A View of Industrial Agglomeration, Air Pollution and Economic Sustainability from Spatial Econometric Analysis of 273 Cities in China

Abstract

Industrial agglomeration in a region changes the economic structure, strategic layout and resource status of a city, and has an important impact on sustainable economic development. The relationship between industrial agglomeration, air pollution and economic sustainability is a key issue concerning the high-quality development of national economy. China is a developing country that once experienced severe air pollution. Now, the Chinese government is aiming to achieve the goal of sustainable and high-quality economic development in China. In this paper, a spatial Dubin model was developed to study the relationship between industry, environment and the economy. The statistical analysis used the air pollutant data of 273 prefecture-level cities in China from 2015 to 2018. The results showed that: (1) there was a positive U-shaped nonlinear relationship between industrial agglomeration and sustainable economic development, and there was a spatial spillover effect. (2) There was a positive U-shaped nonlinear relationship between air pollution and sustainable economic development, and there was a spatial spillover effect between them. (3) The effect of industrial agglomeration on sustainable economic development was influenced by air pollution, an intermediary variable. The existence of air pollution weakens the promoting effect of industrial agglomeration on sustainable economic development.

1. Introduction

Industrial development has always been a fundamental, overarching and strategic issue facing human society [1]. As a carrier of economy, information, technology and capital, industrial agglomeration in a region can change the economic structure, strategic layout and resource status of a city [2], and can have an important impact on sustainable economic development [3]. Data from the National Bureau of Statistics show that over the past 40 years of reform and opening up, China’s economy has been developing rapidly [4,5]. Its Gross Domestic Product (GDP) increased from RMB 367.87 billion in 1978 to RMB 114,366.97 billion in 2021, and the proportion of China’s economic aggregate in the world economic aggregate has continued to rise. The sudden COVID-19 crisis in early 2020 had a huge impact on the global economy, with the global economy shrinking by 4.4% and experiencing the worst recession since World War II [6]. However, China bucked the trend and became the only major economy in the world to achieve positive economic growth. In 2021, the global GDP growth rate was 5.5%, and China’s GDP growth rate was 8.1%, far outpacing other countries. However, with the rapid development of the economy, air pollution has always existed behind it, which has become an unavoidable problem in the process of sustainable economic development. In 2021, 121 of China’s 339 cities at the prefecture level or above exceeded the standard for air quality, accounting for 35.7 percent of the total, according to the Bulletin on the State of China’s Ecological Environment released by the Ministry of Ecology and Environment. Despite great improvements in air quality, solving the problem of air pollution and promoting sustainable economic development are still major tasks that need to be addressed to achieve the goal of sustainable and high-quality economic development in China. Throughout the development history of all countries in the world, environmental pollution, especially air pollution, is inextricably linked with economic development [7]. Among all kinds of pollution, air pollution has the most direct impact on people’s life and does great harm to public health [8]. According to the data of the World Health Organization, about 7 million people worldwide die from indoor and outdoor air pollution every year. The level of exposure of Chinese residents to PM_2.5 is more than 3 times the annual guideline of the World Health Organization (WHO). Air pollution has threatened people’s quality of life and economic development, and is becoming increasingly serious in many fields [9]. Air pollution increases poverty in low-income areas [10], imposes a significant public health and economic burden [11], as well as future health burdens, and poses a major challenge to sustainable economic development [12]. Blue skies and white clouds are a blessing to people’s livelihoods, while environmental pollution is to their detriment. The 14th Five-Year Plan and the 2035 vision outline point out that as China has shifted to a stage of high-quality development, we must keep in mind the overall strategy for national rejuvenation and the major changes unseen in the world in a century, and have a deep understanding of China’s new features and requirements in the new era. The relationship among industry, environment and economy has been the focus of attention in the field of economics.

This study places industry, the environment and the economy in a unified framework for analysis. Different from previous studies, the possible marginal contribution of this study lies in the following: based on the data analysis of 273 urban samples, the research conclusions are general, with a large sample size and wide coverage. At the same time, most studies so far have directly adopted PM_2.5 as the core evaluation index of air pollution. Considering the differences of major sources of air pollution in different cities, this study uses the six environmental air quality evaluation indexes in the Bulletin of the State of Ecological Environment of China to build an air pollution index system, which is more reasonable in measuring air pollution. From the perspective of spatial analysis, the spatial measurement method is introduced to comprehensively consider the spatial spillover effects of industrial agglomeration, air pollution and sustainable economic development. Trying to answer these three questions:

• What is the impact of industrial agglomeration on sustainable economic development?
• What is the impact of air pollution on sustainable economic development?
• Is there a mediating effect of air pollution in the influence of industrial agglomeration on sustainable economic development?

Good coordination of the relationship between industry and environment is a necessary condition for sustainable economic development. The synergy between industry, environment and sustainable economic development has been realized in goals, policies, measures, methods and other aspects, providing a reference for further promoting the realization of sustainable economic development goals.

The rest of the paper is arranged as follows: the Section 2 presents a literature review and theoretical hypothesis; the Section 3 describes the construction of the econometric model; the Section 4 describes the variables and data; the Section 5 and Section 6 presents the empirical analysis and discussion; and the Section 7 presents the conclusion and revelations.

