable.

Elemental Indicators	Variant	Name (of Variable)	Sample Size	Average Value	(Statistics) Standard Deviation	Minimum Value	Maximum Values
environmental quality	air pollution index	api	900	71.79765	16.45496	30.19718	160.3705
Polluting industry agglomeration	Measured by industrial output	agg1	900	1.076527	0.5026508	0.128429	2.586914
	Measured in terms of employment	agg2	900	1.157152	0.5566584	0.2863047	3.759452
Intensity of environmental regulation	Polluting Industries Exhaust Treatment Costs	er1	900	15,646.62	30,926.78	10.84402	454,000.8
	Industrial sulfur dioxide removal rate	er2	900	0.4121832	0.2368057	0	0.9302789
control variable	GDP per capita	pgdp	900	45,783.99	25,492.35	6514	152,099.2
	population density	densp	900	1319.905	1024.097	31.15	11,449.3
	industrial structure	str	900	46.44583	11.27078	8.58	78.66
	technological level	tec	900	2.353328	5.141176	0.01	53.78
	Egyptian openness to the outside world	fdi	900	121,558.9	203,325.8	0	1,518,453

shows the descriptive statistics of the main variables. The air pollution index, which characterizes environmental quality, and the two indicators measuring the degree of agglomeration of polluting industries are relative indicators with small standard deviations. Therefore, their original values are directly used in the regression. In contrast, other indicators exhibit large variances; hence, their logarithmic forms are employed in the regression to reduce heteroscedasticity and enhance the robustness of the results. The baseline regression uses industrial output value to measure the agglomeration of polluting industries (agg1) and the cost of exhaust treatment for polluting industries to measure the intensity of environmental regulation (er1). It also uses agg2 and er2 as robustness tests. In addition, the control variables in the regression include three city characteristic indicators, including whether the city is a provincial capital city (${capital}_i$), whether it is a southern city (${south}_i$), and whether it is a coastal city (${coastal}_i$).

To further understand the characteristics of the core variables in each city, this paper selects eight representative cities and averages the 12-year data for their air pollution index, polluting industry agglomeration, and environmental regulation intensity, as shown in

Table 3. Mean values of core variables for major cities.

City	api	agg1	agg2	er1	er2
Haikou prefecture level city in Hebei	36.2725	0.381823	0.571348	104.5238	0.357221
formerly known as Foochow or Fuchow	58.60378	2.160817	2.071754	26,539.89	0.296653
Wuhan city on Changjiang, subprovincial city and capital of Hubei province	85.28354	1.230619	1.429553	19,644.64	0.372614
Taiyuan prefecture level city in Hebei	94.18916	2.055336	1.677765	47,376.92	0.454339
capital of People’s Republic of China	96.06697	0.976502	0.974894	21,154.23	0.429784
Lanzhou prefecture level city in Zhejiang	113.0925	2.188301	2.076759	42,225.93	0.242694

. Among these cities, Haikou has the highest air pollution index, but its polluting industry agglomeration is relatively low. Due to its unique geographic location and climatic conditions, Haikou’s environmental quality remains high despite the low intensity of environmental regulation. Consequently, the cost of environmental management in Haikou is also lower. Fuzhou City is unique in having a high concentration of polluting industries but a lower-than-average air pollution index. This may be partly due to its environmental regulatory efforts and partly due to its favorable geographic location, which aids in the dispersion of pollutants. Wuhan and Beijing, as representative regional core cities, have medium levels of polluting industry concentration and strong government financial support for environmental regulation. As a result, differences in environmental quality between these cities are likely attributed to variations in industrial organization, technological innovation, and geographic location. In contrast, Taiyuan and Lanzhou, representative of cities with a high concentration of polluting industries, experience worsened air quality due to this over-concentration. However, increased investment in environmental governance and stronger environmental regulation can partially mitigate these negative effects. In conclusion, the environmental performance of cities is not solely dependent on the concentration of polluting industries and the intensity of environmental regulation; it is also closely related to economic development, industrial structure, and geographic and climatic characteristics. This understanding underpins the selection of variables in this paper.

5. Empirical Results and Analysis

5.1. Baseline Regression Results

Figure 1 displays a scatterplot showing the relationship between the concentration of polluting industries in a city (measured by output value) and its air pollution index. The plot suggests a linear positive correlation between the two variables. The key questions are whether this relationship is statistically significant and how the intensity of environmental regulation affects this relationship?