2. Literature Review and Theoretical Hypotheses

The heterogeny of industrial distribution is a universal spatial phenomenon. Industrial density increases from east to central, west and northeast, and the trend of industrial agglomeration is obvious in the east [13]. Any form and level of imbalance will become an obstacle to sustainable development [14]. China’s industrial spatial distribution is not balanced, and there is an excessive agglomeration of regional industries [15], which exert negative effects on sustainable economic development [16]. Lu and Tao (2009) believe that industrial agglomeration is not conducive to economic growth, showing an inverted “U” shape [17]. However, some scholars believe that industrial agglomeration contributes positively to sustainable economic development. Dong et al. (2019) believe that industrial agglomeration can promote sustainable economic development by attracting foreign investment [18]. The relationship between industrial agglomeration and sustainable economic development can be explained according to three different effects: scale effect, technology effect and structure effect [19]. The scale effect means that the more industry gathers, the more resources will be invested, which will produce more pollutants and restrict the sustainable development of the economy [20]. The technological effect describes the process by which industrial agglomeration drives the application and promotion of advanced clean technologies, reducing pollutant emission and promoting sustainable economic development [21]. The structural effect is that by which industrial agglomeration affects changes in industrial structure [22]. Different types of industrial structure have different pollution conditions and play different roles in sustainable economic development. With industrial agglomeration, the above three effects have different impacts on sustainable economic development, and the debate on the relationship between the two is still inconclusive. It should be realized that there is a complex nonlinear relationship between industrial agglomeration and sustainable economic development. Accordingly, the first research hypothesis is as follows:

Hypothesis 1 (H1).

Industrial agglomeration has a spatial spillover effect on sustainable economic development, and there is a nonlinear relationship between them.

Scholars from different countries have adopted various methods to discuss the relationship between air pollution and sustainable economic development from different perspectives [23,24,25]. Many scholars believe that air pollution is one of the obstacles to sustainable economic development [26], and air pollution restricts the realization of sustainable economic development goals. Canh et al. [27] believe that haze pollution has a significant negative impact on economic development. Dong et al. (2021) found that the relationship between air pollution and economic growth in countries with moderate haze concentration presents an inverted “N” shape, while the relationship between air pollution and economic growth in other countries presents a positive “N” shape [28]. Molina-Gómez et al. (2021) believe that air pollution has an impact on people’s living quality and sustainable economic development [29]. Air pollution must be tackled if the economy is to move towards sustainable development [7]. Solving the problem of air pollution is not only the goal of all governments, but also the key to the long-term sustainable development of the international community and economy. Only by actively responding to air pollution, coordinating the benign interaction between ecology and economy [30], adhering to the decoupling between economic growth and air quality optimization, striving to explore the coordinated control strategy of air pollution and sustainable economic development, strengthening environmental regulations and improving the coordinated monitoring mechanism of environmental and economic development, can the sustainable economic development be smoothly promoted. Accordingly, the second research hypothesis is as follows:

Hypothesis 2 (H2).

Air pollution has a spatial spillover effect on sustainable economic development, and there is a nonlinear relationship between them.

There is a strong connection between industrial agglomeration and air pollution. Industrial agglomeration promotes rapid economic development, but also increases the discharge of various pollutants. The pollution gas produced by industrial production is an important reason for the increasing seriousness of air pollution [31]. Industrial agglomeration consists of an “agglomeration effect” and a “scale effect”, and different effects have given rise to a variety of debates. Existing literature on the relationship between industrial agglomeration and air pollution can be summarized into the following three types: first, industrial agglomeration leads to more serious air pollution. Excessive industrial agglomeration increases energy consumption in production and living, resulting in a sharp rise in haze pollution. The “congestion effect” of industrial agglomeration is not conducive to the diffusion of pollutants, and the concentration of air pollutants in a short period of time exceeds the urban environmental capacity, resulting in a decline in urban air quality [32]. Second, industrial agglomeration is conducive to curbing air pollution. Industrial agglomeration usually brings the agglomeration of factors, improves the green awareness of residents [33], promotes technological progress and resource utilization efficiency [34], shortens the travel distance, improves the utilization rate of public transportation, and strengthens the promotion and application of clean technologies, thus achieving the effect of emission reduction. Liu et al. (2023) verified the role of industrial agglomeration in reducing haze pollution [35], while Lin et al. (2022) confirmed that continuous industrial agglomeration in cities promotes industrial consumption of clean energy and public transport services [36], thus reducing sources of air pollution and improving air quality. Third, there is a complex dynamic nonlinear relationship between industrial agglomeration and air pollution. Ref. [37] proved that although the shape of the Environmental Kuznets Curve (EKC) is different in different periods in countries with different levels of development, it generally presents an inverted “U” shape. Ref. [38] also confirmed the inverted “U”-shaped relationship, while Babu and Datta (2013) found a positive “U”-shaped curve relationship between industrial agglomeration and air pollution [39], and Wang and Wang (2019) reported a positive “N”-shaped nonlinear relationship between industrial agglomeration and air pollution [40]. In addition, the air conditions of adjacent areas have strong spatial correlations, and the influence mechanism of industrial agglomeration on air pollution presents various forms. Ghanem (2018) argued that the impact of increased industrial benefits on opportunities for sustainable development depends on the extent to which increased air pollution reduces labor productivity by increasing morbidity [41]. Gouveia et al. (2021) confirmed the relationship between higher urban industrial density and lower PM_2.5 [42]. This shows that industrial agglomeration can promote sustainable economic development by reducing air pollution. The above discussion indicates the existence of an “industrial agglomeration–air pollution–sustainable economic development” mechanism. Accordingly, the third research hypothesis is as follows:

Hypothesis 3 (H3).

The effect of industrial agglomeration on sustainable economic development is mediated by air pollution.

3. Model Research

The STRIPAT model, proposed by Haseeb et al. (2017), is a classic model for studying industrial, economic and environmental issues [43]. This model allows proper decomposition and improvement of impact factors, but it assumes that the relationship between variables is linear. In order to verify the nonlinear relationship between industrial agglomeration and sustainable economic development, the quadratic term of industrial agglomeration is added on the basis of STRIPAT model, and the model is constructed as follows:

$$\ln (su{s}_{it})={\beta }_0+{\beta }_1\ln (in{d}_{it})+{\beta }_2{\ln }^2(po{p}_{it})+{\beta }_3\ln (X_{it})+{\epsilon }_{it}$$

(1)

In the STRIPAT model, the industrial scale (P), the average annual industrial income (A) and the technical level (T) are represented by the industrial density ($in{d}_{it}$), the average annual enterprise income ($din{c}_{it}$) and the number of patents granted ($tec{h}_{it}$), respectively. X represents control variables including average annual income of enterprise cluster ($din{c}_{it}$), technical level ($tec{h}_{it}$), industrial structure ($ind{s}_{it}$) and openness to the outside world ($ope{n}_{it}$). ${\epsilon }_{it}$ is a random perturbation term, with subscripts i and t representing city and year, respectively.

Industrial agglomeration and sustainable economic development both have spatial spillover effects, so it would be biased to continue to analyze them under spatial homogeneity. In order to more objectively measure the spatial correlation and heterogeneity between industrial agglomeration and sustainable economic development in cities across China, and to improve the explanatory ability of spatial econometric models, the spatial Dubin model was constructed by using the spatial weight matrix of geographical distance on the basis of model Formula (1). The model is established in the following form:

$$\begin{array}{l}\ln (su{s}_{it})=\delta {\displaystyle {\sum }_{j=1}^N{\mathrm{W}}_{ij}\ln (su{s}_{it})}+\ln \alpha +{\beta }_1\ln \left(in{d}_{it}\right)+{\beta }_2{\ln }^2\left(in{d}_{it}\right)+{\beta }_3\ln \left(X_{it}\right)+{\theta }_1{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (in{d}_{it})}\\ +{\theta }_2{\displaystyle {\sum }_{j=1}^N{W}_{ij}{\ln }^2\left(in{d}_{it}\right)+{\theta }_3}{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (X_{it})+{\mu }_i}+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(2)

where $\ln (su{s}_{it})$ represents the observed value of sustainable economic development of the explained variable in region i at time t, and $\mathrm{W}ij\ln (su{s}_{it})$ represents the interaction between the dependent variable $\ln (su{s}_{it})$ and the dependent variable $\ln (su{s}_{it})$ of adjacent units. $\delta$ is the spatial autoregressive coefficient, $\alpha$ is the constant term, $\beta$ is the fixed unknown parameter vector, ${\mathrm{W}}_{ij}$ is the element of the spatial weight matrix W, ${\mu }_i$ is the spatial characteristic effect, ${\lambda }_t$ is the period characteristic effect, ${\epsilon }_{it}$ is the random error term.

Both air pollution and sustainable economic development have a certain spatial proximity, which is reflected in the spatial spillover effect of air pollution and sustainable economic development. Based on hypothesis 2, air pollution and sustainable economic development are placed in the perspective of spatial econometric analysis, and the following model is established:

$$\begin{array}{l}\ln (su{s}_{it})=\delta {\displaystyle {\sum }_{j=1}^N{\mathrm{W}}_{ij}\ln (ai{r}_{it})}+\ln \alpha +{\beta }_1\ln \left(ai{r}_{it}\right)+{\beta }_2{\ln }^2\left(ai{r}_{it}\right)+{\beta }_3\ln \left(X_{it}\right)+{\theta }_1{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (ai{r}_{it})}\\ +{\theta }_2{\displaystyle {\sum }_{j=1}^N{W}_{ij}{\ln }^2\left(ai{r}_{it}\right)+{\theta }_3}{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (X_{it})+{\mu }_i}+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(3)

The indirect influence of the explanatory variable (X) on the explained variable (Y) through the intermediate variable (M) is studied, which can be described using the following equation form:

$$Y=cX+e1$$

(4)

$$M=aX+e2$$

(5)

$$Y=cX+bM+e3$$

(6)

Hypothesis 3 shows that industrial agglomeration demonstrates an intermediary effect of air pollution on sustainable economic development. In order to verify hypothesis 3, sustainable economic development ($\ln (sus)$) is regarded as the explained variable Y, air pollution ($\ln air$) as the intermediary variable M to be tested, and industrial agglomeration ($\ln ind$) as the explanatory variable X. The mediating effect test model is established in the following form:

$$\begin{array}{l}\ln (ai{r}_{it})=\delta {\displaystyle {\sum }_{j=1}^N{\mathrm{W}}_{ij}\ln (ai{r}_{it})}+\ln \alpha +{\beta }_1\ln \left(in{d}_{it}\right)+{\beta }_2{\ln }^2\left(in{d}_{it}\right)+{\beta }_3\ln \left(X_{it}\right)+{\theta }_1{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (in{d}_{it})}\\ +{\theta }_2{\displaystyle {\sum }_{j=1}^N{W}_{ij}{\ln }^2\left(in{d}_{it}\right)+{\theta }_3}{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (X_{it})+{\mu }_i}+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(7)

$$\begin{array}{l}\ln (su{s}_{it})=\delta {\displaystyle {\sum }_{j=1}^N{\mathrm{W}}_{ij}\ln (su{s}_{it})}+\ln \alpha +{\beta }_1\ln \left(in{d}_{it}\right)+{\beta }_2{\ln }^2\left(in{d}_{it}\right)+{\beta }_3\ln \left(ai{r}_{it}\right)+{\beta }_4\ln (X_{it})\\ +{\theta }_1{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (in{d}_{it})}+{\theta }_2{\displaystyle {\sum }_{j=1}^N{W}_{ij}{\ln }^2\left(in{d}_{it}\right)+{\theta }_3}{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (ai{r}_{it})}+{\theta }_4{\displaystyle {\sum }_{j=1}^N{W}_{ij}\ln (X_{it})}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(8)

Obviously, Formula (2) corresponds to Formula (4) of the mediation effect test model, while Formulas (7) and (8) respectively correspond to Formulas (5) and (6). Therefore, Formulas (2), (7) and (8) constitute a complete testing process for the mediation effect model.