To address these questions, this paper first employs panel fixed-effects regression to provide preliminary estimates of the impact of polluting industry agglomeration on the air pollution index, as shown in columns (1) and (2) of

Table 4. Baseline Regression Results for Polluting Industry Agglomeration and Air Pollution under Environmental Regulation.

	(fe)	(fe *)	(fe_L)	(fe_iv)	(re_iv)	(re_xthtaylor)
	(1)	(2)	(3)	(4)	(5)	(6)
VARIABLES	api	api	api	api	api	api
agg1	3.657 **	13.49 **	8.378 *	12.54 ***	12.72 ***	14.25 ***
	(1.633)	(6.543)	(4.709)	(4.604)	(4.195)	(3.365)
lner1× agg1		−1.167 *		−1.134 **	−1.100 ***	−1.209 ***
		(0.643)		(0.453)	(0.397)	(0.314)
L.lner1× agg1			−0.751 *
			(0.452)
lnpgdp	−68.16 ***	−58.53 **	−45.35	−40.92 **	−36.23 **	−57.01 ***
	(16.81)	(25.06)	(31.47)	(18.40)	(17.66)	(15.94)
lnpgdp2	3.034 ***	2.531 **	1.895	1.712 **	1.417 *	2.427 ***
	(0.806)	(1.192)	(1.492)	(0.869)	(0.831)	(0.753)
lndensp	−1.876	−0.985	0.837	0.913	3.608 ***	1.618
	(1.485)	(2.238)	(2.098)	(1.493)	(0.990)	(1.215)
lnstr	−12.72 ***	−12.50 **	−12.34 **	−12.21 ***	−11.08 ***	−11.42 ***
	(3.445)	(6.011)	(5.398)	(3.441)	(2.615)	(2.932)
lntec	−0.372	1.254	−0.179	−0.0443	2.694 ***	2.177 ***
	(1.261)	(0.918)	(1.588)	(1.292)	(0.908)	(0.716)
lnfdi	−2.456 ***	−2.592 ***	−2.163 ***	−2.120 ***	−1.038 ***	−1.874 ***
	(0.447)	(0.852)	(0.785)	(0.415)	(0.358)	(0.391)
capital					6.765 **	10.71 ***
					(2.980)	(4.142)
south					−9.147 ***	−9.596 ***
					(1.930)	(3.452)
coastal					−8.581 ***	−7.060 *
					(2.282)	(4.058)
Constant	541.4 ***	485.0 ***	400.3 **	371.9 ***	326.0 ***	453.9 ***
	(91.11)	(140.8)	(175.8)	(101.6)	(95.34)	(86.26)
Huasman				23.84	49.10
				[0.0024]	[0.0000]
Observations	900	900	794	794	794	900
R-squared	0.362	0.347	0.309	0.307	0.430
Number of city	105	105	105	105	105	105

Notation: Robust standard errors in (), p-values of the corresponding statistics in [ ].***, **, and * indicate passing significance tests at 1%, 5%, and 10% significance levels, respectively.

. Regression (1) includes only the core explanatory variable of polluting industry agglomeration, while Regression (1) adds environmental regulation intensity. The results show that the estimated coefficients for the degree of polluting industry agglomeration are positive and statistically significant at the 5% level, indicating that greater agglomeration of polluting industries exacerbates environmental pollution. Furthermore, the coefficient of the interaction term between environmental regulation and polluting industry agglomeration is negative in Regression (2). Although this result is significant only at the 10% level, it suggests that an increase in the level of environmental regulation can mitigate the environmental pollution caused by industry agglomeration. These findings provide initial support for the theoretical hypotheses proposed in this paper.

The regression coefficients of the control variables indicate that optimizing the city’s industrial structure and increasing its level of openness to the outside world reduce the air pollution index and improve environmental quality, with the former having a more significant effect. Since the benchmark regression adopts a panel fixed-effects model, city characteristic indicators that do not change over time are not included in the regression. Additionally, the benchmark regression does not account for the endogeneity of environmental regulation. Scholars often address this issue by using the lagged term of environmental regulation intensity as an instrumental variable. To address this, regression results (3) in Table 4 introduce the lagged term of environmental regulation intensity into the fixed-effects model. The results are consistent with the benchmark regression, showing that an increase in environmental regulation intensity in the current period can reduce the environmental pollution caused by polluting industry agglomeration in the subsequent period.