4. Data and Methodology

4.1. Variable Selection

Industrial agglomeration. Industry density is the number of industries per unit of land area, which is a key index for measuring the spatial distribution of industries and revealing the trend of industrial agglomeration. Industrial agglomeration is measured by industrial density data of each city.

Air pollution. Most studies to date have used PM_2.5 as the core evaluation index of air pollution, as it takes into account differences in the main sources of air pollution in different cities. The six environmental air quality evaluation indexes of fine particulate matter (PM_2.5), ozone (O₃), inhalable particulate matter (PM₁₀), carbon monoxide (CO), nitrogen dioxide (NO₂) and sulfur dioxide (SO₂) are used for reference in the Bulletin of the State of Ecological Environment of China. Based on the measured concentration values of the six aforementioned pollutants, the index describing the air pollution status of each city is determined by weighted average.

Sustainable economic development. The United Nations World Commission on Environment and Development defines sustainable development as “development that meets the needs of the present without jeopardizing the ability of future generations to meet their own needs”. Sustainable economic development involves multiple dimensions such as population, resources, environment, economy and society, and there is no unified measurement standard. We used six different dimensions—GDP per capita, power consumption per unit of GDP, dust emission per unit of GDP, harmless disposal rate of household waste, green coverage rate of built-up areas, and proportion of tertiary industry in GDP—as proxy variables for sustainable economic development. Economic sustainable development index is in essence a multi-dimensional and multi-index comprehensive evaluation system model. The processing method of the economic sustainable development index needs to take into consideration many aspects, such as time span and regional differences. The existing literature has usually adopted subjective and objective comprehensive evaluation methods and entropy methods to solve such problems. Therefore, an entropy method is adopted to construct economic sustainable development index. The specific steps are as follows:

The matrix $R_x$ is constructed as follows:

$$R_x=\left[\begin{array}{ccc}{X}_{11}& \dots & X_{1m}\\ \dots & \dots & \dots \\ X_{n1}& \dots & X_{nm}\end{array}\right]$$

(9)

where $X_{ij}$ represents the i-th indicator data of the j-th item object. This paper involves 273 Chinese cities and six indicators of sustainable economic development.

Preprocessing of index

After forward and dimensionless processing of the original data, the matrix is obtained as follows:

$$x_{ij}^{\prime }=\frac{x_{ij}-\min \left\{x_{ij}\dots x_{nj}\right\}}{\max \left\{x_{ij}\dots x_{nj}\right\}-\min \left\{x_{ij}\dots x_{nj}\right\}}$$

(10)

$$x_{ij}^{\prime }=\frac{\max \left\{x_{ij}\dots x_{nj}\right\}-x_{ij}}{\max \left\{x_{ij}\dots x_{nj}\right\}-\min \left\{x_{ij}\dots x_{nj}\right\}}$$

(11)

$$R_y=\left[\begin{array}{ccc}{y}_{11}& \dots & y_{1m}\\ \dots & \dots & \dots \\ y_{n1}& \dots & y_{nm}\end{array}\right]$$

(12)

Standardized index values

The proportion of the j-th index in the i-th region:

$$P_{ij}=\frac{y_{ij}}{{\displaystyle {\sum }_{i=1}^n{y}_{ij}}}$$

(13)

The entropy of the j-th index:

$$e_j=-\frac{1}{\ln n}{\displaystyle {\sum }_{i=1}^n{P}_{ij}\ln P_{ij}}$$

(14)

As can be seen from the definition of system entropy, if the entropy value of a certain index is small, it is more effective for identifying the target, because its degree of order is low, and the impact of this index on the target shows a high difference.

The index entropy weight $w_j$ is calculated as follows:

$$w_j=\frac{1-e_j}{{\displaystyle {\sum }_{i=1}^{n}1-e_j}}$$

(15)

$w_j$ is the lifting weight of the indicators, namely the weight indicating the role of each indicator in the comprehensive evaluation.

The comprehensive score $S_i$ is calculated as follows

$$S_i={\displaystyle {\sum }_{j=1}^m{w}_j}\times R_y$$

(16)

Control variable. The average annual income of enterprise clusters is an effective tool for reflecting the economic situation of a region, and can be used to reflect the development level of enterprises and has an important impact on the sustainable economic development. The average annual income of urban enterprise clusters is taken as the measurement. There is uncertainty about the impact of technological level on sustainable economic development. On the one hand, technological progress with a green bias is conducive to energy conservation and emissions reduction, and will promote sustainable economic development. On the other hand, productivity-oriented technological progress is not conducive to energy conservation and emission reduction due to large-scale production, and the environmental protection effect brought by technological progress has a long cycle. In their pursuit of performance, local governments are more inclined to seek the economic growth effect brought about by technological progress. Therefore, technological progress may not necessarily promote sustainable economic development. Following the general practice of the existing literature, the number of patents granted is used to measure the technical level. The impact of industrial structure on sustainable economic development is mainly reflected in pollution emission. Generally speaking, the industrial layout dominated by the secondary industry has the typical characteristics of high energy consumption, high emissions and high pollution, which is not conducive to sustainable economic development. Therefore, the proportion of the secondary industry in GDP is adopted to measure the industrial structure. Ref. [35] confirmed that pollution transfer is common among multinational corporations, which usually transfer pollution-intensive enterprises to countries and regions with lower standards. However, it should also be realized that opening to the outside world is conducive to attracting foreign enterprises with higher energy saving and emission reduction technologies to carry out greener production, thus contributing to emission reduction. Therefore, the impact of opening to the outside world on sustainable economic development is uncertain, and actual foreign investment is used to measure the degree of opening to the outside world. See

Table 1. Statistical description of major variables.

Variable	Obs	Mean	Std. Dev.	Min	Max
Ind	1092	5.8296	0.8901	1.7462	7.8817
Air	1092	3.7860	0.2478	2.9997	4.5540
Sus	1092	0.2933	0.1218	0.0944	0.8277
Dinc	1092	10.7759	0.5105	9.4319	12.2807
Tech	1092	7.5479	1.4652	4.2047	11.8475
Inds	1092	3.7946	0.2344	2.7047	4.3245
Open	1092	10.2678	1.8291	1.0986	14.9469

for a statistical description of the above main variables.

4.2. Data Source

In 2014, the Ministry of Environmental Protection organized the third phase of monitoring and implementation of new air quality standards, and released real-time monitoring data and air quality index of six air quality indicators to the public. Based on six air pollutants, in this study, the air pollution status of various cities and regions is considered. Therefore, the starting year for the data was 2015, and the time span of the data was 2015–2018. The data were mainly obtained from the China Urban Statistical Yearbook, the China Environmental Statistical Yearbook and the China Urban Construction Statistical Yearbook. The data cover 31 provinces and cities in China (excluding Hong Kong, Macao and Taiwan), and the final sample size comprised 273 cities in China after removing some cities with serious data shortages. For cities with light individual data missing in the sample size, the linear interpolation method was used to complete the data.

5. Results

5.1. Preliminary Study of Basic Regression Results

Based on Equation (1), traditional non-spatial panel regression analysis was used to analyze the relationship between industrial agglomeration and sustainable economic development. The results are reported in

Table 2. The influence of industrial agglomeration on sustainable economic development.

	M1	M2	M3	M4	M5	M6
ln(ind)	0.2079 *** (0.0685)	−0.5452 *** (0.1867)	−0.2613 *** (0.1302)	−0.2821 ** (0.1103)	−0.2148 ** (0.0839)	−0.2145 ** (0.0843)
ln(ind2)		0.0650 *** (0.0195)	0.0263 ** (0.0117)	0.0268 *** (0.0098)	0.0234 *** (0.0078)	0.0233 *** (0.0078)
ln(dinc)			0.2335 *** (0.0089)	0.1945 *** (0.0104)	0.2149 *** (0.0071)	0.2146 *** (0.0072)
ln(tech)				0.0225 *** (0.0029)	−0.0025 (0.0018)	−0.0025 (0.0018)
ln(inds)					−0.1648 *** (0.0072)	−0.1649 *** (0.0072)
ln(open)						0.0002 (0.0005)
_cons	−0.9200 ** (0.3996)	1.2106 *** (0.4660)	−1.6143 *** (0.3866)	−1.2614 *** (0.3319)	−0.9384 *** (0.2494)	−0.9376 *** (0.2512)
N	1365	1365	1365	1365	1365	1365
R2	0.0383	0.0611	0.7346	0.7722	0.8879	0.8879

Note: ** p < 0.01; *** p < 0.001.

. Model 1 only considers the impact of industrial agglomeration on sustainable economic development, and the results are significant at the 1% level. Based on Model 1, Model 2 includes the quadratic term of industrial agglomeration, and the results are still significant, indicating that there is a nonlinear relationship between industrial agglomeration and sustainable economic development. From model 3 to model 6, control variables such as average annual income, technical level, industrial structure and openness of enterprise clusters were successively included. The results are still significant, and the primary term coefficient of industrial agglomeration is negative, while the secondary term coefficient is positive, indicating a positive “U” shape between industrial agglomeration and sustainable economic development. The nonlinear relationship between industrial agglomeration and sustainable economic development has been preliminarily verified.

5.2. Regression Analysis of Spatial Dubin Model

Before performing spatial model parameter estimation, the spatial correlation test must be carried out to determine whether the spatial econometric analysis method is suitable. The spatial correlation test of the core variables industrial agglomeration ($\ln ind$) and sustainable economic development ($\ln (sus)$) was conducted, and the test results are reported in

Table 3. Moran’s I based on geographic distance matrix.

Year	lnind		lnsus
	Moran’s I	p	Moran’s I	p
2014	−0.056 ***	0.000	−0.028 ***	0.000
2015	−0.057 ***	0.000	−0.021 ***	0.000
2016	−0.058 ***	0.000	−0.019 ***	0.000
2017	−0.059 ***	0.000	−0.018 ***	0.000
2018	−0.059 ***	0.000	−0.019 ***	0.000

Note: *** p < 0.001.

. According to the results in Table 3, the Global Moran index (Moran’s I) [44] test results of the two variables—industrial agglomeration and economic sustainable development—are both negative and significant at the 1% level, indicating that industrial agglomeration and economic sustainable development have significant negative spatial effects, which can be analyzed by spatial econometric analysis.