5.2. Instrumental Variable (IV) Regression Results

This paper uses the ivxtreg2 command to examine the relationship between the instrumental variables and the endogenous variables in Regression (4), as well as to test for weak identification and over-identification issues. The results confirm that the selected instrumental variables are valid and appropriate. The estimation results in Regression (4) reveal that the coefficient for polluting industry agglomeration is positive and statistically significant at the 1% level, indicating that the agglomeration of polluting industries exacerbates environmental pollution. Additionally, the coefficient for the interaction term between environmental regulation and polluting industry agglomeration is negative and significant at the 5% level, with an improved significance level compared to prior models. This result suggests that while polluting industry agglomeration worsens environmental quality, an increase in environmental regulation can mitigate the pernicious environmental impact of such agglomeration.

Column (5) of Table 4 reports the regression results for the instrumental variable method under the random effects model. The random effects model was chosen because the instrumental variables approach requires that the instrumental variables are uncorrelated with the error term, satisfying the exogeneity assumption. The random effects model helps to satisfy this assumption to some extent. Because the random effects model treats individual heterogeneities as random factors, the exogeneity of the instrumental variables is better ensured when they are uncorrelated with these random factors. The results show that the coefficients of the core explanatory variables do not change significantly, and the estimated coefficients of the capital, south, and coastal dummy variables are positive, negative, and negative, respectively. This suggests that the air pollution indices of provincial capitals and municipalities are significantly higher than those of other cities, southern cities are significantly less polluted than northern cities, and coastal cities have air quality significantly higher than that of inland cities. These findings may be attributed to the obvious differences among provincial capitals and municipalities, which tend to gather a large number of people and industries, the heavy reliance on coal energy for heating during winter in northern cities, and the favorable diffusion conditions and ecological characteristics of coastal cities.

However, the results of the Huasman test show that the regression results of the instrumental variables approach under the random effects model are inconsistent. Therefore, this paper further employs the Huasman–Taylor (1981) method using the xthtaylor command to perform the regression, with the results reported in Column (6) of Table 4. The estimation results show that the coefficients of the core explanatory variables remain largely unchanged, with the level of significance further increased. This finding confirms that the agglomeration of polluting industries degrades environmental quality, while an increase in the level of environmental regulation mitigates the environmental pollution caused by such agglomeration. These results validate the theoretical assumptions of this paper. A comprehensive analysis of the regression coefficients of the control variables across all models further reveals that optimizing the industrial structure of a city and increasing its level of openness to the outside world reduce the air pollution index and improve environmental quality. Additionally, the air pollution indices of provincial capital cities and municipalities directly under the central government are significantly higher than those of other cities. Pollution levels in southern cities are notably lower than in northern cities, and the air quality of coastal cities is significantly higher than that of inland cities.

In addition, this paper employs the instrumental variable method to test the nonlinear effect of polluting industry agglomeration on environmental quality by including the squared and cubic terms of agg1 in the regression analysis (the results are not shown in the text due to space constraints). The results show that, within the scope of the sample, there is no significant non-linear relationship between the agglomeration of polluting industries and the quality of the environment, implying that there is no possibility of an “inflection point” at which the degree of impact is slowed down or even improved. The possible reason for this is that, in order to attract more enterprises to cluster, local government environmental and industrial management departments often adopt a low-standard threshold policy for enterprise pollution emissions, and the public facilities and means to control pollution and reduce emissions are not perfect, so that the advantages of joint emission reduction have not been effectively utilized. Even though agglomerative development may generate positive externalities such as knowledge spillover and equipment sharing, such externalities cannot offset the environmental pollution brought about by the agglomeration of polluting industries, and therefore the agglomeration of polluting industries shows a linear negative correlation with the quality of the urban environment, and there is no inflection point for environmental pollution.

5.3. Robustness Tests

To obtain more robust research conclusions, this paper conducts robustness testing by replacing the core explanatory variables, as shown in

Table 5. Robustness test.

	(fe_iv)	(re_iv)	(fe_iv)	(re_iv)
	(1)	(2)	(3)	(4)
VARIABLES	api	api	api	api
agg2	34.70 ***	39.42 ***
	(11.72)	(12.90)
lner1× agg2	−3.821 ***	−3.948 ***
	(1.227)	(1.266)
agg1			43.31 ***	45.33 ***
			(14.36)	(15.51)
lner2× agg1			−4.362 ***	−4.316 ***
			(1.495)	(1.527)
control variable	Y	Y	Y	Y
Observations	794	794	794	794
R-squared	0.170	0.187	0.175	0.209
Number of city	105	105	105	105

Note: Robust standard errors within (). ***, indicate significance at the 1% level.

. In Table 5, Columns (1) and (2) replace the measure of polluting industry agglomeration from output value to employment-based characterization. Column (1) presents the estimation results of the instrumental variable method under fixed effects, while Column (2) shows the results under random effects. The direction of the coefficient estimates does not change significantly. The reason for the different values of the two regression results is that fixed effects focus only on the net effect of the explanatory variables on the explanatory variables and exclude inter-individual differences, which will make the estimated coefficients relatively small. Random effects will include a part of inter-individual differences in the estimation as well, which will lead to a larger coefficient of the estimated coefficients. These differences also indirectly suggest that external factors moderate the relationship between polluting industry agglomeration and air pollution. In the results reported in Columns (3) and (4), the characterization of polluting industry agglomeration by output value remains unchanged, but the measure of input-type environmental regulation is replaced with effect-type environmental regulation. The regressions are conducted using the instrumental variable method under fixed and random effects, respectively. Based on the estimations, the negative impact of employment-based polluting industry agglomeration on environmental quality is stronger than that of output-based agglomeration. Furthermore, environmental regulation characterized by the sulfur dioxide removal rate is more effective in improving environmental quality compared to regulation based on exhaust gas pollution control. The direction of the estimated coefficients remains consistent and significant at the 1% level across all models, further validating the findings of this paper.

5.4. Heterogeneity Analysis

The analysis above confirms the significant deterioration in urban environmental quality resulting from increased polluting industrial agglomeration across the entire sample. It also demonstrates that enhanced environmental regulation mitigates the negative impact of polluting industrial agglomeration on environmental quality, improving pollution levels to a certain extent. However, cities vary in their stages of development and geo-ecological characteristics, which influence these relationships. Based on the regression results in Table 4, cities have significant differences in air quality. These variations may be attributed to factors such as the concentration of people and industries in provincial capitals, the high consumption of coal energy during winter in northern cities, and the favorable diffusion conditions and ecological advantages of coastal cities. Given these differences, it is essential to categorize cities based on their characteristics to further explore the relationship between the core variables.

Table 6. Regression results by urban characteristics.

	Provincial Capitals (Including Municipalities Directly Under the Central Government)	Other Cities	agg1 > 1.07	agg1 < 1.07	Coastal Cities	Landlocked Cities	Southern Cities	Northern Cities
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
variant	api	api	api	api	api	api	api	api
agg1	23.99 ***	12.62 ***	19.65 ***	2.446	20.47 ***	28.02 ***	17.45 ***	30.09 ***
	(4.810)	(4.464)	(4.835)	(6.732)	(4.469)	(4.364)	(3.091)	(5.777)
lner1× agg1	−2.008 ***	−0.458	−1.734 ***	0.999	−0.840 **	−2.337 ***	−1.666 ***	−2.404 ***
	(0.439)	(0.430)	(0.425)	(0.679)	(0.402)	(0.414)	(0.296)	(0.520)
control variable	Y	Y			Y	Y	Y	Y
observed value	348	552	352	548	259	641	469	431
Number of cities	29	76	63	73	24	81	51	54

Note: Robust standard errors within (). ***, **, indicate significance at the 1%, 5% levels, respectively.

reports the regression results stratified by city characteristics, providing insights into how these factors differ across various city types.

Columns (1) and (2) of Table 6 report the regression results for provincial capital cities (including municipalities directly under the central government) and other cities, respectively. The findings indicate that polluting industrial agglomeration in municipalities and provincial capitals leads to more severe environmental pollution. Moreover, environmental regulations in provincial capitals (including municipalities) are more effective in ameliorating pollution, whereas regulations in other cities are less effective in addressing agglomeration-related pollution. The reasons for this difference are twofold. First, as regional centers, capital cities (including municipalities directly under the central government) tend to attract large populations and industries. Polluting industries, such as intermediate demand industries, exhibit strong correlations with final demand, thereby causing more significant pollution in these cities. Second, the stronger financial capacity of provincial capital cities (including municipalities directly under the central government) [47] determines their stronger ability to combat environmental pollution. This financial advantage, combined with the diminishing marginal returns effect of environmental control, enables these cities to achieve notable reductions in emissions. This conclusion is further supported by the regression results in Columns (3) and (4) of Table 6. These results indicate that cities with higher-than-average levels of polluting industry agglomeration (1.07) experience more severe pollution but also benefit from stronger governance. Conversely, cities with lower levels of polluting industry agglomeration do not exhibit the same patterns. These findings suggest that polluting industries should be decentralized as much as possible. For cities with significant polluting industry agglomerations, efforts should focus on reducing such agglomeration and enhancing environmental quality through stronger environmental regulation.