In order to select an appropriate spatial metering model, a series of tests needed to be carried out, and the test results are reported in

Table 4. Test results of spatial model.

	Index	Statistic Value	p
LM test	LM_Error_test	37.299 ***	0.000
	R_LM_Error_test	41.235 ***	0.000
	LM_Lag_test	4.212 **	0.040
	R_LM_Lag_test	8.148 ***	0.004
Wald test	LR-SAR	118.41 ***	0.000
	LR-SEM	108.93 ***	0.000
	Wald-SAR	382.52 ***	0.000
	Wald-SEM	116.66 ***	0.000

Note: ** p < 0.01; *** p < 0.001.

. First, the LM test was conducted. LMerr and R-LM-err [45] rejected the null hypothesis at a significance level of 1%, while lm-lag and R-LM-lag rejected the null hypothesis at significance levels of 5% and 1%, respectively, indicating that the model had a spatial lag term and a spatial error effect. The space error model (SEM) and the space lag model (SAR) can be used [46]. In this case, it is more appropriate to directly select the SDM model in combination with SEM and SAR. Both the LR test and the Wald test results rejected the null hypothesis at a significance level of 1%, that is, the SDM model will not degenerate to the SAR model or SEM model, and the SDM model is to be preferred.

The influence of industrial agglomeration on sustainable economic development was deduced based on Equation (2), and the results are reported in

Table 5. Regression results of spatial model of industrial agglomeration on sustainable economic development.

Variable	Coefficient		Effect Estimation
	x	Wx	Direct Effect	Indirect Effect	Total Effect
ln(ind)	−0.1948 *** (0.0453)	−1.9375 ** (0.8226)	−0.1693 * (0.0984)	−0.4016 * (0.2334)	−0.5709 ** (0.2278)
ln(ind2)	0.0220 *** (0.0038)	0.2136 *** (0.0712)	0.0191 * (0.0111)	0.0437 ** (0.0221)	0.0629 *** (0.0190)
ln(dinc)	0.1970 *** (0.0045)	0.4960 *** (0.0471)	0.1984 *** (0.0062)	−0.0127 (0.0119)	0.1857 *** (0.0102)
ln(tech)	−0.0047 *** (0.0013)	−0.0201 ** (0.0079)	−0.0045 *** (0.0016)	−0.0021 (0.0026)	−0.0067 *** (0.0021)
ln(inds)	−0.1491 *** (0.0058)	−0.6305 *** (0.0615)	−0.1451 *** (0.0223)	−0.0627 ** (0.0279)	−0.2078 *** (0.0151)
Ln(open)	−0.0000 (0.0004)	0.0242 *** (0.0034)	−0.0005 (0.0020)	0.0069 *** (0.0022)	0.0064 *** (0.0009)
ρ	−2.7420 *** (0.1471)
N	1092
R2	0.7371

Note: * p < 0.05; ** p < 0.01; *** p < 0.001.

. The ρ value of the spatial Durbin model is significant (−2.7420), and not zero, indicating that industrial agglomeration has a significant negative spatial effect on sustainable economic development. Firstly, the spatial coefficient was analyzed, and the primary coefficient (−1.9375) of the impact of industrial agglomeration on sustainable economic development is negative and the secondary coefficient (0.2136) is positive, and significant at the levels of 5% and 1%, respectively. This shows that there is a “U”-shaped nonlinear relationship between industrial agglomeration and sustainable economic development, and the indirect effect of industrial agglomeration significantly indicates that industrial agglomeration has a spatial spillover effect on sustainable economic development; therefore, hypothesis 1 is verified. The influence of each control variable was analyzed. The average annual income of enterprise clusters has a significant positive impact on sustainable economic development, indicating that regional industrial development plays an important role in sustainable economic development. The coefficient of technological level is negative and significant, indicating that the improvement of technological level has not made a positive contribution to the sustainable development of economy. Combined with the actual development situation, the following explanation can be proposed: technological progress is more productivity oriented, focusing on the economic growth effect of technological progress rather than the effect of energy saving and emission reduction, and technology promotion has a time lag, so it has failed to make a positive contribution to sustainable economic development. The significant negative impact of industrial structure on sustainable economic development indicates that the higher the proportion of secondary industry in GDP, the more detrimental to sustainable economic development, and the heavy pollution of the secondary industry is not conducive to sustainable economic development. Therefore, in order to achieve sustainable economic development, we should vigorously promote the optimization of industrial structure. There is a significant positive correlation between openness to the outside world and sustainable economic development. Cities with high openness to the outside world are often able to attract foreign enterprises with high energy saving and emission reduction technologies to carry out green production. The sample data just verifies the positive contribution of openness to the outside world, which plays a good role in promoting sustainable economic development. The effect decomposition of the spatial Dubin model is discussed below. The direct effect measures the influence of the change in the independent variable on the local dependent variable. The indirect effect, also known as spatial spillover effect, measures the influence of independent variable changes on dependent variables in neighboring areas. The total effect is the influence of the explanatory variable on the explained variable in the whole area. For this region, the excessive speed in the early stage of industrial agglomeration leads to increased energy consumption, increased emissions and increased congestion effect, aggravated air pollution and inhibited sustainable economic development. With further industrial agglomeration and slowing down, people’s awareness of green development is enhanced, and various advanced technologies are constantly promoted and applied, thus promoting sustainable economic development.

The impact of air pollution on sustainable economic development was explored based on Equation (3), and the results are reported in

Table 6. Regression results of spatial model of air pollution on sustainable economic development.