Columns (5) and (6) of Table 6 report the regression results for coastal and inland cities, respectively. The results show that the agglomeration of polluting industries in inland cities leads to a more severe deterioration of urban environmental quality, and environmental regulations have a stronger ameliorating effect on pollution. Columns (7) and (8) of Table 6 report the regression results for southern and northern cities, respectively. The findings indicate that, in northern cities, the agglomeration of polluting industries causes a more significant deterioration of urban environmental quality, and environmental regulations are more effective in mitigating pollution. Relatively speaking, the environmental performance of cities in the southern and coastal cities is better. The possible reasons for this are that, on the one hand, the ecological conditions in the southern cities are better, with a stronger carrying capacity, which is conducive to the absorption and diffusion of pollutants, while the ecological conditions in the northern and inland cities are not conducive to the absorption and diffusion of pollutants, and thus the agglomeration of polluting industries in these cities has resulted in more serious environmental damage. On the other hand, northern cities traditionally have a higher concentration of heavy industry, urban pollution problems are more prominent, the government, enterprises, and the public have a stronger demand and willingness for pollution control, and the media report more frequently on environmental pollution in northern cities, which can form public opinion pressure and push the government and enterprises to take active action to control pollution, so as to make the environmental regulation policy better implemented. At the same time, strict environmental regulations have prompted northern enterprises to adopt more advanced, cleaner production technologies and pollution control equipment, which have reduced emissions more significantly and effectively improved air quality.

6. Conclusions and Policy Implications

This paper introduces environmental regulation and environmental quality into the new economic geography model to theoretically explore the impact of polluting industrial agglomeration on urban environmental quality and the role of environmental regulation in this process. Based on the theoretical analysis, the paper focuses on key monitoring cities of the MEP. The study employs the instrumental variables method to examine the effects of industrial exhaust pollution, industrial exhaust environmental regulation levels, and their interactions on urban environmental quality in these key cities while considering the relevant characteristics of the cities and the endogenous nature of environmental regulation. The following conclusions are drawn: The significant feature of polluting industries is that an increase in the level of agglomeration will linearly degrade urban environmental quality, with no non-linear relationship observed. It means that there is no possibility of an “inflection point” where the impact slows down or even improves. However, improvements in environmental regulation can reduce the negative impacts of polluting industries’ agglomeration on urban environmental quality, and, to a certain extent, help mitigate the environmental pollution caused by the agglomeration of polluting industries.

The main policy implications of the article are as follows. (1) From the perspective of sustainable development, unlike general industries where agglomeration effects can be pursued, polluting industries should be decentralized as much as possible rather than clustered in certain key cities. The government should formulate a series of policies to encourage the decentralized development of polluting industries, such as granting tax incentives and financial subsidies to enterprises that build factories in remote or environmentally large-capacity areas, so as to reduce the operating costs of enterprises in those areas and increase their motivation. At the same time, in regional planning, the functional positioning and industrial development direction of different regions should be clarified to guide the rational distribution of polluting industries. Northern and inland cities should be more likely to decentralize the development of polluting industries, and the degree of environmental regulation should be more intensive. According to the environmental carrying capacity, resource endowment, and infrastructure conditions of each region, the type and scale of polluting industries suitable for development should be determined, so as to avoid over-concentration in specific regions. This will help protect the public space and ecological corridors of the city, increase the green space and open space in the city, and enhance the quality of life of urban residents, which is in line with the vision of Sustainable Development Goals 11.

(2) Cities that already have a high concentration of polluting industries should improve their environmental quality by means of increased environmental regulation. More detailed, stringent, and operational local environmental regulations and policies should be formulated, taking into account the city’s own situation and clarifying the standards for the discharge of various types of pollutants and the penalties for non-compliance. At the same time, special scientific research funds should be set up to encourage enterprises to carry out research and development of pollution control technology and clean production technology to improve the environmental performance of polluting industries. With the help of big data, artificial intelligence, satellite remote sensing, and other technological means, the production and business activities and environmental behavior of polluting industries are intelligently supervised, so as to detect and warn of environmental problems in a timely manner and improve the efficiency and accuracy of supervision. Precise environmental management can help improve environmental performance across the industry, reduce the impact of pollutant emissions on the climate system, and contribute to the achievement of greenhouse gas emission reduction and climate adaptation targets. This is consistent with the vision of Sustainable Development Goals 13. However, in this process, attention needs to be paid to the strength and efficiency of policy implementation, the game of interests of various parties, and the cumbersome administrative process.