Variable	Coefficient		Effect Estimation
	x	Wx	Direct Effect	Indirect Effect	Total Effect
ln(air)	0.0529 (0.0509)	−0.8611 * (0.4889)	0.0655 (0.0555)	−0.2878 ** (0.1408)	−0.2222 (0.1586)
ln(air2)	−0.0085 (0.0066)	0.1182 * (0.0642)	−0.0103 (0.0072)	0.0404 ** (0.0184)	0.0301 (0.0207)
ln(dinc)	0.1967 *** (0.0048)	0.4813 *** (0.0534)	0.1974 *** (0.0047)	−0.0071 (0.0077)	0.1903 *** (0.0084)
ln(tech)	−0.0047 *** (0.0014)	−0.0378 ** (0.0067)	−0.0045 *** (0.0018)	−0.0074 *** (0.0025)	−0.0119 *** (0.0017)
ln(inds)	−0.1485 *** (0.0061)	−0.7519 *** (0.0682)	−0.1449 *** (0.0164)	−0.1067 *** (0.0204)	−0.2517 *** (0.0131)
ln(open)	0.0000 (0.0004)	0.0312 *** (0.0069)	−0.0003 (0.0015)	0.0090 *** (0.0023)	0.0087 *** (0.0018)
ρ	−2.5680 *** (0.2068)
N	1092
R2	0.6564

Note: * p < 0.05; ** p < 0.01; *** p < 0.001.

. The coefficients of primary and secondary terms of air pollution are significant at the 1% level, and the coefficient of the primary term is negative, and the coefficient of the secondary term is positive, indicating that there is a typical positive “U”-shaped curve relationship between air pollution and sustainable economic development. The indirect effect of air pollution is significant, indicating that there is an obvious spatial spillover effect of air pollution. Therefore, hypothesis 2 is verified. That is to say, air pollution can inhibit and then promote sustainable economic development. This is because the aggravation of air pollution will inevitably inhibit the sustainable development of the economy. However, air pollution cannot be allowed to continue increasing. After reaching a certain value, it will promote the sustainable development of the economy under the dual role of environmental regulation and the promotion and application of advanced clean technology.

5.3. Analysis of Mediating Effect Test

The influence of industrial agglomeration on air pollution was analyzed using Equation (7), and the results are shown in column (2) of

Table 7. Test results of air pollution mediating effect.

Variable	(1) lnsus Coefficient		(2) lnair Coefficient		(3) lnsus Coefficient
	x	Wx	x	Wx	x	Wx
ln(ind)	−0.1948 *** (0.0453)	−1.9375 ** (0.8226)	0.1153 *** (0.0344)	17.0795 *** (3.9690)	−0.1978 *** (0.0453)	−1.8548 ** (0.9332)
ln(ind2)	0.0220 *** (0.0038)	0.2136 *** (0.0712)	−0.0048 (0.0032)	−1.7516 *** (0.3999)	0.0220 *** (0.0038)	0.2017 ** (0.0870)
ln(air)	-	-	-	-	−0.0111 *** (0.0042)	−0.0047 *** (0.013)
ln(dinc)	0.1970 *** (0.0045)	0.4960 *** (0.0471)	−0.0653 *** (0.0111)	−2.6273 *** (0.6342)	0.1972 *** (0.0047)	0.4919 *** (0.0522)
ln(tech)	−0.0047 *** (0.0013)	−0.0201 ** (0.0079)	0.0333 *** (0.0051)	2.3404 *** (0.3904)	−0.0050 *** (0.0013)	−0.0220 ** (0.0086)
ln(inds)	−0.1491 *** (0.0058)	−0.6305 *** (0.0615)	0.2088 *** (0.0204)	−6.3692 *** (0.8589)	−0.1481 *** (0.0059)	−0.6213 *** (0.0900)
ln(open)	−0.0000 (0.0004)	0.0242 *** (0.0034)	0.0034 (0.0034)	−0.3220 *** (0.0918)	−0.0001 (0.0004)	0.0219 ** (0.0093)
N	1092	1092	1092	1092	1092	1092
ρ	−2.7420 *** (0.1471)		−0.9534 *** (0.3045)		−2.7440 *** (0.1465)
Variance sigma2_e	0.0001 *** (0.0000)		0.0197 *** (0.0007)		0.0001 *** (0.0000)
R2	0.7371		0.0126		0.7980

Note: ** p < 0.01; *** p < 0.001.

. The ρ value of the spatial Dubin model is −0.9534, which is, significantly, not zero, indicating that industrial agglomeration has a significant negative spatial effect on air pollution. Based on the analysis of the spatial coefficient, the primary term coefficient of industrial agglomeration’s influence on air is positive and the secondary term coefficient is negative, both of which are significant at the level of 1%, indicating an inverted “U”-shaped nonlinear relationship between industrial agglomeration and air pollution.