(3) The optimization of industrial structure has a significant effect on improving urban environmental quality. The study shows that an increase in the proportion of the tertiary sector in the economy, as well as a rise in the actual utilization of foreign capital, reduces the air pollution index and improves urban air quality. A higher proportion of the tertiary sector means that final-demand industries replace a portion of pollution-intensive industries, which in turn reduces the concentration of polluting industries and enhances sustainable development capacity. Therefore, on the one hand, steady growth in the advanced manufacturing sector should be promoted, and changes and improvements in energy utilization patterns should be accelerated. Through technological progress, pollution emissions from equipment should be reduced, and an industrial structure of high development and low emissions should be realized. On the other hand, there is a need to vigorously promote the growth of the tertiary sector, in particular, the development of modernized service industries, and to promote the harmonious development of the economy and the ecological environment.

7. Discussion

We compared the findings of this study with those of the existing literature using similar methods. On the one hand, they have two main similarities: First, it can be confirmed that industrial agglomeration has a significant impact on air pollution, which is consistent with the studies of related scholars [20,21,22,23]. Second, environmental regulation has a positive effect on air pollution, which is consistent with the findings of existing studies [35,36,37]. On the other hand, there are three differences: First, this paper reveals that there is a negative linear relationship between polluting industry agglomeration and air quality, rather than a nonlinear relationship [24,25,26]. Second, this paper accurately identifies and tests the moderating effect of environmental regulations and finds that the moderating effect of environmental regulations on the agglomeration of polluting industries and air quality exhibits spatial heterogeneity. This research perspective and conclusion provide new insights into understanding the spatial effects of environmental regulation. Third, most of the existing studies have neglected the impact of urban eco-geographical features. By incorporating regional geo-ecological features into a unified analytical framework, this paper can more accurately capture the spatial dependence between polluting industry agglomeration, environmental regulation, and air quality, which will help polluting industries to be more flexibly located in urban planning, reduce the burden of cities in infrastructure construction and maintenance, improve the overall operational efficiency and stability of urban infrastructure, and enhance the sustainable development of cities.

During the time period studied in this paper, China was in the stage of rapid industrialization and economic transformation, during which there were characteristics of unbalanced development among regions and an urgent need to upgrade the industrial structure. The policy recommendations on the layout of polluting industries derived from this paper are also of some reference significance to the decision-making behavior of the governments of the less-developed countries that are engaged in industrial transfer, and can provide reference for some developing countries in the early stage of industrialization.

Despite the insightful conclusions obtained in this study, the following limitations remain, which will be further improved in future research endeavors:

Firstly, the limitation of sample selection. We use the panel data of prefecture-level cities in China as the study sample, without considering other countries or regions, and the conclusions we draw are still limited, although they can provide some lessons for other developing countries. Air quality is not only a concern for China, but also an environmental problem that other countries are committed to solving. Therefore, future research could expand the sample to collect relevant data from other typical developed or developing countries, conduct empirical tests, and compare them with the results of this study to obtain richer policy recommendations.

Secondly, the limitations of the data used. Restricted by the year of China’s industrial enterprise database, the time interval of this paper’s research is 2011–2012. Some research with high timeliness requirements, such as the research on emerging industries and market dynamic changes, may not be able to obtain the most real-time data, which affects the timeliness and relevance of the research. Therefore, future studies can further update the year and extend the study period.

Thirdly, the spatial and temporal characteristics of polluting industry agglomeration and air pollution can be further analyzed. This paper mainly pursued study from the perspective of theoretical model construction and empirical tests and lacked the analysis of the spatial and temporal evolution characteristics of polluting industrial agglomerations and air pollution. Therefore, in the future, the data can be further collected and graphically plotted using ARCGIS 10.2 software to show the characteristics of their spatial and temporal distribution changes, which will help us to more accurately identify the heterogeneous effects of polluting industrial agglomerations and environmental regulation on air pollution.