The indirect effect of the explanatory variable on the explained variable through the intermediate variable is called the mediation effect. Hypothesis 3 was empirically tested by combining Equations (2), (7) and (8), and the results are reported in Table 7. Based on the testing steps of the mediation effect mentioned above, the coefficients in the three Equations (2), (7) and (8) were tested. It was found that the coefficients of industrial agglomeration are significant, and the coefficients of air pollution are also significant and consistent with expectations, which proves that air pollution is an intermediary variable in the effect of industrial agglomeration on sustainable economic development, and verifies the validity of hypothesis 3. The results of industrial agglomeration on sustainable economic development are reported in Table 7, in column (1), indicating that there is a positive “U”-shaped nonlinear relationship between industrial agglomeration and sustainable economic development without considering the impact of air pollution on sustainable economic development. The results in column (2) show that there is an inverted “U”-shaped non-linear relationship between industrial agglomeration and air pollution. The results in column (3) show that there is still a “U”-shaped relationship between industrial agglomeration and sustainable economic development after considering the influence of air pollution. The results show that there is an explanatory mechanism of “industrial agglomeration–air pollution–sustainable economic development”. In the short term, industrial agglomeration leads to the increase in air pollution, which restricts the sustainable economic development. With the further acceleration of industrial agglomeration, the change in development consciousness and concept and the application of advanced technology, the decrease in air pollution will promote sustainable economic development. Increased air pollution will weaken the promoting effect of industrial agglomeration on economic sustainable development, which again confirms that air pollution can be an intermediary variable of industrial agglomeration affecting economic sustainable development.

6. Discussion

6.1. Theoretical Implications

Theoretically speaking, the influence of industrial agglomeration on sustainable economic development is bidirectional. Some scholars believe that industrial agglomeration will reduce air pollution through the use of clean energy and investing in public transportation [47]. However, another scholar believes that industrial agglomeration has a significant negative impact on air pollution concentration through spillover effect [48].

Specifically, first, industrial agglomeration affects air pollution through a “U”-shaped agglomeration effect. The increase in industry intensifies the demand for housing, infrastructure and private cars [49], leading to greater energy consumption [50]. In addition, improper disposal of a large amount of household garbage leads to a rise in air pollution. From another perspective, industrial agglomeration will promote investment and use of infrastructure such as public transport [51], reduce the use of single-use fuels in rural areas [52], and thus reduce air pollution. The results show that industrial agglomeration plays a positive role in promoting sustainable economic development. However, the scale of industrial agglomeration does not directly lead to sustainable economic development, but through various indirect activities and demands.

Industrial agglomeration encourages more rural residents to move to cities, leading to an increase in the use of private cars and air pollution [53]. However, with improvements in environmental awareness, people are more inclined to travel green and use clean energy. The regression results show that the influence of industrial agglomeration is negative in both years. Social development and progress have raised the level of education, increased the inflow of highly qualified personnel, and generally increased the awareness of environmental problems. Coupled with the development of public transport in recent years, the increasing number of buses and the emergence of new fuel-efficient buses have reduced the use of private cars and offset some of the pollution produced by fossil fuels, thus reducing air pollution.

6.2. Managerial Implications

Industrial agglomeration itself is a dynamic development concept involving many fields, while at the same time, industrial agglomeration is a process that is constantly changing, driven by both time and space dimensions. How to coordinate the relationship between industrial agglomeration and sustainable economic development is a topic worthy of further study in the future. In the short term, industrial agglomeration leads to urban congestion, urban energy consumption and emission increase, which aggravates air pollution and forms a certain constraint to sustainable economic development. In the long term, industrial agglomeration attracts all factors to agglomeration, and the innovation ability enhances the technology level and the development level of comprehensive urban governance to reduce the environmental burden on the sustainable economic development constraints. The “U”-shaped relationship between industrial agglomeration and sustainable economic development does not mean that pollution should be made it possible to take place before treatment, but that more attention should be paid to the constraints on sustainable economic development caused by rapid industrial agglomeration in the short term, and to the negative impact of air pollution caused by excessive industrial agglomeration and the resulting constraints on sustainable economic development. Upgrading the “temperature” of the city, expanding the “width” of the city, increasing the “concentration” of the city, focusing on the “density” of the city to guide industries to gather in a reasonable and appropriate manner within the region, and to form a reasonable layout for industrial clustering. This will reduce the negative impact of industrial agglomeration on air pollution by technological innovation and productivity improvement, thus reducing the constraints on sustainable economic development caused by air pollution and promoting sustainable green economic development.

7. Conclusions

In order to reveal the underlying mechanism of industrial agglomeration, air pollution and sustainable economic development in China, in this paper, the mediating effects of industrial agglomeration and sustainable economic development, air pollution and sustainable economic development, and industrial agglomeration and air pollution on sustainable economic development were analyzed using the spatial Dubin model. The following conclusions are drawn:

First, the relationship between industrial agglomeration and sustainable economic development presents a positive “U”-shaped nonlinear relationship with a spatial spillover effect. Data from 273 cities show that the rapid industrial agglomeration in the short term increases energy consumption in production and living, causes a prominent congestion effect and excessive energy consumption in resources, which inhibits sustainable economic development. However, with the further industrial agglomeration, the slowing down of agglomeration speed also results in the agglomeration of more factors, technological progress and resource utilization efficiency are significantly improved, and production efficiency is greatly improved, while industrial structure is constantly optimized, which makes a positive contribution to promoting the sustainable development of the economy.

Second, there is a positive “U”-shaped nonlinear relationship between air pollution and sustainable economic development, and there is a spatial spillover effect between them. The increase in air pollution directly restricts the sustainable development of economy, but when the air pollution reaches a certain peak, under the dual role of environmental regulation and the promotion and application of advanced clean technology, the gradual reduction of air pollution will promote the sustainable development of economy.

Thirdly, the influence of industrial agglomeration on sustainable economic development is influenced by air pollution, an intermediary variable. The existence of air pollution weakens the promoting effect of industrial agglomeration on sustainable economic development.

The limitation of this paper is that it mainly discusses the impact of industrial agglomeration on six types of air pollutants. The data come from developing countries, and the data from developed countries will be used for comparison in the future. It also indicates the direction China must go to achieve the strategic goal of carbon neutrality and carbon peak.